Several methods for detecting the face and extracting the facial features and components exist in the
literature. These methods are different in their complexity, performance, type and nature of the images and
the targeted application. The facial features and components are used in security applications, robotics and
assistance for the disabled. We use these components and characteristics to determine the state of alertness
and fatigue for medical diagnoses. In this work we use plain color background images whose color is
different from the skin and which contain a single face. We are interested in FPGA implementation of this
application. This implementation must meet two constraints, which are the execution time and the FPGA
resources. We have selected and have associated a face detection algorithm based on the skin detection
(using the RGB space) with a facial-feature extraction algorithm based on tracking the gradient and the
geometric model.
Regenerating face images from multi-spectral palm images using multiple fusio...TELKOMNIKA JOURNAL
This paper established a relationship between multi-spectral palm images and a face image based
on multiple fusion methods. The first fusion method to be considered is a feature extraction between different
multi-spectral palm images, where multi-spectral CASIA database was used. The second fusion method to
be considered is a score fusion between two parts of an output face image. Our method suggests that both
right and left hands are used, and that each hand aims to produce a significant part of
a face image by using a Multi-Layer Perceptron (MLP) network. This will lead to the second fusion part to
reconstruct the full-face image, in order to examine its appearance. This topology provided interesting results
of Equal Error Rate (EER) equal to 1.99%.
This document discusses principal component analysis (PCA) for face recognition. It begins with an introduction to face recognition and PCA. PCA works by calculating eigenvectors from a set of face images, which represent the principal components that account for the most variance in the image data. These eigenvectors are called "eigenfaces" and can be used to reconstruct the face images. The document then discusses how the system is implemented, including preparing a face database, normalizing the training images, calculating the eigenfaces/principal components, projecting the face images into this reduced space, and recognizing faces by calculating distances between projected test images and training images.
This document presents a novel two-step approach for skull stripping MRI brain images. The first step uses morphological reconstruction operations to generate a mask of the brain. The second step applies thresholding to the mask to extract the brain. The method was tested on axial PD and FLAIR MRI images. Results found Jaccard and Dice similarity scores above 0.8 and 0.9 respectively, indicating the method efficiently extracts the brain from the skull.
A novel approach for efficient skull stripping using morphological reconstruc...eSAT Journals
This document presents a novel two-step approach for skull stripping of MRI brain images. The first step uses morphological reconstruction operations including erosion, opening by reconstruction, dilation, and opening-closing by reconstruction to generate a primary segmentation mask. The second step applies thresholding to the primary mask to extract the final skull-stripped brain image. The method is tested on axial PD and FLAIR MRI images and achieves high Jaccard and Dice similarity scores compared to manually stripped images, demonstrating its effectiveness at skull stripping.
Multi modal face recognition using block based curvelet featuresijcga
In this paper, we present multimodal 2D +3D face recognition method using block based curvelet features. The 3D surface of face (Depth Map) is computed from the stereo face images using stereo vision technique. The statistical measures such as mean, standard deviation, variance and entropy are extracted from each block of curvelet subband for both depth and intensity images independently.In order to compute the decision score, the KNN classifier is employed independently for both intensity and depth map. Further, computed decision scoresof intensity and depth map are combined at decision level to improve the face recognition rate. The combination of intensity and depth map is verified experimentally using benchmark face database. The experimental results show that the proposed multimodal method is better than individual modality.
In this paper, we present an automatic application of 3D face recognition system using geodesic distance in Riemannian geometry. We consider, in this approach, the three dimensional face images as residing in Riemannian manifold and we compute the geodesic distance using the Jacobi iterations as a solution of the Eikonal equation. The problem of solving the Eikonal equation, unstructured simplified meshes of 3D face surface, such as tetrahedral and triangles are important for accurately modeling material interfaces and curved domains, which are approximations to curved surfaces in R3. In the classifying steps, we use: Neural Networks (NN), K-Nearest Neighbor (KNN) and Support Vector Machines (SVM). To test this method and evaluate its performance, a simulation series of experiments were performed on 3D Shape REtrieval Contest 2008 database (SHREC2008).
Fourier mellin transform based face recognitioniaemedu
This document presents a face recognition algorithm based on Fourier Mellin Transform. It begins with an introduction to face recognition and challenges of illumination and pose variations. It then describes extracting illumination invariant features by computing depth maps from input images using a shape from shading algorithm. Fourier Mellin Transform is applied to the depth maps to extract features. Experiments on the ORL database showed the approach achieved 100% recognition with 4 training images and 95.7% recognition with 3 training images, demonstrating robustness to illumination and pose variations.
International Journal of Engineering Research and Applications (IJERA) is an open access online peer reviewed international journal that publishes research and review articles in the fields of Computer Science, Neural Networks, Electrical Engineering, Software Engineering, Information Technology, Mechanical Engineering, Chemical Engineering, Plastic Engineering, Food Technology, Textile Engineering, Nano Technology & science, Power Electronics, Electronics & Communication Engineering, Computational mathematics, Image processing, Civil Engineering, Structural Engineering, Environmental Engineering, VLSI Testing & Low Power VLSI Design etc.
Regenerating face images from multi-spectral palm images using multiple fusio...TELKOMNIKA JOURNAL
This paper established a relationship between multi-spectral palm images and a face image based
on multiple fusion methods. The first fusion method to be considered is a feature extraction between different
multi-spectral palm images, where multi-spectral CASIA database was used. The second fusion method to
be considered is a score fusion between two parts of an output face image. Our method suggests that both
right and left hands are used, and that each hand aims to produce a significant part of
a face image by using a Multi-Layer Perceptron (MLP) network. This will lead to the second fusion part to
reconstruct the full-face image, in order to examine its appearance. This topology provided interesting results
of Equal Error Rate (EER) equal to 1.99%.
This document discusses principal component analysis (PCA) for face recognition. It begins with an introduction to face recognition and PCA. PCA works by calculating eigenvectors from a set of face images, which represent the principal components that account for the most variance in the image data. These eigenvectors are called "eigenfaces" and can be used to reconstruct the face images. The document then discusses how the system is implemented, including preparing a face database, normalizing the training images, calculating the eigenfaces/principal components, projecting the face images into this reduced space, and recognizing faces by calculating distances between projected test images and training images.
This document presents a novel two-step approach for skull stripping MRI brain images. The first step uses morphological reconstruction operations to generate a mask of the brain. The second step applies thresholding to the mask to extract the brain. The method was tested on axial PD and FLAIR MRI images. Results found Jaccard and Dice similarity scores above 0.8 and 0.9 respectively, indicating the method efficiently extracts the brain from the skull.
A novel approach for efficient skull stripping using morphological reconstruc...eSAT Journals
This document presents a novel two-step approach for skull stripping of MRI brain images. The first step uses morphological reconstruction operations including erosion, opening by reconstruction, dilation, and opening-closing by reconstruction to generate a primary segmentation mask. The second step applies thresholding to the primary mask to extract the final skull-stripped brain image. The method is tested on axial PD and FLAIR MRI images and achieves high Jaccard and Dice similarity scores compared to manually stripped images, demonstrating its effectiveness at skull stripping.
Multi modal face recognition using block based curvelet featuresijcga
In this paper, we present multimodal 2D +3D face recognition method using block based curvelet features. The 3D surface of face (Depth Map) is computed from the stereo face images using stereo vision technique. The statistical measures such as mean, standard deviation, variance and entropy are extracted from each block of curvelet subband for both depth and intensity images independently.In order to compute the decision score, the KNN classifier is employed independently for both intensity and depth map. Further, computed decision scoresof intensity and depth map are combined at decision level to improve the face recognition rate. The combination of intensity and depth map is verified experimentally using benchmark face database. The experimental results show that the proposed multimodal method is better than individual modality.
In this paper, we present an automatic application of 3D face recognition system using geodesic distance in Riemannian geometry. We consider, in this approach, the three dimensional face images as residing in Riemannian manifold and we compute the geodesic distance using the Jacobi iterations as a solution of the Eikonal equation. The problem of solving the Eikonal equation, unstructured simplified meshes of 3D face surface, such as tetrahedral and triangles are important for accurately modeling material interfaces and curved domains, which are approximations to curved surfaces in R3. In the classifying steps, we use: Neural Networks (NN), K-Nearest Neighbor (KNN) and Support Vector Machines (SVM). To test this method and evaluate its performance, a simulation series of experiments were performed on 3D Shape REtrieval Contest 2008 database (SHREC2008).
Fourier mellin transform based face recognitioniaemedu
This document presents a face recognition algorithm based on Fourier Mellin Transform. It begins with an introduction to face recognition and challenges of illumination and pose variations. It then describes extracting illumination invariant features by computing depth maps from input images using a shape from shading algorithm. Fourier Mellin Transform is applied to the depth maps to extract features. Experiments on the ORL database showed the approach achieved 100% recognition with 4 training images and 95.7% recognition with 3 training images, demonstrating robustness to illumination and pose variations.
International Journal of Engineering Research and Applications (IJERA) is an open access online peer reviewed international journal that publishes research and review articles in the fields of Computer Science, Neural Networks, Electrical Engineering, Software Engineering, Information Technology, Mechanical Engineering, Chemical Engineering, Plastic Engineering, Food Technology, Textile Engineering, Nano Technology & science, Power Electronics, Electronics & Communication Engineering, Computational mathematics, Image processing, Civil Engineering, Structural Engineering, Environmental Engineering, VLSI Testing & Low Power VLSI Design etc.
Face Detection for identification of people in Images of Internetijceronline
One method for searching the internet faces in images is proposed by using digital processing topological with descriptors. Location in real time with the development of a database that stores addresses of internet downloaded images, in which the search is done by text, but by finding facial image, is achieved. Face recognition in images of Internet has proved to be a difficult task, because the images vary considerably depending on viewpoint, illumination, expression, pose, accessories, etc. The descriptors for general information: containing low-level descriptors. Developments on face recognition systems have improved significantly since the first system; image analysis is a topic on which much emphasis is being given in order to identify parameters, visual features in the image that provide environment data that it is represented in the image, but without the intervention of a person. In this project raises its realization using the method of viola and jones as face descriptor. We can distinguish even different parts of the face such as eyes, eyebrows, nose and mouth.One method for searching faces in image taken from internet intends to use digital processing using topological descriptors. It is located the face in real time.
LOCAL REGION PSEUDO-ZERNIKE MOMENT- BASED FEATURE EXTRACTION FOR FACIAL RECOG...aciijournal
In the domain of image processing, face recognition is one of the most well-known research field. When
humans have very similar biometric properties, such as identical twins, the face recognition system is
considered as a challengeable problem. In this paper, the AdaBoost method is utilized to detect the
facial area of input image.
Medoid based model for face recognition using eigen and fisher facesijscmcj
Biometric technologies have gained a remarkable impetus in high security applications. Various biometric modalities are widely being used these days. The need for unobtrusive biometric recognition can be fulfilled through Face recognition which is the most natural and non intrusive authentication system. However the vulnerability to changes owing to variations in face due to various factors like pose,
illumination, ageing, emotions, expressions etc make it necessary to have robust face recognition systems.
Various statistical models have been developed so far with varying degree of accuracy and efficiency. This
paper discusses a new approach to utilize Eigen face and Fisher face methodology by using medoid instead
of mean as a statistic in calculating the Eigen faces and Fisher faces. The method not only requires lesser training but also demonstrates better time efficiency and performance compared to the conventional method of using mean
Face skin color based recognition using local spectral and gray scale featureseSAT Journals
Abstract Human face conveys more information about identity of person. Human face recognition is one of the most challenging problem and it can be used in many applications at different security places in airports, defense and banking sectors etc.In this work used colored features obtained from color segmentation because in real time scenario color provides the more information than gray scale image but it has a drawback. To overcome this drawback gray scale feature extracted from co-occurrence matrix of an image and for efficient face recognition of human Face under different illumination conditions spectral features can be extracted from face texture. These three feature vectors concatenated into a single feature vector and applied Lenc-Kral matching technique to measure similarity between the database and query image, the similarity is high then face is recognized. Keywords: Face recognition, illumination condition, local texture features, color segmentation.
A survey on human face recognition invariant to illuminationIAEME Publication
This document provides a survey of techniques for human face recognition that are invariant to illumination. It begins with an introduction to face recognition and the challenges posed by illumination variation. It then describes several preprocessing techniques that have been used to address illumination issues, including histogram equalization, adaptive histogram equalization, discrete cosine transform, discrete wavelet transform, and local binary pattern methods. The document concludes by comparing the performance of these various preprocessing techniques on standard face recognition databases based on error rates reported in other studies.
An efficient feature extraction method with pseudo zernike moment for facial ...ijcsity
Face recognition is one of the most challenging problems in the domain of image processing and machine
vision. Face recognition system is critical when individuals have very similar biometric signature such as
identical twins. In this paper, new efficient facial-based identical twins recognition is proposed according
to the geometric moment. The utilized geometric moment is Pseudo-Zernike Moment (PZM) as a feature
extractor inside the facial area of identical twins images. Also, the facial area inside an image is detected
using Ada Boost approach. The proposed method is evaluated on two datasets, Twins Days Festival and
Iranian Twin Society which contain scaled, which contain the shifted and rotated facial images of identical
twins in different illuminations. The results prove the ability of proposed method to recognize a pair of
identical twins. Also, results show that the proposed method is robust to rotation, scaling and changing
illumination.
Pupil Detection Based on Color Difference and Circular Hough Transform IJECEIAES
Human pupil eye detection is a significant stage in iris segmentation which is representing one of the most important steps in iris recognition. In this paper, we present a new method of highly accurate pupil detection. This method is consisting of many steps to detect the boundary of the pupil. First, the read eye image (R, G, B), then determine the work area which is consist of many steps to detect the boundary of the pupil. The determination of the work area contains many circles which are larger than pupil region. The work area is necessary to determine pupil region and neighborhood regions afterward the difference in color and intensity between pupil region and surrounding area is utilized, where the pupil region has color and intensity less than surrounding area. After the process of detecting pupil region many steps on the resulting image is applied in order to concentrate the pupil region and delete the others regions by using many methods such as dilation, erosion, canny filter, circle hough transforms to detect pupil region as well as apply optimization to choose the best circle that represents the pupil area. The proposed method is applied for images from palacky university, it achieves to 100 % accuracy
An Assimilated Face Recognition System with effective Gender Recognition RateIRJET Journal
This document summarizes an assimilated face recognition system that can also perform gender recognition. The system conducts experiments using databases like GENDER-FERET and Cambridge AT&T. For face recognition, it uses the Eigenfaces algorithm to extract features and classify faces. For gender recognition, it uses a trainable COSFIRE filter with Gabor filters to obtain face descriptors, which are classified using an SVM classifier. The experiments achieve a gender recognition rate of over 90%. The paper shows that the gender recognition approach outperforms other methods using handcrafted features and raw pixels.
HVDLP : HORIZONTAL VERTICAL DIAGONAL LOCAL PATTERN BASED FACE RECOGNITION sipij
Face image is an efficient biometric trait to recognize human beings without expecting any co-operation from a person. In this paper, we propose HVDLP: Horizontal Vertical Diagonal Local Pattern based face recognition using Discrete Wavelet Transform (DWT) and Local Binary Pattern (LBP). The face images of different sizes are converted into uniform size of 108×990and color images are converted to gray scale images in pre-processing. The Discrete Wavelet Transform (DWT) is applied on pre-processed images and LL band is obtained with the size of 54×45. The Novel concept of HVDLP is introduced in the proposed method to enhance the performance. The HVDLP is applied on 9×9 sub matrix of LL band to consider HVDLP coefficients. The local Binary Pattern (LBP) is applied on HVDLP of LL band. The final features are generated by using Guided filters on HVDLP and LBP matrices. The Euclidean Distance (ED) is used to compare final features of face database and test images to compute the performance parameters.
ABSTRACT: a rigorous work on static and dynamic appearance based classification systems for face is on but, it is proving to be a challenging task for researchers to design a proper system since human face is complex one. Decades of work was and is focussed on how to classify a face and on how to increase the rate of classification but, little attention was paid to overcome redundancy in image classification. This paper presents a novel idea which focuses on redundancy check and its elimination. The paper after drawing inferences from previous work gives out a novel idea for exact face classification and elimination of redundancy.
EV-SIFT - An Extended Scale Invariant Face Recognition for Plastic Surgery Fa...IJECEIAES
This paper presents a new technique called Entropy based SIFT (EV-SIFT) for accurate face recognition after the plastic surgery. The corresponding feature extracts the key points and volume of the scale-space structure for which the information rate is determined. This provides least effect on uncertain variations in the face since the entropy is the higher order statistical feature. The corresponding EV-SIFT features are applied to the Support vector machine for classification. The normal SIFT feature extracts the key points based on the contrast of the image and the V- SIFT feature extracts the key points based on the volume of the structure. However, the EV- SIFT method provides both the contrast and volume information. Thus EV-SIFT provide better performance when compared with PCA, normal SIFT and VSIFT based feature extraction.
Extraction of texture features by using gabor filter in wheat crop disease de...eSAT Journals
This document discusses a method for detecting diseases in wheat crops using image processing and artificial neural networks. It involves taking digital images of wheat crop leaves and preprocessing the images by applying Gaussian and median filters to reduce noise. The images are then segmented using CIELAB color space. Texture features like area, perimeter, contrast, and energy are extracted from the images using Gabor filters. These features are then fed into an artificial neural network classifier to identify the type of disease present in the wheat crop. The method aims to help farmers more quickly and accurately detect diseases so they can better manage their crops and increase agricultural productivity.
The document describes a novel method for localizing the pupil in eye gaze tracking systems. The proposed method fuses existing pupil localization techniques, including Hough transform, gray projection, and coarse positioning. This fusion approach aims to improve localization accuracy while lowering false detections. The method first preprocesses the eye image using filtering, edge detection, and thresholding/binarization. It then applies the fused techniques of Hough transform, gray projection, and coarse positioning to the preprocessed image to accurately detect the pupil boundary and center coordinates in a robust manner. The results are expected to provide better pupil localization compared to existing individual techniques.
Binary operation based hard exudate detection and fuzzy based classification ...IJECEIAES
Diabetic retinopathy (DR) is one of the most considerable reasons for visual impairment. The main objective of this paper is to automatically detect and recognize DR lesions like hard exudates, as it helps in diagnosing and screening of the disease. Here, binary operation based image processing for detecting lesions and fuzzy logic based extraction of hard exudates on diabetic retinal images are discused. In the initial stage, the binary operations are used to identify the exudates. Similarly, the RGB channel space of the DR image is used to create fuzzy sets and membership functions for extracting the exudates. The membership directives obtained from the fuzzy rule set are used to detect the grade of exudates. In order to evaluate the proposed approach, experiment tests are carriedout on various set of images and the results are verified. From the experiment results, the sensitivity obtained is 98.10%, specificity is 96.96% and accuracy is 98.2%. These results suggest that the proposed method could be a diagnostic aid for ophthalmologists in the screening for DR.
Content Based Image Retrieval Approach Based on Top-Hat Transform And Modifie...cscpconf
In this paper a robust approach is proposed for content based image retrieval (CBIR) using texture analysis techniques. The proposed approach includes three main steps. In the first one, shape detection is done based on Top-Hat transform to detect and crop object part of the image. Second step is included a texture feature representation algorithm using color local binary patterns (CLBP) and local variance features. Finally, to retrieve mostly closing matching images to the query, log likelihood ratio is used. The performance of the proposed approach is evaluated using Corel and Simplicity image sets and it compared by some of other well-known approaches in terms of precision and recall which shows the superiority of the proposed approach. Low noise sensitivity, rotation invariant, shift invariant, gray scale invariant and low computational complexity are some of other advantages.
This document summarizes a research paper on face recognition using Gabor features and PCA. It begins with an introduction to face recognition and discusses challenges like lighting, pose, and orientation. It then describes how the proposed system uses Gabor wavelets for preprocessing to reduce variations from pose, lighting, etc. Principal component analysis (PCA) is used to extract low dimensional and discriminating feature vectors from the preprocessed images. These feature vectors are then used for classification with k-nearest neighbors. The proposed system was tested on the Yale face database containing 100 images of 10 subjects with variable illumination and expressions.
Single frontal face detection by finding dark pixel group and comparing xyIAEME Publication
The document describes a method for detecting single frontal faces in color images. The method first detects skin color regions using a skin color model. It then finds connected components within the skin regions and tests each one for facial features. Facial features like eyes, nose, and mouth are extracted by analyzing the locations and shapes of dark pixel groups and comparing the relative positions of the features. Faces are identified by drawing lines between the features and comparing the shapes of the resulting triangles. The detected face region is enclosed within a rectangle defined by the feature locations. The method was tested on images containing faces with variations in lighting, features, and expressions.
International Journal of Engineering Research and Applications (IJERA) is an open access online peer reviewed international journal that publishes research and review articles in the fields of Computer Science, Neural Networks, Electrical Engineering, Software Engineering, Information Technology, Mechanical Engineering, Chemical Engineering, Plastic Engineering, Food Technology, Textile Engineering, Nano Technology & science, Power Electronics, Electronics & Communication Engineering, Computational mathematics, Image processing, Civil Engineering, Structural Engineering, Environmental Engineering, VLSI Testing & Low Power VLSI Design etc.
This document discusses principal component analysis (PCA) for face recognition. It begins with an introduction to face recognition and PCA. PCA works by calculating eigenvectors from a set of face images, which represent the principal components that account for the most variance in the image data. These eigenvectors are called "eigenfaces" and can be used to reconstruct the face images. The document then discusses how the system is implemented, including preparing a face database, normalizing the training images, calculating the eigenfaces/principal components, projecting the face images into this reduced space, and recognizing faces by calculating distances between projected test images and training images.
The document discusses challenges and approaches for facial emotion recognition. It aims to develop a model-based approach for real-time driver emotion recognition on an embedded platform using parallel processing. Model-based approaches can overcome issues like illumination and pose variations. The document reviews several state-of-the-art methods and discusses challenges like occlusion, lighting distortions, and complex backgrounds. It describes exploring both 2D and 3D techniques for facial feature extraction and expression recognition.
Fourier mellin transform based face recognitioniaemedu
This document presents a face recognition algorithm based on Fourier Mellin Transform. It begins with an introduction to face recognition and challenges of illumination and pose variations. It then describes extracting illumination invariant features by computing depth maps from input images using a shape from shading algorithm. Fourier Mellin Transform is applied to the depth maps to extract features. Experiments on the ORL database showed the approach achieved 100% recognition with 4 training images and 95.7% recognition with 3 training images, demonstrating robustness to illumination and pose variations.
This document summarizes a research paper on a Fourier Mellin transform based face recognition algorithm. The algorithm extracts illumination invariant features from depth maps of face images obtained through a shape from shading algorithm. These features are transformed using Fourier Mellin transform and classified using k-NN classification based on L1 norm distance. Experiments on the ORL database showed the transformed shape features are robust to illumination and pose variations, achieving better recognition performance compared to other algorithms. The document provides background on challenges in face recognition and reviews existing algorithms before describing the proposed Fourier Mellin transform based approach.
Face Detection for identification of people in Images of Internetijceronline
One method for searching the internet faces in images is proposed by using digital processing topological with descriptors. Location in real time with the development of a database that stores addresses of internet downloaded images, in which the search is done by text, but by finding facial image, is achieved. Face recognition in images of Internet has proved to be a difficult task, because the images vary considerably depending on viewpoint, illumination, expression, pose, accessories, etc. The descriptors for general information: containing low-level descriptors. Developments on face recognition systems have improved significantly since the first system; image analysis is a topic on which much emphasis is being given in order to identify parameters, visual features in the image that provide environment data that it is represented in the image, but without the intervention of a person. In this project raises its realization using the method of viola and jones as face descriptor. We can distinguish even different parts of the face such as eyes, eyebrows, nose and mouth.One method for searching faces in image taken from internet intends to use digital processing using topological descriptors. It is located the face in real time.
LOCAL REGION PSEUDO-ZERNIKE MOMENT- BASED FEATURE EXTRACTION FOR FACIAL RECOG...aciijournal
In the domain of image processing, face recognition is one of the most well-known research field. When
humans have very similar biometric properties, such as identical twins, the face recognition system is
considered as a challengeable problem. In this paper, the AdaBoost method is utilized to detect the
facial area of input image.
Medoid based model for face recognition using eigen and fisher facesijscmcj
Biometric technologies have gained a remarkable impetus in high security applications. Various biometric modalities are widely being used these days. The need for unobtrusive biometric recognition can be fulfilled through Face recognition which is the most natural and non intrusive authentication system. However the vulnerability to changes owing to variations in face due to various factors like pose,
illumination, ageing, emotions, expressions etc make it necessary to have robust face recognition systems.
Various statistical models have been developed so far with varying degree of accuracy and efficiency. This
paper discusses a new approach to utilize Eigen face and Fisher face methodology by using medoid instead
of mean as a statistic in calculating the Eigen faces and Fisher faces. The method not only requires lesser training but also demonstrates better time efficiency and performance compared to the conventional method of using mean
Face skin color based recognition using local spectral and gray scale featureseSAT Journals
Abstract Human face conveys more information about identity of person. Human face recognition is one of the most challenging problem and it can be used in many applications at different security places in airports, defense and banking sectors etc.In this work used colored features obtained from color segmentation because in real time scenario color provides the more information than gray scale image but it has a drawback. To overcome this drawback gray scale feature extracted from co-occurrence matrix of an image and for efficient face recognition of human Face under different illumination conditions spectral features can be extracted from face texture. These three feature vectors concatenated into a single feature vector and applied Lenc-Kral matching technique to measure similarity between the database and query image, the similarity is high then face is recognized. Keywords: Face recognition, illumination condition, local texture features, color segmentation.
A survey on human face recognition invariant to illuminationIAEME Publication
This document provides a survey of techniques for human face recognition that are invariant to illumination. It begins with an introduction to face recognition and the challenges posed by illumination variation. It then describes several preprocessing techniques that have been used to address illumination issues, including histogram equalization, adaptive histogram equalization, discrete cosine transform, discrete wavelet transform, and local binary pattern methods. The document concludes by comparing the performance of these various preprocessing techniques on standard face recognition databases based on error rates reported in other studies.
An efficient feature extraction method with pseudo zernike moment for facial ...ijcsity
Face recognition is one of the most challenging problems in the domain of image processing and machine
vision. Face recognition system is critical when individuals have very similar biometric signature such as
identical twins. In this paper, new efficient facial-based identical twins recognition is proposed according
to the geometric moment. The utilized geometric moment is Pseudo-Zernike Moment (PZM) as a feature
extractor inside the facial area of identical twins images. Also, the facial area inside an image is detected
using Ada Boost approach. The proposed method is evaluated on two datasets, Twins Days Festival and
Iranian Twin Society which contain scaled, which contain the shifted and rotated facial images of identical
twins in different illuminations. The results prove the ability of proposed method to recognize a pair of
identical twins. Also, results show that the proposed method is robust to rotation, scaling and changing
illumination.
Pupil Detection Based on Color Difference and Circular Hough Transform IJECEIAES
Human pupil eye detection is a significant stage in iris segmentation which is representing one of the most important steps in iris recognition. In this paper, we present a new method of highly accurate pupil detection. This method is consisting of many steps to detect the boundary of the pupil. First, the read eye image (R, G, B), then determine the work area which is consist of many steps to detect the boundary of the pupil. The determination of the work area contains many circles which are larger than pupil region. The work area is necessary to determine pupil region and neighborhood regions afterward the difference in color and intensity between pupil region and surrounding area is utilized, where the pupil region has color and intensity less than surrounding area. After the process of detecting pupil region many steps on the resulting image is applied in order to concentrate the pupil region and delete the others regions by using many methods such as dilation, erosion, canny filter, circle hough transforms to detect pupil region as well as apply optimization to choose the best circle that represents the pupil area. The proposed method is applied for images from palacky university, it achieves to 100 % accuracy
An Assimilated Face Recognition System with effective Gender Recognition RateIRJET Journal
This document summarizes an assimilated face recognition system that can also perform gender recognition. The system conducts experiments using databases like GENDER-FERET and Cambridge AT&T. For face recognition, it uses the Eigenfaces algorithm to extract features and classify faces. For gender recognition, it uses a trainable COSFIRE filter with Gabor filters to obtain face descriptors, which are classified using an SVM classifier. The experiments achieve a gender recognition rate of over 90%. The paper shows that the gender recognition approach outperforms other methods using handcrafted features and raw pixels.
HVDLP : HORIZONTAL VERTICAL DIAGONAL LOCAL PATTERN BASED FACE RECOGNITION sipij
Face image is an efficient biometric trait to recognize human beings without expecting any co-operation from a person. In this paper, we propose HVDLP: Horizontal Vertical Diagonal Local Pattern based face recognition using Discrete Wavelet Transform (DWT) and Local Binary Pattern (LBP). The face images of different sizes are converted into uniform size of 108×990and color images are converted to gray scale images in pre-processing. The Discrete Wavelet Transform (DWT) is applied on pre-processed images and LL band is obtained with the size of 54×45. The Novel concept of HVDLP is introduced in the proposed method to enhance the performance. The HVDLP is applied on 9×9 sub matrix of LL band to consider HVDLP coefficients. The local Binary Pattern (LBP) is applied on HVDLP of LL band. The final features are generated by using Guided filters on HVDLP and LBP matrices. The Euclidean Distance (ED) is used to compare final features of face database and test images to compute the performance parameters.
ABSTRACT: a rigorous work on static and dynamic appearance based classification systems for face is on but, it is proving to be a challenging task for researchers to design a proper system since human face is complex one. Decades of work was and is focussed on how to classify a face and on how to increase the rate of classification but, little attention was paid to overcome redundancy in image classification. This paper presents a novel idea which focuses on redundancy check and its elimination. The paper after drawing inferences from previous work gives out a novel idea for exact face classification and elimination of redundancy.
EV-SIFT - An Extended Scale Invariant Face Recognition for Plastic Surgery Fa...IJECEIAES
This paper presents a new technique called Entropy based SIFT (EV-SIFT) for accurate face recognition after the plastic surgery. The corresponding feature extracts the key points and volume of the scale-space structure for which the information rate is determined. This provides least effect on uncertain variations in the face since the entropy is the higher order statistical feature. The corresponding EV-SIFT features are applied to the Support vector machine for classification. The normal SIFT feature extracts the key points based on the contrast of the image and the V- SIFT feature extracts the key points based on the volume of the structure. However, the EV- SIFT method provides both the contrast and volume information. Thus EV-SIFT provide better performance when compared with PCA, normal SIFT and VSIFT based feature extraction.
Extraction of texture features by using gabor filter in wheat crop disease de...eSAT Journals
This document discusses a method for detecting diseases in wheat crops using image processing and artificial neural networks. It involves taking digital images of wheat crop leaves and preprocessing the images by applying Gaussian and median filters to reduce noise. The images are then segmented using CIELAB color space. Texture features like area, perimeter, contrast, and energy are extracted from the images using Gabor filters. These features are then fed into an artificial neural network classifier to identify the type of disease present in the wheat crop. The method aims to help farmers more quickly and accurately detect diseases so they can better manage their crops and increase agricultural productivity.
The document describes a novel method for localizing the pupil in eye gaze tracking systems. The proposed method fuses existing pupil localization techniques, including Hough transform, gray projection, and coarse positioning. This fusion approach aims to improve localization accuracy while lowering false detections. The method first preprocesses the eye image using filtering, edge detection, and thresholding/binarization. It then applies the fused techniques of Hough transform, gray projection, and coarse positioning to the preprocessed image to accurately detect the pupil boundary and center coordinates in a robust manner. The results are expected to provide better pupil localization compared to existing individual techniques.
Binary operation based hard exudate detection and fuzzy based classification ...IJECEIAES
Diabetic retinopathy (DR) is one of the most considerable reasons for visual impairment. The main objective of this paper is to automatically detect and recognize DR lesions like hard exudates, as it helps in diagnosing and screening of the disease. Here, binary operation based image processing for detecting lesions and fuzzy logic based extraction of hard exudates on diabetic retinal images are discused. In the initial stage, the binary operations are used to identify the exudates. Similarly, the RGB channel space of the DR image is used to create fuzzy sets and membership functions for extracting the exudates. The membership directives obtained from the fuzzy rule set are used to detect the grade of exudates. In order to evaluate the proposed approach, experiment tests are carriedout on various set of images and the results are verified. From the experiment results, the sensitivity obtained is 98.10%, specificity is 96.96% and accuracy is 98.2%. These results suggest that the proposed method could be a diagnostic aid for ophthalmologists in the screening for DR.
Content Based Image Retrieval Approach Based on Top-Hat Transform And Modifie...cscpconf
In this paper a robust approach is proposed for content based image retrieval (CBIR) using texture analysis techniques. The proposed approach includes three main steps. In the first one, shape detection is done based on Top-Hat transform to detect and crop object part of the image. Second step is included a texture feature representation algorithm using color local binary patterns (CLBP) and local variance features. Finally, to retrieve mostly closing matching images to the query, log likelihood ratio is used. The performance of the proposed approach is evaluated using Corel and Simplicity image sets and it compared by some of other well-known approaches in terms of precision and recall which shows the superiority of the proposed approach. Low noise sensitivity, rotation invariant, shift invariant, gray scale invariant and low computational complexity are some of other advantages.
This document summarizes a research paper on face recognition using Gabor features and PCA. It begins with an introduction to face recognition and discusses challenges like lighting, pose, and orientation. It then describes how the proposed system uses Gabor wavelets for preprocessing to reduce variations from pose, lighting, etc. Principal component analysis (PCA) is used to extract low dimensional and discriminating feature vectors from the preprocessed images. These feature vectors are then used for classification with k-nearest neighbors. The proposed system was tested on the Yale face database containing 100 images of 10 subjects with variable illumination and expressions.
Single frontal face detection by finding dark pixel group and comparing xyIAEME Publication
The document describes a method for detecting single frontal faces in color images. The method first detects skin color regions using a skin color model. It then finds connected components within the skin regions and tests each one for facial features. Facial features like eyes, nose, and mouth are extracted by analyzing the locations and shapes of dark pixel groups and comparing the relative positions of the features. Faces are identified by drawing lines between the features and comparing the shapes of the resulting triangles. The detected face region is enclosed within a rectangle defined by the feature locations. The method was tested on images containing faces with variations in lighting, features, and expressions.
International Journal of Engineering Research and Applications (IJERA) is an open access online peer reviewed international journal that publishes research and review articles in the fields of Computer Science, Neural Networks, Electrical Engineering, Software Engineering, Information Technology, Mechanical Engineering, Chemical Engineering, Plastic Engineering, Food Technology, Textile Engineering, Nano Technology & science, Power Electronics, Electronics & Communication Engineering, Computational mathematics, Image processing, Civil Engineering, Structural Engineering, Environmental Engineering, VLSI Testing & Low Power VLSI Design etc.
This document discusses principal component analysis (PCA) for face recognition. It begins with an introduction to face recognition and PCA. PCA works by calculating eigenvectors from a set of face images, which represent the principal components that account for the most variance in the image data. These eigenvectors are called "eigenfaces" and can be used to reconstruct the face images. The document then discusses how the system is implemented, including preparing a face database, normalizing the training images, calculating the eigenfaces/principal components, projecting the face images into this reduced space, and recognizing faces by calculating distances between projected test images and training images.
The document discusses challenges and approaches for facial emotion recognition. It aims to develop a model-based approach for real-time driver emotion recognition on an embedded platform using parallel processing. Model-based approaches can overcome issues like illumination and pose variations. The document reviews several state-of-the-art methods and discusses challenges like occlusion, lighting distortions, and complex backgrounds. It describes exploring both 2D and 3D techniques for facial feature extraction and expression recognition.
Fourier mellin transform based face recognitioniaemedu
This document presents a face recognition algorithm based on Fourier Mellin Transform. It begins with an introduction to face recognition and challenges of illumination and pose variations. It then describes extracting illumination invariant features by computing depth maps from input images using a shape from shading algorithm. Fourier Mellin Transform is applied to the depth maps to extract features. Experiments on the ORL database showed the approach achieved 100% recognition with 4 training images and 95.7% recognition with 3 training images, demonstrating robustness to illumination and pose variations.
This document summarizes a research paper on a Fourier Mellin transform based face recognition algorithm. The algorithm extracts illumination invariant features from depth maps of face images obtained through a shape from shading algorithm. These features are transformed using Fourier Mellin transform and classified using k-NN classification based on L1 norm distance. Experiments on the ORL database showed the transformed shape features are robust to illumination and pose variations, achieving better recognition performance compared to other algorithms. The document provides background on challenges in face recognition and reviews existing algorithms before describing the proposed Fourier Mellin transform based approach.
This document presents a face recognition algorithm based on Fourier Mellin Transform. It begins with an introduction to face recognition and challenges of illumination and pose variations. It then describes extracting illumination invariant features by computing depth maps from input images using a shape from shading algorithm. Fourier Mellin Transform is applied to the depth maps to extract features. Experiments on the ORL database showed the approach achieved 100% recognition with 4 training images and 95.7% recognition with 3 training images, demonstrating robustness to illumination and pose variations.
Unimodal Multi-Feature Fusion and one-dimensional Hidden Markov Models for Lo...IJECEIAES
The objective of low-resolution face recognition is to identify faces from small size or poor quality images with varying pose, illumination, expression, etc. In this work, we propose a robust low face recognition technique based on one-dimensional Hidden Markov Models. Features of each facial image are extracted using three steps: firstly, both Gabor filters and Histogram of Oriented Gradients (HOG) descriptor are calculated. Secondly, the size of these features is reduced using the Linear Discriminant Analysis (LDA) method in order to remove redundant information. Finally, the reduced features are combined using Canonical Correlation Analysis (CCA) method. Unlike existing techniques using HMMs, in which authors consider each state to represent one facial region (eyes, nose, mouth, etc), the proposed system employs 1D-HMMs without any prior knowledge about the localization of interest regions in the facial image. Performance of the proposed method will be measured using the AR database.
A New Approach of Iris Detection and RecognitionIJECEIAES
This paper proposes an IRIS recognition and detection model for measuring the e-security. This proposed model consists of the following blocks: segmentation and normalization, feature encoding and feature extraction, and classification. In first phase, histogram equalization and canny edge detection is used for object detection. And then, Hough Transformation is utilized for detecting the center of the pupil of an IRIS. In second phase, Daugmen’s Rubber Sheet model and Log Gabor filter is used for normalization and encoding and as a feature extraction method GNS (Global Neighborhood Structure) map is used, finally extracted feature of GNS is feed to the SVM (Support Vector Machine) for training and testing. For our tested dataset, experimental results demonstrate 92% accuracy in real portion and 86% accuracy in imaginary portion for both eyes. In addition, our proposed model outperforms than other two conventional methods exhibiting higher accuracy.
SYMMETRICAL WEIGHTED SUBSPACE HOLISTIC APPROACH FOR EXPRESSION RECOGNITIONijcsit
This document proposes a new method called Symmetrical Weighted 2D Principal Component Analysis (SW2DPCA) for facial expression recognition. SW2DPCA aims to reduce the dimensionality of the feature space extracted from face images using Gabor filters, while removing redundant information. It does this by decomposing the training images into odd and even symmetrical components, and applying weighted PCA to equalize the variance of principal components. The method is tested on the JAFFE database, achieving a recognition accuracy of 95.24% - higher than the 2DPCA baseline method.
The document discusses appearance-based face recognition using PCA and LDA algorithms. It summarizes the steps of each algorithm and compares their performance on preprocessed face images from the Faces94 database. Image preprocessing techniques like grayscale conversion and modified histogram equalization are applied before PCA and LDA to enhance image quality and improve recognition rates. The paper aims to study PCA and LDA with respect to recognition accuracy and dimensionality.
A Novel Mathematical Based Method for Generating Virtual Samples from a Front...CSCJournals
This paper deals with one sample face recognition which is a new challenging problem in pattern recognition. In the proposed method, the frontal 2D face image of each person divided to some sub-regions. After computing the 3D shape of each sub-region, a fusion scheme is applied on sub-regions to create a total 3D shape for whole face image. Then, 2D face image is added to the corresponding 3D shape to construct 3D face image. Finally by rotating the 3D face image, virtual samples with different views are generated. Experimental results on ORL dataset using nearest neighbor as classifier reveal an improvement about 5% in recognition rate for one sample per person by enlarging training set using generated virtual samples. Compared with other related works, the proposed method has the following advantages: 1) only one single frontal face is required for face recognition and the outputs are virtual images with variant views for each individual 2) need only 3 key points of face (eyes and nose) 3) 3D shape estimation for generating virtual samples is fully automatic and faster than other 3D reconstruction approaches 4) it is fully mathematical with no training phase and the estimated 3D model is unique for each individual.
IRJET - Facial Recognition based Attendance System with LBPHIRJET Journal
This document presents a facial recognition based attendance system using LBPH (Local Binary Pattern Histograms). It begins with an abstract describing the system which takes student attendance using facial identification from classroom camera images. It then discusses related work in attendance and face recognition systems. The proposed system workflow is described involving face detection, feature extraction using LBPH, template matching, and attendance recording. Experimental results demonstrate the system's ability to detect multiple faces and record attendance accurately in an Excel sheet with date/time. The conclusion discusses how the system reduces human effort for attendance and increases learning time compared to traditional methods.
A Face Recognition Using Linear-Diagonal Binary Graph Pattern Feature Extract...ijfcstjournal
This document presents a new face recognition method called Linear-Diagonal Binary Graph Pattern (LD-BGP) feature extraction. LD-BGP operates on image blocks and assigns an 8-bit code to each pixel by considering its linear and diagonal neighbors. This captures both straight and diagonal connectivity. Testing on the ORL face database achieved an accuracy of 95.3% for identification using Euclidean distance matching. The method was shown to be effective even with variation in lighting, expressions and other factors. Face recognition has applications in access control, surveillance, and other areas that require verifying or identifying individuals from images.
A FACE RECOGNITION USING LINEAR-DIAGONAL BINARY GRAPH PATTERN FEATURE EXTRACT...ijfcstjournal
Face recognition is one the most interesting topic in the field in computer vision and image processing.
Face recognition is a processing system that recognizes and identifies individuals human by their faces.
Automatic face recognition is powerful way to provide, authorized access to control their system. Face
recognition has many challenging problems (like face pose, face expression variation, illumination
variation, face orientation and noise) in the field of image analysis and computer vision. This method is
work on feature extraction part of face recognition. New way to extract face feature using LD-BGP code
operator it is like LGS and LBP feature extraction operator. In our LD-BGP-code operator work in two
direction first linear then diagonal. In both direction, its create eight digits code to every pixel of image.
Means of these two directional are taken so that is cover all neighbor of center pixel. First linear direction,
only horizontal and vertical pixel are taken. Second diagonal direction only diagonal pixels taken. In
matching phase, we use Euclidean distance to match a face image. We perform the Linear and diagonal
directional operator method on face database ORL. We get accuracy 95.3 %. LD-BGP method also works
on different type image like illuminated and expression variation image.
A FACE RECOGNITION USING LINEAR-DIAGONAL BINARY GRAPH PATTERN FEATURE EXTRACT...ijfcstjournal
Face recognition is one the most interesting topic in the field in computer vision and image processing.
Face recognition is a processing system that recognizes and identifies individuals human by their faces.
Automatic face recognition is powerful way to provide, authorized access to control their system. Face
recognition has many challenging problems (like face pose, face expression variation, illumination
variation, face orientation and noise) in the field of image analysis and computer vision. This method is
work on feature extraction part of face recognition. New way to extract face feature using LD-BGP code
operator it is like LGS and LBP feature extraction operator. In our LD-BGP-code operator work in two
direction first linear then diagonal. In both direction, its create eight digits code to every pixel of image.
Means of these two directional are taken so that is cover all neighbor of center pixel. First linear direction,
only horizontal and vertical pixel are taken. Second diagonal direction only diagonal pixels taken. In
matching phase, we use Euclidean distance to match a face image. We perform the Linear and diagonal
directional operator method on face database ORL. We get accuracy 95.3 %. LD-BGP method also works
on different type image like illuminated and expression variation image.
Local Region Pseudo-Zernike Moment- Based Feature Extraction for Facial Recog...aciijournal
In the domain of image processing, face recognition is one of the most well-known research field. When
humans have very similar biometric properties, such as identical twins, the face recognition system is
considered as a challengeable problem. In this paper, the AdaBoost method is utilized to detect the
facial area of input image. After that the facial area is divided into some local regions. Finally, new
efficient facial-based identical twins feature extractor based on the geometric moment is applied into
local regions of face image. The used feature extractor is Pseudo-Zernike Moment (PZM) which is
employed inside the local regions of facial area of identical twins images. To evaluate the proposed
method, two datasets, Twins Days Festival and Iranian Twin Society, are collected where the datasets
includes scaled and rotated facial images of identical twins in different illuminations. The experimental
results demonstrates the ability of proposed method to recognize a pair of identical twins in
different situations such as rotation, scaling and changing illumination
Local Region Pseudo-Zernike Moment- Based Feature Extraction for Facial Recog...aciijournal
In the domain of image processing, face recognition is one of the most well-known research field. When humans have very similar biometric properties, such as identical twins, the face recognition system is considered as a challengeable problem. In this paper, the AdaBoost method is utilized to detect the facial area of input image. After that the facial area is divided into some local regions. Finally, new efficient facial-based identical twins feature extractor based on the geometric moment is applied into local regions of face image. The used feature extractor is Pseudo-Zernike Moment (PZM) which is employed inside the local regions of facial area of identical twins images. To evaluate the proposed method, two datasets, Twins Days Festival and Iranian Twin Society, are collected where the datasets includes scaled and rotated facial images of identical twins in different illuminations. The experimental results demonstrates the ability of proposed method to recognize a pair of identical twins in different situations such as rotation, scaling and changing illumination
3D Face Modelling of Human Faces using GANsIRJET Journal
This document describes a model for 3D face modelling using GANs with less training data. Convolutional neural networks (CNNs) are used to extract features from 2D face images. These features are then used to generate 3D face models using generative adversarial networks (GANs). The model measures distances between facial features and calculates regional areas of faces from different views to capture spatial data needed for 3D modelling. Evaluation on datasets shows the model can generate realistic 3D face models with high accuracy, even with limited training data.
AN EFFICIENT FEATURE EXTRACTION METHOD WITH LOCAL REGION ZERNIKE MOMENT FOR F...ieijjournal
Face recognition is one of the most challenging problems in the domain of image processing and machine vision. The face recognition system is critical when individuals have very similar biometric signature such as identical twins. In this paper, the facial area in an image is detected using AdaBoost approach. After that the facial area is divided into some local regions. Finally, new efficient facial-based identical twins feature extractor based on the geometric moment is applied into local regions of face image.The utilized geometric moment is Zernike Moment (ZM) as a feature extractor inside the local regions of facial area of identical twins images. The proposed method is evaluated on two datasets, Twins Days Festival and Iranian Twin Society which contain scaled and rotated facial images of identical twins in different illuminations. The results prove the ability of proposed method to recognize a pair of identical twins.Also, results show that the proposed method is robust to rotation, scaling and changing illumination.
Similar to International Journal of Computer Science, Engineering and Information Technology (IJCSEIT) (20)
ANALYSIS OF EXISTING TRAILERS’ CONTAINER LOCK SYSTEMS IJCSEIT Journal
Trailers carry large containers to various destinations in the world. These are manually locked on to
trailers as they move through these long distances. Security mainly refers to the safety of a state,
organization, property, and individuals against criminal activity. The study was made to analyze the
existing trailer locks and the insecurity being experienced currently. The study also focused on creating a
background to building an automated lock system for auto-mobiles. Findings showed that there are various
container types like the General Purpose containers, the Hard-Top containers and the Open-Top among
others. Similarly, the twist locks were the ones revised for this study. The study also discussed the
weaknesses of the twist locks, most especially the non-notification on unsecured locks. This causes leads to
accidents and wastage of lives and property. The study finally proposed an automated lock system to
overcome these weaknesses to some good extent.
A MODEL FOR REMOTE ACCESS AND PROTECTION OF SMARTPHONES USING SHORT MESSAGE S...IJCSEIT Journal
The smartphone usage among people is increasing rapidly. With the phenomenal growth of smartphone
use, smartphone theft is also increasing. This paper proposes a model to secure smartphones from theft as
well as provides options to access a smartphone through other smartphone or a normal mobile via Short
Message Service. This model provides option to track and secure the mobile by locking it. It also provides
facilities to receive the incoming call and sms information to the remotely connected device and enables the
remote user to control the mobile through SMS. The proposed model is validated by the prototype
implementation in Android platform. Various tests are conducted in the implementation and the results are
discussed.
BIOMETRIC APPLICATION OF INTELLIGENT AGENTS IN FAKE DOCUMENT DETECTION OF JOB...IJCSEIT Journal
The Job selection process in today’s globally competitive economy can be a daunting task for prospective
employees no matter their experience level. Although many years of research has been devoted to job
search and application resulting in good integration with information technology including the internet and
intelligent agent-based architectures, there are still many areas that need to be enhanced. Two such areas
include the quality of jobs associated with applicants in the job search by profiling the needs of employers
against the needs of prospective employees and the security and verifications schemes integrated to reduce
the instances of fraud and identity theft. The integration of mobile, intelligent agent, and cryptography
technologies provide benefits such as improved accessibility wirelessly, intelligent dynamic profiling, and
increased security. With this in mind we propose the intelligent mobile agents instead of human agents to
perform the Job search using fuzzy preferences which is been published elsewhere and application
operations incorporating the use of agents with a trust authority to establish employer trust and validate
applicant identity and accuracy. Our proposed system incorporates design methodologies to use JADELEAP
and Android to provide a robust, secure, user friendly solution.
FACE RECOGNITION USING DIFFERENT LOCAL FEATURES WITH DIFFERENT DISTANCE TECHN...IJCSEIT Journal
A face recognition system using different local features with different distance measures is proposed in this
paper. Proposed method is fast and gives accurate detection. Feature vector is based on Eigen values,
Eigen vectors, and diagonal vectors of sub images. Images are partitioned into sub images to detect local
features. Sub partitions are rearranged into vertically and horizontally matrices. Eigen values, Eigenvector
and diagonal vectors are computed for these matrices. Global feature vector is generated for face
recognition. Experiments are performed on benchmark face YALE database. Results indicate that the
proposed method gives better recognition performance in terms of average recognized rate and retrieval
time compared to the existing methods.
BIOMETRICS AUTHENTICATION TECHNIQUE FOR INTRUSION DETECTION SYSTEMS USING FIN...IJCSEIT Journal
Identifying attackers is a major apprehension to both organizations and governments. Recently, the most
used applications for prevention or detection of attacks are intrusion detection systems. Biometrics
technology is simply the measurement and use of the unique characteristics of living humans to distinguish
them from one another and it is more useful as compare to passwords and tokens as they can be lost or
stolen so we have choose the technique biometric authentication. The biometric authentication provides the
ability to require more instances of authentication in such a quick and easy manner that users are not
bothered by the additional requirements. In this paper, we have given a brief introduction about
biometrics. Then we have given the information regarding the intrusion detection system and finally we
have proposed a method which is based on fingerprint recognition which would allow us to detect more
efficiently any abuse of the computer system that is running.
PERFORMANCE ANALYSIS OF FINGERPRINTING EXTRACTION ALGORITHM IN VIDEO COPY DET...IJCSEIT Journal
A video fingerprint is a recognizer that is derived from a piece of video content. The video fingerprinting
methods obtain unique features of a video that differentiates one video clip from another. It aims to identify
whether a query video segment is a copy of video from the video database or not based on the signature of
the video. It is difficult to find whether a video is a copied video or a similar video, since the features of the
content are very similar from one video to the other. The main focus of this paper is to detect that the query
video is present in the video database with robustness depending on the content of video and also by fast
search of fingerprints. The Fingerprint Extraction Algorithm and Fast Search Algorithms are adopted in
this paper to achieve robust, fast, efficient and accurate video copy detection. As a first step, the
Fingerprint Extraction algorithm is employed which extracts a fingerprint through the features from the
image content of video. The images are represented as Temporally Informative Representative Images
(TIRI). Then, the second step is to find the presence of copy of a query video in a video database, in which
a close match of its fingerprint in the corresponding fingerprint database is searched using inverted-filebased
method. The proposed system is tested against various attacks like noise, brightness, contrast,
rotation and frame drop. Thus the performance of the proposed system on an average shows high true
positive rate of 98% and low false positive rate of 1.3% for different attacks.
Effect of Interleaved FEC Code on Wavelet Based MC-CDMA System with Alamouti ...IJCSEIT Journal
In this paper, the impact of Forward Error Correction (FEC) code namely Trellis code with interleaver on
the performance of wavelet based MC-CDMA wireless communication system with the implementation of
Alamouti antenna diversity scheme has been investigated in terms of Bit Error Rate (BER) as a function of
Signal-to-Noise Ratio (SNR) per bit. Simulation of the system under proposed study has been done in M-ary
modulation schemes (MPSK, MQAM and DPSK) over AWGN and Rayleigh fading channel incorporating
Walsh Hadamard code as orthogonal spreading code to discriminate the message signal for individual
user. It is observed via computer simulation that the performance of the interleaved coded based proposed
system outperforms than that of the uncoded system in all modulation schemes over Rayleigh fading
channel.
FUZZY WEIGHTED ASSOCIATIVE CLASSIFIER: A PREDICTIVE TECHNIQUE FOR HEALTH CARE...IJCSEIT Journal
In this paper we extend the problem of classification using Fuzzy Association Rule Mining and propose the
concept of Fuzzy Weighted Associative Classifier (FWAC). Classification based on Association rules is
considered to be effective and advantageous in many cases. Associative classifiers are especially fit to
applications where the model may assist the domain experts in their decisions. Weighted Associative
Classifiers that takes advantage of weighted Association Rule Mining is already being proposed. However,
there is a so-called "sharp boundary" problem in association rules mining with quantitative attribute
domains. This paper proposes a new Fuzzy Weighted Associative Classifier (FWAC) that generates
classification rules using Fuzzy Weighted Support and Confidence framework. The naïve approach can be
used to generating strong rules instead of weak irrelevant rules. where fuzzy logic is used in partitioning
the domains. The problem of Invalidation of Downward Closure property is solved and the concept of
Fuzzy Weighted Support and Fuzzy Weighted Confidence frame work for Boolean and quantitative item
with weighted setting is generalized. We propose a theoretical model to introduce new associative classifier
that takes advantage of Fuzzy Weighted Association rule mining.
GENDER RECOGNITION SYSTEM USING SPEECH SIGNALIJCSEIT Journal
In this paper, a system, developed for speech encoding, analysis, synthesis and gender identification is
presented. A typical gender recognition system can be divided into front-end system and back-end system.
The task of the front-end system is to extract the gender related information from a speech signal and
represents it by a set of vectors called feature. Features like power spectrum density, frequency at
maximum power carry speaker information. The feature is extracted using First Fourier Transform (FFT)
algorithm. The task of the back-end system (also called classifier) is to create a gender model to recognize
the gender from his/her speech signal in recognition phase. This paper also presents the digital processing
of a speech signals (pronounced “A” and “B”) which are taken from 10 persons, 5 of them are Male and
the rest of them are Female. Power Spectrum Estimation of the signal is examined .The frequency at
maximum power of the English Phonemes is extracted from the estimated power spectrum. The system uses
threshold technique as identification tool. The recognition accuracy of this system is 80% on average.
DETECTION OF CONCEALED WEAPONS IN X-RAY IMAGES USING FUZZY K-NNIJCSEIT Journal
Scanning baggage by x-ray and analysing such images have become important technique for detecting
illicit materials in the baggage at Airports. In order to provide adequate security, a reliable and fast
screening technique is needed for baggage examination.This paper aims at providing an automatic method
for detecting concealed weapons, typically a gun in the baggage by employing image segmentation method
to extract the objects of interest from the image followed by applying feature extraction methods namely
Shape context descriptor and Zernike moments. Finally the objects are classified using fuzzy KNN as illicit
or non-illicit object.
META-HEURISTICS BASED ARF OPTIMIZATION FOR IMAGE RETRIEVALIJCSEIT Journal
The document proposes an approach combining automatic relevance feedback and particle swarm optimization for image retrieval. It constructs a visual feature database from image features like color moments and Gabor filters. For a query image, it retrieves similar images and generates automatic relevance feedback by labeling images as relevant or irrelevant. It then uses particle swarm optimization to re-weight features and retrieve more relevant images over multiple iterations, splitting the swarm in later iterations. An experiment on Corel images over 5 classes showed the approach could effectively retrieve relevant images through this meta-heuristic process without human interaction.
ERROR PERFORMANCE ANALYSIS USING COOPERATIVE CONTENTION-BASED ROUTING IN WIRE...IJCSEIT Journal
In Wireless Ad hoc network, cooperation of nodes can be achieved by more interactions at higher protocol
layers, particularly the MAC (Medium Access Control) and network layers play vital role. MAC facilitates
a routing protocol based on position location of nodes at network layer specially known as Beacon-less
geographic routing (BLGR) using Contention-based selection process. This paper proposes two levels of
cross-layer framework -a MAC network cross-layer design for forwarder selection (or routing) and a
MAC-PHY for relay selection. Wireless networks suffers huge number of communication at the same time
leads to increase in collision and energy consumption; hence focused on new Contention access method
that uses a dynamical change of channel access probability which can reduce the number of contention
times and collisions. Simulation result demonstrates the best Relay selection and the comparative of direct
mode with the cooperative networks. And also demonstrates the Performance evaluation of contention
probability with Collision avoidance.
M-FISH KARYOTYPING - A NEW APPROACH BASED ON WATERSHED TRANSFORMIJCSEIT Journal
Karyotyping is a process in which chromosomes in a dividing cell are properly stained, identified and
displayed in a standard format, which helps geneticist to study and diagnose genetic factors behind various
genetic diseases and for studying cancer. M-FISH (Multiplex Fluorescent In-Situ Hybridization) provides
color karyotyping. In this paper, an automated method for M-FISH chromosome segmentation based on
watershed transform followed by naive Bayes classification of each region using the features, mean and
standard deviation, is presented. Also, a post processing step is added to re-classify the small chromosome
segments to the neighboring larger segment for reducing the chances of misclassification. The approach
provided improved accuracy when compared to the pixel-by-pixel approach. The approach was tested on
40 images from the dataset and achieved an accuracy of 84.21 %.
Steganography is the technique of hiding a confidential message in an ordinary message and the extraction
of that secret message at its destination. Different carrier file formats can be used in steganography.
Among these carrier file formats, digital images are the most popular. For this work, digital images are
used. Here steganography is done on the skin portion of an image. First skin portion of an image is
detected. Random pixels are selected from that detected region using a pseudo-random number generator.
The bits of the secret message will be embedded on the LSB of these random pixels. An analysis is done to
check the efficiency and robustness of the proposed method. The aim of this work is to show that
steganography done using random pixel selection is less prone to outside attacks.
A NOVEL WINDOW FUNCTION YIELDING SUPPRESSED MAINLOBE WIDTH AND MINIMUM SIDELO...IJCSEIT Journal
In many applications like FIR filters, FFT, signal processing and measurements, we are required (~45 dB)
or less side lobes amplitudes. However, the problem is usual window based FIR filter design lies in its side
lobes amplitudes that are higher than the requirement of application. We propose a window function,
which has better performance like narrower main lobe width, minimum side lobe peak compared to the
several commonly used windows. The proposed window has slightly larger main lobe width of the
commonly used Hamming window, while featuring 6.2~22.62 dB smaller side lobe peak. The proposed
window maintains its maximum side lobe peak about -58.4~-52.6 dB compared to -35.8~-38.8 dB of
Hamming window for M=10~14, while offering roughly equal main lobe width. Our simulated results also
show significant performance upgrading of the proposed window compared to the Kaiser, Gaussian, and
Lanczos windows. The proposed window also shows better performance than Dolph-Chebyshev window.
Finally, the example of designed low pass FIR filter confirms the efficiency of the proposed window.
CSHURI – Modified HURI algorithm for Customer Segmentation and Transaction Pr...IJCSEIT Journal
Association rule mining (ARM) is the process of generating rules based on the correlation between the set
of items that the customers purchase.Of late, data mining researchers have improved upon the quality of
association rule mining for business development by incorporating factors like value (utility), quantity of
items sold (weight) and profit. The rules mined without considering utility values (profit margin) will lead
to a probable loss of profitable rules.
The advantage of wealth of the customers’ needs information and rules aids the retailer in designing his
store layout[9]. An algorithm CSHURI, Customer Segmentation using HURI, is proposed, a modified
version of HURI [6], finds customers who purchase high profitable rare items and accordingly classify the
customers based on some criteria; for example, a retail business may need to identify valuable customers
who are major contributors to a company’s overall profit. For a potential customer arriving in the store,
which customer group one should belong to according to customer needs, what are the preferred functional
features or products that the customer focuses on and what kind of offers will satisfy the customer, etc.,
finds the key in targeting customers to improve sales [9], which forms the base for customer utility mining.
AN EFFICIENT IMPLEMENTATION OF TRACKING USING KALMAN FILTER FOR UNDERWATER RO...IJCSEIT Journal
The exploration of oceans and sea beds is being made increasingly possible through the development of
Autonomous Underwater Vehicles (AUVs). This is an activity that concerns the marine community and it
must confront the existence of notable challenges. However, an automatic detecting and tracking system is
the first and foremost element for an AUV or an aqueous surveillance network. In this paper a method of
Kalman filter was presented to solve the problems of objects track in sonar images. Region of object was
extracted by threshold segment and morphology process, and the features of invariant moment and area
were analysed. Results show that the method presented has the advantages of good robustness, high
accuracy and real-time characteristic, and it is efficient in underwater target track based on sonar images
and also suited for the purpose of Obstacle avoidance for the AUV to operate in the constrained
underwater environment.
USING DATA MINING TECHNIQUES FOR DIAGNOSIS AND PROGNOSIS OF CANCER DISEASEIJCSEIT Journal
Breast cancer is one of the leading cancers for women in developed countries including India. It is the
second most common cause of cancer death in women. The high incidence of breast cancer in women has
increased significantly in the last years. In this paper we have discussed various data mining approaches
that have been utilized for breast cancer diagnosis and prognosis. Breast Cancer Diagnosis is
distinguishing of benign from malignant breast lumps and Breast Cancer Prognosis predicts when Breast
Cancer is to recur in patients that have had their cancers excised. This study paper summarizes various
review and technical articles on breast cancer diagnosis and prognosis also we focus on current research
being carried out using the data mining techniques to enhance the breast cancer diagnosis and prognosis.
FACTORS AFFECTING ACCEPTANCE OF WEB-BASED TRAINING SYSTEM: USING EXTENDED UTA...IJCSEIT Journal
Advancement in information system leads organizations to apply e-learning system to train their employees
in order to enhance its performance. In this respect, applying web based training will enable the
organization to train their employees quickly, efficiently and effectively anywhere at any time. This
research aims to extend Unified Theory of Acceptance and Use Technology (UTAUT) using some factors
such flexibility of web based training system, system interactivity and system enjoyment, in order to explain
the employees’ intention to use web based training system. A total of 290 employees have participated in
this study. The findings of the study revealed that performance expectancy, facilitating conditions, social
influence and system flexibility have direct effect on the employees’ intention to use web based training
system, while effort expectancy, system enjoyment and system interactivity have indirect effect on
employees’ intention to use the system.
PROBABILISTIC INTERPRETATION OF COMPLEX FUZZY SETIJCSEIT Journal
The document summarizes research on complex fuzzy sets. Complex fuzzy sets extend traditional fuzzy sets by allowing membership functions to range over the unit circle in the complex plane rather than just [0,1]. The paper defines complex fuzzy sets and operations like union, intersection, and complement. It presents examples and properties of these operations. It also introduces a probabilistic interpretation of complex fuzzy sets to distinguish them from probability.
Harnessing WebAssembly for Real-time Stateless Streaming PipelinesChristina Lin
Traditionally, dealing with real-time data pipelines has involved significant overhead, even for straightforward tasks like data transformation or masking. However, in this talk, we’ll venture into the dynamic realm of WebAssembly (WASM) and discover how it can revolutionize the creation of stateless streaming pipelines within a Kafka (Redpanda) broker. These pipelines are adept at managing low-latency, high-data-volume scenarios.
6th International Conference on Machine Learning & Applications (CMLA 2024)ClaraZara1
6th International Conference on Machine Learning & Applications (CMLA 2024) will provide an excellent international forum for sharing knowledge and results in theory, methodology and applications of on Machine Learning & Applications.
Hierarchical Digital Twin of a Naval Power SystemKerry Sado
A hierarchical digital twin of a Naval DC power system has been developed and experimentally verified. Similar to other state-of-the-art digital twins, this technology creates a digital replica of the physical system executed in real-time or faster, which can modify hardware controls. However, its advantage stems from distributing computational efforts by utilizing a hierarchical structure composed of lower-level digital twin blocks and a higher-level system digital twin. Each digital twin block is associated with a physical subsystem of the hardware and communicates with a singular system digital twin, which creates a system-level response. By extracting information from each level of the hierarchy, power system controls of the hardware were reconfigured autonomously. This hierarchical digital twin development offers several advantages over other digital twins, particularly in the field of naval power systems. The hierarchical structure allows for greater computational efficiency and scalability while the ability to autonomously reconfigure hardware controls offers increased flexibility and responsiveness. The hierarchical decomposition and models utilized were well aligned with the physical twin, as indicated by the maximum deviations between the developed digital twin hierarchy and the hardware.
Using recycled concrete aggregates (RCA) for pavements is crucial to achieving sustainability. Implementing RCA for new pavement can minimize carbon footprint, conserve natural resources, reduce harmful emissions, and lower life cycle costs. Compared to natural aggregate (NA), RCA pavement has fewer comprehensive studies and sustainability assessments.
ACEP Magazine edition 4th launched on 05.06.2024Rahul
This document provides information about the third edition of the magazine "Sthapatya" published by the Association of Civil Engineers (Practicing) Aurangabad. It includes messages from current and past presidents of ACEP, memories and photos from past ACEP events, information on life time achievement awards given by ACEP, and a technical article on concrete maintenance, repairs and strengthening. The document highlights activities of ACEP and provides a technical educational article for members.
Understanding Inductive Bias in Machine LearningSUTEJAS
This presentation explores the concept of inductive bias in machine learning. It explains how algorithms come with built-in assumptions and preferences that guide the learning process. You'll learn about the different types of inductive bias and how they can impact the performance and generalizability of machine learning models.
The presentation also covers the positive and negative aspects of inductive bias, along with strategies for mitigating potential drawbacks. We'll explore examples of how bias manifests in algorithms like neural networks and decision trees.
By understanding inductive bias, you can gain valuable insights into how machine learning models work and make informed decisions when building and deploying them.
Low power architecture of logic gates using adiabatic techniquesnooriasukmaningtyas
The growing significance of portable systems to limit power consumption in ultra-large-scale-integration chips of very high density, has recently led to rapid and inventive progresses in low-power design. The most effective technique is adiabatic logic circuit design in energy-efficient hardware. This paper presents two adiabatic approaches for the design of low power circuits, modified positive feedback adiabatic logic (modified PFAL) and the other is direct current diode based positive feedback adiabatic logic (DC-DB PFAL). Logic gates are the preliminary components in any digital circuit design. By improving the performance of basic gates, one can improvise the whole system performance. In this paper proposed circuit design of the low power architecture of OR/NOR, AND/NAND, and XOR/XNOR gates are presented using the said approaches and their results are analyzed for powerdissipation, delay, power-delay-product and rise time and compared with the other adiabatic techniques along with the conventional complementary metal oxide semiconductor (CMOS) designs reported in the literature. It has been found that the designs with DC-DB PFAL technique outperform with the percentage improvement of 65% for NOR gate and 7% for NAND gate and 34% for XNOR gate over the modified PFAL techniques at 10 MHz respectively.
DEEP LEARNING FOR SMART GRID INTRUSION DETECTION: A HYBRID CNN-LSTM-BASED MODELgerogepatton
As digital technology becomes more deeply embedded in power systems, protecting the communication
networks of Smart Grids (SG) has emerged as a critical concern. Distributed Network Protocol 3 (DNP3)
represents a multi-tiered application layer protocol extensively utilized in Supervisory Control and Data
Acquisition (SCADA)-based smart grids to facilitate real-time data gathering and control functionalities.
Robust Intrusion Detection Systems (IDS) are necessary for early threat detection and mitigation because
of the interconnection of these networks, which makes them vulnerable to a variety of cyberattacks. To
solve this issue, this paper develops a hybrid Deep Learning (DL) model specifically designed for intrusion
detection in smart grids. The proposed approach is a combination of the Convolutional Neural Network
(CNN) and the Long-Short-Term Memory algorithms (LSTM). We employed a recent intrusion detection
dataset (DNP3), which focuses on unauthorized commands and Denial of Service (DoS) cyberattacks, to
train and test our model. The results of our experiments show that our CNN-LSTM method is much better
at finding smart grid intrusions than other deep learning algorithms used for classification. In addition,
our proposed approach improves accuracy, precision, recall, and F1 score, achieving a high detection
accuracy rate of 99.50%.
CHINA’S GEO-ECONOMIC OUTREACH IN CENTRAL ASIAN COUNTRIES AND FUTURE PROSPECTjpsjournal1
The rivalry between prominent international actors for dominance over Central Asia's hydrocarbon
reserves and the ancient silk trade route, along with China's diplomatic endeavours in the area, has been
referred to as the "New Great Game." This research centres on the power struggle, considering
geopolitical, geostrategic, and geoeconomic variables. Topics including trade, political hegemony, oil
politics, and conventional and nontraditional security are all explored and explained by the researcher.
Using Mackinder's Heartland, Spykman Rimland, and Hegemonic Stability theories, examines China's role
in Central Asia. This study adheres to the empirical epistemological method and has taken care of
objectivity. This study analyze primary and secondary research documents critically to elaborate role of
china’s geo economic outreach in central Asian countries and its future prospect. China is thriving in trade,
pipeline politics, and winning states, according to this study, thanks to important instruments like the
Shanghai Cooperation Organisation and the Belt and Road Economic Initiative. According to this study,
China is seeing significant success in commerce, pipeline politics, and gaining influence on other
governments. This success may be attributed to the effective utilisation of key tools such as the Shanghai
Cooperation Organisation and the Belt and Road Economic Initiative.
CHINA’S GEO-ECONOMIC OUTREACH IN CENTRAL ASIAN COUNTRIES AND FUTURE PROSPECT
International Journal of Computer Science, Engineering and Information Technology (IJCSEIT)
1. International Journal of Computer Science, Engineering and Information Technology (IJCSEIT), Vol.3, No.2,April2013
DOI : 10.5121/ijcseit.2013.3204 51
FPGA ARCHITECTURE FOR FACIAL-FEATURES AND
COMPONENTS EXTRACTION
Nadia NACER1
, Bouraoui MAHMOUD2
, Mohamed Hedi BEDOUI3
1,3 Laboratory of Medical Technology and Imaging, Faculty of Medicine at Monastir,
University of Monastir – Tunisia
nd.nacer@gmail.com, Medhedi.bedoui@fmm.rnu.t
2 Laboratory of Medical Technology and Imaging (FMM-Monastir), National
Engineering School at Sousse,
University of Sousse-Tunisia
bouraoui.mahmoud@eniso.rnu.tn
ABSTRACT
Several methods for detecting the face and extracting the facial features and components exist in the
literature. These methods are different in their complexity, performance, type and nature of the images and
the targeted application. The facial features and components are used in security applications, robotics and
assistance for the disabled. We use these components and characteristics to determine the state of alertness
and fatigue for medical diagnoses. In this work we use plain color background images whose color is
different from the skin and which contain a single face. We are interested in FPGA implementation of this
application. This implementation must meet two constraints, which are the execution time and the FPGA
resources. We have selected and have associated a face detection algorithm based on the skin detection
(using the RGB space) with a facial-feature extraction algorithm based on tracking the gradient and the
geometric model.
KEYWORDS
Face detection, face components, face features, skin detection, RGB, gradient, implementation, FPGA
1. INTRODUCTION
The localization of facial features and components is an important step in many applications of
security (biometrics, surveillance), robotics, assistance for disabled (face communication) and
driving safety (detection of decreased alertness, fatigue). The location of these features allows the
facial expressions analysis. The facial expressions consist in a temporary distortion of a facial
structure. There are two types of facial structure: permanent (mouth, eyes, hair, deep wrinkles,
and brow) and transient (wrinkles, swelling). We use these components and features to determine
the state of alertness and fatigue for medical diagnoses.
The embedded systems for acquiring, processing and analyzing facial expressions require
increasingly complex algorithms. These algorithms require computational power and a large
memory space exceeding the possibilities offered by most conventional processors. The FPGA
(Field Programmable Gate Array) offers a solution that combines the programming flexibility and
with the specialized-architecture power. The implementation of application of acquiring,
processing and analyzing the facial expressions must meet a real-time execution (the camera
frame rate), while minimizing the resource consumption if we aim at low cost systems.
2. International Journal of Computer Science, Engineering and Information Technology (IJCSEIT), Vol.3, No.2,April2013
52
Several studies in the literature have proposed face detection implementation son processor and
on FPGA. The authors in [1] proposed a face detection implementation (Haar-classifier) on
FPGA (Viertx5Xilinxfamily) of images (320*240) with a time execution of 34.7ms
(28.8 frames/s). The same application implemented on a computer (Intel Core 2 CPU
(2.80 GHz), 2.98 GB DDR2 SDRAM (800 MHz), Windows XP Professional, and Visual Studio)
allows treating 0.71 frames/s [1]. The work done by [2] is a face detection parallel
implementation (Convolutional Face Finder) on FPGAs (Virtex 4) designing 32 Elementary
Processors (PE) that operate at a frequency of 350MHz. In [3] the authors implemented a face
detection algorithm (color model) for an image (800*600) on FPGA (Virtex II) which was used to
treat up to 90frames/s. In [4] the authors made a face detection implementation (color model) of
an image (176*144) on Altera APEX FPGA that could handle 434 frames/s. Several other studies
have conducted on FPGA implementations such as [5] (processing time is 200 ms for an image
(180*180) on FPGA Stratix II), [6] (processing speed is 50 frames/s ),[7] (face detection using
neural network for images (256*256) on Virtex FPGA with an 99.67% occupancy rate).
We remark that the face detection part with various methods is not an easy task. It requires
material resources with a varied execution time depending on the method used and the image
size. Our objective is implementing an application of localizing and extracting facial-features and
components on an FPGA circuit. We plan to exploit the algorithm parallelism and the memory
management by reducing the memory space and the execution time. In this work we are
interested in two main parts: face detection and interest-region localization (components) (eyes,
mouth and forehead). We have chosen and have combined several methods and algorithms that
have a reduced complexity and a maximum performance. We have validated these algorithms
using MATLAB Tool. Then we realize the description of these algorithms by VHDL hardware
language. Having made this choice we are also interested in the optimized implementation of
these algorithms on FPGA.
2. FACE DETECTION AND FACIAL COMPONENTS EXTRACTION ALGORITHM
Several methods for face detection and facial-feature extraction exist in the literature. These
methods are different in their complexity, performance, type and nature of the images and the
application. In this work we use plain color background images whose color is different from the
skin and witch contain a single face. The image size (n,m) is (192*256) (figure1).We have chosen
a face detection algorithm based on the skin detection (using RGB space) developed in
[8,9,10,11] and an algorithm of extracting the facial-features based on tracking the gradient
developed in [12] and the geometric model developed in [13].The method consists of two parts.
The first part is the face detection. We have chosen a simple algorithm of detection that does not
have a lot of calculations and constraints. The algorithm eliminates the regions that do not have
the skin color using the RGB space [8,9,10,11](Figure 1.b).This first part comprises the following
step: segmentation and binarization, determination of the coordinates of the face, and the face
extraction.
A pixel is said a skin pixel, if the components R (Red), G (Green) and B (Blue) of the color
satisfy the following conditions (segmentation and binarization) [8,9,10,11]:
( > 95) ( > 40) ( > 20)
((Max[ , , ] − Min[ , , ]) > 15)
( ( − ) > 15) ( > )
( > )
(1)
(0)
(1)
3. International Journal of Computer Science, Engineering and Information Technology (IJCSEIT), Vol.3, No.2,April2013
53
After selecting the skin pixels, we obtain a binary image whose various-pixel values are equal to
0 or 1.The principle of determining the coordinates of the face is to achieve a horizontal
projection (HIb) and a vertical one (VIb) of the binary image.
= ( , )
=1
= ( , )
=1
We search the positions of the first two values that are different from the zero of both sides of the
projection vectors HIb and VIb. These positions are the coordinates of the face in the original
image (L1, C1, L2, C2) (Figure 1.a).
The second part is the extraction of the facial-features. Several extraction methods use the color
information [14]. These methods have two limitations. They work only with color images, so
illumination changes can affect the results. There are some methods based on the gradient
information that are robust to illumination changes [15,16,17]. We have chosen a method
developed in [12]. The method is based on the fact that human faces are constructed in the same
geometrical configuration. To model the face with a geometric model, the method uses a gradient
tracking to locate each component of a frontal face on plain color background. Thus, it presents
robustness against small rotation angles, lighting conditions, skin color or accessories (mustache,
glasses, beards)[12]. We search through this method the extraction of two eyes, the mouth and the
forehead. The method uses a gradient tracking to locate each facial component [12]. By
exploiting the property of the face (since it is parameterized), it is easy to retrieve an object
characteristic referring to its coordinates in the face with regards to the center of the face and the
level of the eyes.
To locate the facial components, we first determine their horizontal axes, and then we determine
the area of each component by applying a geometric model [13]. This assumes that the vertical
distances between the eyes and the nose and between the eyes and the mouth are proportional to
the horizontal distance between the centers of the eyes [13]. Distances related to this model will
be described later. The proposed method comprises the following steps:
• Grayscale conversion (Figure 2.a),
The grayscale RGB space transformation algorithm is based on the following formula [14]:
= ∗ 0.299 + ∗ 0.587 + ∗ 0.114
Where I expresses the gray level for a given pixel, and R, G and B are the color components of
the pixel
• Calculating the gradient in the horizontal direction (Figure 2.c):
∇ =
Where y and j is the index and the horizontal direction, and I is the grayscale image.
• Eye level location (Neyes): realizing of the horizontal projection of the gradient image, and
searching for the maximum of the horizontal projection (Figure 2.d).
= ∇ ( , )
=1
= ( )
• location of the central axis (Ncenter): search for the position of the highest gray level on the
eye axis (Figure 2.b):
(2)
(3)
(4)
(5)
(6)
4. International Journal of Computer Science, Engineering and Information Technology (IJCSEIT), Vol.3, No.2,April2013
54
= (∇ ( , ))
• Location of the nose (Nnose): search for the highest gradient in a narrow window by
horizontal projection, below the eyes axis and around the central axis with a width of
pixels ∆x (Figure 2):
ose
= ∇ ( + + ∆ , )
= +∆
= −∆
= ( nose
)
• Location of the mouth (Nmouth): same principle as the previous step (Figure 2):
mouth
= ∇ ( + + ∆ , )
= +∆
= −∆
mouth = ( mouth
)
• Application of the geometric model (Figure 3).
The distance between the two-eye centers is D. The geometric face model (Figure 3) and the
related distances are described below [18]:
−The vertical distance between the center of the mouth and the two eyes is D.
−The vertical distance between the center of the nostrils and the two-eyes is 0.6D.
−The mouth width is D.
−The nose width is 0.8D.
−The vertical distance between the eyebrows and the eyes is 0.4D.
With D = Neyes – Nmouth
a. Original image (192*256) b. framing the face c. Detected face
(119*74)
Figure 1: Detection and extraction of the face
Once the eyes axis and the mouth axe are located, we apply the geometric model of the face [13].
Figure 3 shows the results obtained for the extraction of the facial components (eyes, forehead
and mouth).
The validation of these algorithms is carried out under MATLAB tool on a computer
(CPU: Intel Corp. Duo Frequency: 1.8 GHz RAM: 3 GB). Table 1 shows the execution time of
C1 C2
L1
L2
(7)
(8)
(9)
(10)
5. International Journal of Computer Science, Engineering and Information Technology (IJCSEIT), Vol.3, No.2,April2013
55
this method. The execution is done in a sequential manner on the computer processor; it explains
the high execution time (Table 1).
In the next section we use the VHDL language and FPGA circuit implementation to reduce
execution time of the application and optimize the memory space.
Figure2: Location of the axes of the eyes, the nostril, the mouth and the median
Figure 3: Application of the geometric model and extraction of the eyes, the mouth and the forehead.
Table 1: Execution time of the method
Algorithm Time (ms)
Face detection 261.8
Facial component extraction 23
Total 284.8
(3.5 images/s)
3. IMPLEMENTATION OF THE ALGORITHM OF FACE DETECTION AND
COMPONENTS EXTRACTION
The architecture of the application is composed of two blocks: a block of face detection and a
block of components extraction (interest regions) (Figure 4). We explore the possible execution
parallelism in algorithms to design the optimal architecture of the application. We designed an
architecture for each block. We realized the descriptions of these tow architecture in VHDL, the
simulation with ModelSim, the synthesis and implementation with Xilinx ISE tool.
OG_A1
OG_D1
OD_A1
OD_D1
B_A1
B_D1
F_A1
F_D1
050100150200
0
500
1000
1500
2000
2500
0 5 0 1 0 0 1 5 0
0
5 0
1 0 0
1 5 0
2 0 0
Left eye Right eye Mouth Forehead
D0.6D
0.4D
(a)
(b)
(c) (d)
6. International Journal of Computer Science, Engineering and Information Technology (IJCSEIT), Vol.3, No.2,April2013
56
Figure 4: Architecture of application
3.1. Face Detection
The description of this part for an implementation is shown in Figure 5.The architecture
comprises 8 blocks: two memory blocks (for the original image and the image gradient), a block
of reading the component pixels (R,G,B), a block of segmentation and binarization, a block of
search for line L1 and L2 using horizontal projection, a block of searching for columns C1 and
C2 using vertical projection, a block for face extraction and a block of grayscale to RGB
conversion.
Figure 5: Block diagram of the hardware architecture of the face detection part using the RGB component
Organization and size of the memory space for the face detection portion architecture is presented
in figure6.
Original image (color)
3 * (192*256)
Binary image
(192*256)
Horizontal projection
table
192
Original image (color)
3 * (192*256)
Vertical projection
table
256
Vertical projection
table 256
Face image (color)
3*(L2-L1)*(C2-C1)
Gray image of face
(L2-L1)*(C2-C1)
Gray image of face
(L2-L1)*(C2-C1)
Figure 6: Size of the memory space before and after optimization
Face
detection
Component
extraction
Input
image
Output
images
7. International Journal of Computer Science, Engineering and Information Technology (IJCSEIT), Vol.3, No.2,April2013
57
In this hardware description we have optimized the memory space and we have exploited the
parallelism of some functions in order to decrease the execution time.
3.1.1. Parallel execution of bloc
We proceed to perform in parallel the segmentation of the skin pixel algorithm, the research-line
(L1 and L2) algorithm and the search-column (C1 and C2) algorithm. The face-color extraction
(RGB) algorithm, the grayscale conversion algorithm and the in memory recording are performed
in parallel.
Figure 7: Architecture of the hardware description for selecting and segmenting regions of the skin
algorithm
for i = 1 … n do
for j = 1 … m do
if ((R(i,j) > 95) AND (G(i,j) > 40) AND (B(i,j) > 20)) AND
(Max(R(i,j),G(i,j),B(i,j))-Min(R(i,j),G(i,j),B(i,j)))>15) AND
( abs(R(i,j) - G(i,j)) > 15) AND (R(i,j) > G(i,j)) AND (R(i,j) >
B(i,j))then
S(i,j) = 1;
else
S(i,j) = 0;
End if
Enf for
End for
8. International Journal of Computer Science, Engineering and Information Technology (IJCSEIT), Vol.3, No.2,April2013
58
Figure 7 represents the hardware description of the algorithm for selecting and segmenting the
skin regions for an FPGA implementation. In this description we read the three components of the
pixel in parallel. We reserve 3 memories (each of size (192*256 bytes)) for the three matrices of
the original image. We realize the execution of the absolute value of (RG), the search for the
maximum of (R, G, B) and the minimum of (R, G, B) in parallel. Different comparisons are
performed in parallel.
3.1.2. Memory Optimization (figure 6)
Depending on the method used, we have 4 images: original color image (3 matrices (192 * 256),
binary image (one matrix (192 * 256)), RGB color image extracted from face
(3 matrices (L2-L1)*(C2-C1)) and gray level image (1 matrix (L2-L1)*(C2-C1)) : there is also an
array of size 192 bytes for the horizontal projection and another of size 256 bytes for the vertical
projection. The all the memory space is:
Memory space = 4 * (192 * 256) + 4 * (L2-L1) * (L2-L1) + 448 = 232280 bytes
Since the binary image is used only once, and since the determination of the coordinates L1, C1,
L2, C2 is in parallel, we do not need to save the binary image in memory. The horizontal
projection allows defining the two lines L1 and L2; therefore we have used a method to determine
these two indices without creating a memory table. Using the coordinates we extract the facial
image from the original image and for each read pixel, we execute the grayscale conversion. The
new memory space required for this party becomes:
Memory space = 3 * (192 * 256) + 1 * (L2 - L1) * (C2 - C1) + 256 bytes= 156518 bytes
3.2. Facial components extraction
The hardware description of this part for an implementation is shown in Figure 8. In this
description we have optimized the memory space by exploiting the parallelism of functions in
order to minimize the execution time. The architecture comprises 10 blocks: five memory blocks
(for gradient image, image of left eye, image of right eye, mouth image and forehead image), a
block for gradient calculation, a block for search for the eye level, a block for search for the
center, a block for search the nose and mouth level, a block for application of the geometric
model. The block of grayscale RGB pixel conversion and the memory block of grayscale image
are belonging to the first part. They are present only to show the complete data path in the second
part.
9. International Journal of Computer Science, Engineering and Information Technology (IJCSEIT), Vol.3, No.2,April2013
59
Figure 8: Block diagram of the components extraction hardware architecture
3.2.2. Parallel bloc execution
We have chosen that the memorization of grayscale calculated in the memory, the calculation of
the gradient and its storage in the memory and the eyes level localization are to be executed in
parallel.
Figure 9 shows the description of the grayscale RGB conversion function. Using the indices of
the lines (L1, L2) and the columns (C1, C2) of the face rectangle, we convert them in to addresses
and we read the components R, G and B of the three memories at the same time. We execute the
three multiplications in parallel (figure 9).
R G B
X X X
+
+
Coef 1 Coef 2 Coef 3
Y: pixel
niveau de fris
With coef1 = 0.299, coef2=0.587 and coef3= 0.114
Figure 9: Hardware description of the algorithm for grayscale RGB conversion of the face image
Figure 10 shows the description of the gradient calculation algorithm of the grayscale image. The
gradient of a pixel is the difference between the pixel (i,j+1) and the pixel (i,j). We have
calculated the gradient for each pixel converted in to a grayscale. We perform the gradient
calculation in parallel with the grayscale conversion. We apply a delay of the j pixel and then we
calculate the difference in P(i,j +1) and P(i,j) (with P is the pixel value).
10. International Journal of Computer Science, Engineering and Information Technology (IJCSEIT), Vol.3, No.2,April2013
60
P: grayscale
pixel
Delay
Pi Pi+1
-
Gradient
pixel
(n-1)*(m-1)
Figure 10: Description of the gradient calculation algorithm
3.2.2. Memory Optimization (Table 11):
In this section we have many memories: memory of gradient image ((L2-L1)*(C2-C1)) bytes,
memory for the horizontal projection of size (L2-L1) byte of the gradient image for localizing the
axis of the eyes, memory for the horizontal projection of size ((L2-L1)-Neyes) bytes of a part of the
gradient image for localizing the nose and the mouth, memory for the left eye image with (10*28)
bytes, memory for the right eye image with (10*28) bytes, memory for the mouth image with
(16*40) bytes, and memory for the front image with (20*68) bytes.
Once we have determined the eyes level, we determine the face center. We search for the face
center on the line that corresponds to the eye level in the grayscale image. Using the luminance
component face, the light intensity at this level is clearer (value of highest intensity) which
matches the center of the face. We need just to find the maximum intensity of the line. We use the
eye level and the face center to locate the levels of the nostril and the mouth. Then we take an
area of the gradient image. This area is centered on the median axis of the face. This space has a
width of±∆y of the center of the face and the length of
(((L2-L1)-Neyes)+∆y), with ∆y=20.This area includes the nose and the mouth. The choice of this
area reduces the effect of the beard, mustache and other structures in the face. The horizontal
projection of the area has two peaks. The first greater peak is for the nose level and the second
one is for the mouth. We need just to find these two peaks. Organization and size of the memory
space for the component extraction part are presented in figure 11.
11. International Journal of Computer Science, Engineering and Information Technology (IJCSEIT), Vol.3, No.2,April2013
61
gradient image
(L2-L1)*(C2-C1) 119*74
Horizontal projection of the
image gradient
(L2-L1) = 119
gradient image
(L2-L1)*(C2-C1)
(119*74)
Horizontal projection
(L2-L1) – (Neyes +∆x) = 56
Right eye image
(10*28)
Right eye image
(10*28)
Left eye image
(10*28)
Left eye image
(10*28)
Image of the mouth
(16*39)
Image of the mouth
(16*39)
Image of the forehead
(20*68)
Image of the forehead
(20*68)
Memory befor optimisation =
11045 bytes
Memory after optimisation
= 10870 bytes
Figure 11: Size of the memory space before and after optimization
3.2.3. Application of the geometric model
Once we have determined the eyes level indices, the nostril level, the mouth level and the center
of the face, we can determine the coordinates of the eyes, the mouth and the forehead while
applying the geometric model. This assumes that the vertical distances between the eyes and the
nose and between the eyes and mouth are proportional to the horizontal distance between the
centers of the eyes D. the distances related to this model are described in [13].
3.3. Results of the simulation, synthesis and implementation
To implement the algorithms already presented, a VHDL description is simulated and synthesized
to obtain an adequate description for the desired operation of the design system. We have used an
FPGA family Xilinx (Virtex 5 XC5VLX30). Material resources necessary for the implementation
of this application are given in Table 2. The results presented in Table 3 and 4 are given by the
simulation, synthesis and implementation on Xilinx ISE tool, and we have chosen as target FPGA
Virtex 5 XC5VLX30. We have not done a real physical implementation on FPGA, just a
simulation and emulation on the Xilinx ISE tool.
Table 2: Resource FPGA circuit for the implementation
FPGA Circuit Slice BlocRAM DSP
Virtex 5
XC5VLX30
1799 20 6
37.4% 31.2% 18.75%
12. International Journal of Computer Science, Engineering and Information Technology (IJCSEIT), Vol.3, No.2,April2013
62
Table 3 shows the memory space optimization in the application
Table 3: memory space optimization for the implementation
Algorithm Memory before
optimization
Memory after
optimization
gain
optimization
Face detection and
grayscale conversion
232280 bytes 156518 bytes 32.61%
Components extraction 11045 bytes 10870 bytes 1.58%
Total 243325 bytes 167388 bytes 31.2%
Table 4: shows the execution time of the application. This time allows us to treat 490 frames/s
with a clock frequency of 50MHz.
Table 4: Execution time of application (simulation result with a frequency of 50MHz)
Algorithm Execution time (en ms)
Face detection and grayscale
conversion
1.1646
Component extraction 0.8734
Total 2.038
(490 images/s)
Figure 12 shows the simulation results of the VHDL code architecture (left eye image, right eye
image, mouth image and forehead image) obtained by the developed architecture.
Figure 12: Images extraction Results of developed hardware architecture
IV. CONCLUSION
We have chosen and associated two algorithms that have reduced complexity with a maximum
performance. The face detection algorithm is based on the use of the color information. The
algorithm of localizing the facial components is based on the gradient which provides robustness
against illumination changes. The architectural description of the application developed in this
work allows us to process 490 frames/s with a frequency of 50MHz. This allows us to consider
the implementation of other algorithmic parts of analyzing the facial expressions. In addition, we
have performed an optimization of the memory space with a gain of 31.2%. This optimization is
obtained by exploiting the implementation parallelism offered by the FPGA circuit and the
execution parallelism that we have achieved in the algorithms of this application.
Left eye Right eye Mouth Forehead
13. International Journal of Computer Science, Engineering and Information Technology (IJCSEIT), Vol.3, No.2,April2013
63
REFERENCES
[1]Junguk Cho, ShahnamMirzaei, Jason Oberg, Ryan Kastner, (2009) “Fpga-based face detection system
using Haar classifiers” Proceeding of the ACM/SIGDA international symposium on Field programmable
gate arrays, California, USA, Pages 103-112.
[2]Nicolas Farrugia, Franck Mamalet, S´ebastien Roux, Michel Paindavoine, Fan Yang, (2007) “F
Implantation parallèle de la detection de visages : Méthodologie et implantation sur FPGA”. Colloque
GRETSI, pp 573-576, Troyes, 11-14 septembre 2007.
[3]Mohammad S. Sadri, Nasim Shams, MasihRahmaty, IrajHosseini, ReihaneChangiz,
ShahedMortazavian, ShimaKheradmand, RoozbehJafari, (2004) “An FPGA Based Fast Face Detector”,
GSP 2004 pp. 1039, 2004.
[4]Paschalakis, S., Bober, M., (2003) “A Low Cost FPGA System for High Speed Face Detection and
Tracking”, In Proc. 2003 2nd IEEE International Conference on Field Programmable Technology (FPT
’03), Tokyo, Japan, Dec. 15-17, pp. 214-221, 2003.
[5]L. PIERREFEU, J. JAY., (2007) “Détection et localisation de visage pour un système grand public
d’authentification de visage implantable sur FPGA”. Colloque GRETSI, Troyes, pp 361-364, 11-14
septembre 2007.
[6]T.Theocharides, G.Link, N.Vijaykrishnan, M.J.Irwin, and W.Wolf, (2004). “Embedded hardware face
detection” Proceedings of the 17th International Conference on VLSI Design (VLSID 04), 2004.
[7]Smach, .M Atri, J. Mitéran and M. Abid, (2006) “Design of a Neural Networks Classifier for Face
Detection” Journal of Computer Science 2 (3): 257-260, 2006.
[8]Garcia C. ,Tziritas G., (1999) “Face Detection Using Quantized Skin Color Regions Merging and
Wavelet Packet Analysis” IEEE Transactions on Multimedia, 1(3), p.264-277, September 1999.
[9]BencherietChemesse-Ennehar, BouallegAbd El halim, Tebbikh Hicham, (2007) “Segmentation de la
Couleur de Peau par Seuillage selon Différents Espaces de Couleur” JIG’2007 - 3èmes Journées
Internationales sur l’Informatique Graphique, pp 207-211, INRIA Sophia-Antipolis, France, 2007.
[10]F. Gasparini and R. Schettini, (2006) “Skin segmentation using multiple thresholding in Internet
Imaging” vol. 6061 of Proceedings of SPIE, pp. 1–8, San Jose, Calif, USA, January 2006.
[11]J. Kovac, P. Peer, and F. Solina, (2003) “2D versus 3D colour space face detection” in Proceedings of
the 4th EURASIP Conference on Video/Image Processing and Multimedia Communications, vol. 2, pp.
449–454, Zagreb, Croatia, July 2003.
[12]F.ABDAT, C.MAAOUI et A.PRUSKI, (2007) “Suivi du gradient pour la localisation des
caractéristiques faciales dans des images statiques” Colloque GRETSI, Troyes 11-14 septembre 2007.
[13]Frank Y. Shih, Chao-Fa Chuang, (2004) “Automatic extraction of head and face boundaries and facial
features” Information Sciences 158 pp. 117–130, 2004.
[14]Michael J. JONES and James M. REHG. (1999) “Statistical Color Models with Application to Skin
Detection” In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, volume 1,
page 1274, Fort Collins, Colorado, June 1999.
[15]Deng X., Chang C. H. and Brandle E. (2004) “A new method for eye extraction from facial image“.
Proc. 2nd IEEE international workshop on electronic design test and applications (DELTA), 2 NO.4:29–34,
Perth, Australia, 2004.
[16]Maio D. and Maltoni D. (2000) “Real-time face location on grayscale static images” Pattern
Recognition, 33:1525– 1539, 2000.
[17]Tsekeridou S. and Pitas I. (1998) “Facial feature extraction in frontal views using biometric analogies”
Proc. 9th European Signal Processing Conference, 1:315–318, September 8-11 Island of Rhodes, Greece,
1998.
[18]A. Nikolaidis, I. Pitas, (2000) “Facial feature extraction and pose determination” Pattern Recognition
33 (2000) 1783–1791.
14. International Journal of Computer Science, Engineering and Information Technology (IJCSEIT), Vol.3, No.2,April2013
64
Authors
Nadia NASR received the Bachelor’s and Master‘s Degree in the Faculty of
Science at Monastir of Manastir University, in 2007 and 2009 respectively. Since
2010, she has been working as a Research Scientist at the Research team of
Medical Imaging Technology (TIM) at Faculty of Medicine at Monastir of
Monastir University where she prepares his thesis. She has also two published
paper in international conference in the same areas. His areas of interest include
dedicated hardware architecture of image processing.
Bouraoui MAHMOUD received the Bachelor’s, Master‘s Degree and doctorate in
the Faculty of Science at Monastir of Monastir University, in 1997, 1999 and 2006
respectively. Since 2008, He is currently Assistant Professor in the Department of
Industrial Electronics in the National Engineering School of Sousse, University of
Sousse. He has six published papers in international journals. His areas of interest
include embedded processor, embedded system, image processing Hardware
architecture, codesign HW/SW.
Mohamed Hedi BEDOUI Since 2011, He is currently Professor in Biophysics in the Faculty of Medecine
at Monastir of Monastir University. He has many published papers in international journals. His areas of
interest include Biophysics, medical imaging processing, and embedded system, codesign HW/SW.