This document presents a method for classifying gender from human gait using silhouette images. The silhouette is divided into two regions, and features like moments and ellipse parameters are extracted from each region. These features are used to form a feature vector for each frame. Mean and standard deviation features are then calculated across frames. Different classifiers like K-Nearest Neighbor and Support Vector Machine are tested on the CASIA gait dataset, with SVM found to perform better, achieving a 93.33% recognition rate using linear kernels.
Hybrid Domain based Face Recognition using DWT, FFT and Compressed CLBPCSCJournals
The characteristics of human body parts and behaviour are measured with biometrics, which are used to authenticate a person. In this paper, we propose Hybrid Domain based Face Recognition using DWT, FFT and Compressed CLBP. The face images are preprocessed to enhance sharpness of images using Discrete Wavelet Transform (DWT) and Laplacian filter. The Compound Local Binary Pattern (CLBP) is applied on sharpened preprocessed face image to compute magnitude and sign components. The histogram is applied on CLBP components to compress number of features. The Fast Fourier Transformation (FFT) is applied on preprocessed image and compute magnitudes. The histogram features and FFT magnitude features are fused to generate final feature. The Euclidian Distance (ED) is used to compare final features of test face images with data base face images to compute performance parameters. It is observed that the percentage recognition rate is high in the case of proposed algorithm compared to existing algorithms.
IRJET- Detection of Static Body Poses using Skin Detection Techniques to Asse...IRJET Journal
This paper proposes a method to detect static body poses from images to assess a person's accessibility during human-robot interaction. It uses skin detection techniques like superpixels and the YCbCr color space to segment the skin regions. Five static poses are classified into categories of accessibility - not accessible, accessible, and neutral - using Hidden Markov Models. Geometric features of the segmented skin regions are used for classification. The results show the method can accurately categorize images into the different accessibility levels based on the body poses depicted. This could help robots interpret human body language and nonverbal cues during social interactions.
Human Segmentation Using Haar-ClassifierIJERA Editor
Segmentation is an important process in many aspects of multimedia applications. Fast and perfect segmentation of moving objects in video sequences is a basic task in many computer visions and video investigation applications. Particularly Human detection is an active research area in computer vision applications. Segmentation is very useful for tracking and recognition the object in a moving clip. The motion segmentation problem is studied and reviewed the most important techniques. We illustrate some common methods for segmenting the moving objects including background subtraction, temporal segmentation and edge detection. Contour and threshold are common methods for segmenting the objects in moving clip. These methods are widely exploited for moving object segmentation in many video surveillance applications, such as traffic monitoring, human motion capture. In this paper, Haar Classifier is used to detect humans in a moving video clip some features like face detection, eye detection, full body, upper body and lower body detection.
Robust Human Tracking Method Based on Apperance and Geometrical Features in N...csandit
This paper proposes a robust tracking method which concatenates appearance and geometrical
features to re-identify human in non-overlapping views. A uniformly-partitioning method is
proposed to extract local HSV(Hue, Saturation, Value) color features in upper and lower
portion of clothing. Then adaptive principal view selecting algorithm is presented to locate
principal view which contains maximum appearance feature dimensions captured from different
visual angles. For each appearance feature dimension in principal view, all its inner frames get
involved in training a support vector machine (SVM). In matching process, human candidate
filtering is first operated with an integrated geometrical feature which connects height estimate
with gait feature. The appearance features of the remaining human candidates are later tested
by SVMs to determine the object’s existence in new cameras. Experimental results show the
feasibility and effectiveness of this proposal and demonstrate the real-time in appearance
feature extraction and robustness to illumination and visual angle change.
The document discusses multispectral palm image fusion for biometric authentication using ant colony optimization. It introduces intra-modal fusion of palmprint images from multiple spectra to improve accuracy. The key steps involve detecting the region of interest, fusing the images using wavelet transforms, extracting Gabor features, selecting optimal features using ant colony optimization, and classifying with support vector machines. Experimental results and conclusions are also presented.
A Methodology for Extracting Standing Human Bodies from Single Imagesjournal ijrtem
Abstract: Extraction of the image of human body in unconstrained still images is challenging due to several factors, including shading, image noise, occlusions, background clutter, the high degree of human body deformability, and the unrestricted positions due to in and out of the image plane rotations. we propose a bottom-up approach for human body segmentation in static images. We decompose the problem into three sequential problems: Face detection, upper body extraction, and lower body extraction, since there is a direct pair wise correlation among them. Index Terms: Skin segmentation, Torso, Face recognition, Thresholding, Ethnicity, Morphology.
Face Recognition based on STWT and DTCWT using two dimensional Q-shift Filters IJERA Editor
The Biometrics is used to recognize a person effectively compared to traditional methods of identification. In this paper, we propose a Face recognition based on Single Tree Wavelet Transform (STWT) and Dual Tree Complex Wavelet transform (DTCWT). The Face Images are preprocessed to enhance quality of the image and resize. DTCWT and STWT are applied on face images to extract features. The Euclidian distance is used to compare features of database image with test face images to compute performance parameters. The performance of STWT is compared with DTCWT. It is observed that the DTCWT gives better results compared to STWT technique.
IJERA (International journal of Engineering Research and Applications) is International online, ... peer reviewed journal. For more detail or submit your article, please visit www.ijera.com
Hybrid Domain based Face Recognition using DWT, FFT and Compressed CLBPCSCJournals
The characteristics of human body parts and behaviour are measured with biometrics, which are used to authenticate a person. In this paper, we propose Hybrid Domain based Face Recognition using DWT, FFT and Compressed CLBP. The face images are preprocessed to enhance sharpness of images using Discrete Wavelet Transform (DWT) and Laplacian filter. The Compound Local Binary Pattern (CLBP) is applied on sharpened preprocessed face image to compute magnitude and sign components. The histogram is applied on CLBP components to compress number of features. The Fast Fourier Transformation (FFT) is applied on preprocessed image and compute magnitudes. The histogram features and FFT magnitude features are fused to generate final feature. The Euclidian Distance (ED) is used to compare final features of test face images with data base face images to compute performance parameters. It is observed that the percentage recognition rate is high in the case of proposed algorithm compared to existing algorithms.
IRJET- Detection of Static Body Poses using Skin Detection Techniques to Asse...IRJET Journal
This paper proposes a method to detect static body poses from images to assess a person's accessibility during human-robot interaction. It uses skin detection techniques like superpixels and the YCbCr color space to segment the skin regions. Five static poses are classified into categories of accessibility - not accessible, accessible, and neutral - using Hidden Markov Models. Geometric features of the segmented skin regions are used for classification. The results show the method can accurately categorize images into the different accessibility levels based on the body poses depicted. This could help robots interpret human body language and nonverbal cues during social interactions.
Human Segmentation Using Haar-ClassifierIJERA Editor
Segmentation is an important process in many aspects of multimedia applications. Fast and perfect segmentation of moving objects in video sequences is a basic task in many computer visions and video investigation applications. Particularly Human detection is an active research area in computer vision applications. Segmentation is very useful for tracking and recognition the object in a moving clip. The motion segmentation problem is studied and reviewed the most important techniques. We illustrate some common methods for segmenting the moving objects including background subtraction, temporal segmentation and edge detection. Contour and threshold are common methods for segmenting the objects in moving clip. These methods are widely exploited for moving object segmentation in many video surveillance applications, such as traffic monitoring, human motion capture. In this paper, Haar Classifier is used to detect humans in a moving video clip some features like face detection, eye detection, full body, upper body and lower body detection.
Robust Human Tracking Method Based on Apperance and Geometrical Features in N...csandit
This paper proposes a robust tracking method which concatenates appearance and geometrical
features to re-identify human in non-overlapping views. A uniformly-partitioning method is
proposed to extract local HSV(Hue, Saturation, Value) color features in upper and lower
portion of clothing. Then adaptive principal view selecting algorithm is presented to locate
principal view which contains maximum appearance feature dimensions captured from different
visual angles. For each appearance feature dimension in principal view, all its inner frames get
involved in training a support vector machine (SVM). In matching process, human candidate
filtering is first operated with an integrated geometrical feature which connects height estimate
with gait feature. The appearance features of the remaining human candidates are later tested
by SVMs to determine the object’s existence in new cameras. Experimental results show the
feasibility and effectiveness of this proposal and demonstrate the real-time in appearance
feature extraction and robustness to illumination and visual angle change.
The document discusses multispectral palm image fusion for biometric authentication using ant colony optimization. It introduces intra-modal fusion of palmprint images from multiple spectra to improve accuracy. The key steps involve detecting the region of interest, fusing the images using wavelet transforms, extracting Gabor features, selecting optimal features using ant colony optimization, and classifying with support vector machines. Experimental results and conclusions are also presented.
A Methodology for Extracting Standing Human Bodies from Single Imagesjournal ijrtem
Abstract: Extraction of the image of human body in unconstrained still images is challenging due to several factors, including shading, image noise, occlusions, background clutter, the high degree of human body deformability, and the unrestricted positions due to in and out of the image plane rotations. we propose a bottom-up approach for human body segmentation in static images. We decompose the problem into three sequential problems: Face detection, upper body extraction, and lower body extraction, since there is a direct pair wise correlation among them. Index Terms: Skin segmentation, Torso, Face recognition, Thresholding, Ethnicity, Morphology.
Face Recognition based on STWT and DTCWT using two dimensional Q-shift Filters IJERA Editor
The Biometrics is used to recognize a person effectively compared to traditional methods of identification. In this paper, we propose a Face recognition based on Single Tree Wavelet Transform (STWT) and Dual Tree Complex Wavelet transform (DTCWT). The Face Images are preprocessed to enhance quality of the image and resize. DTCWT and STWT are applied on face images to extract features. The Euclidian distance is used to compare features of database image with test face images to compute performance parameters. The performance of STWT is compared with DTCWT. It is observed that the DTCWT gives better results compared to STWT technique.
IJERA (International journal of Engineering Research and Applications) is International online, ... peer reviewed journal. For more detail or submit your article, please visit www.ijera.com
MultiModal Identification System in Monozygotic TwinsCSCJournals
This document presents a multimodal biometric system for identifying identical twins using face, fingerprint, and iris recognition. It utilizes Fisher's linear discriminant analysis to extract features from faces, principal component analysis for fingerprints, and local binary pattern features for iris matching. These features are then fused for identification. The system is tested on a database of 50 pairs of identical twins and shows promising results compared to other techniques. Receiver operating characteristics also indicate the proposed method performs better than other studied techniques in distinguishing identical twins.
The International Journal of Engineering and Science (The IJES)theijes
The International Journal of Engineering & Science is aimed at providing a platform for researchers, engineers, scientists, or educators to publish their original research results, to exchange new ideas, to disseminate information in innovative designs, engineering experiences and technological skills. It is also the Journal's objective to promote engineering and technology education. All papers submitted to the Journal will be blind peer-reviewed. Only original articles will be published.
The papers for publication in The International Journal of Engineering& Science are selected through rigorous peer reviews to ensure originality, timeliness, relevance, and readability
Artículo presentado por la Universidad de Vigo durante la jornada HOIP'10 organizada por la Unidad de Sistemas de información e interacción de TECNALIA.
Más información en http://www.tecnalia.com/es/ict-european-software-institute/index.htm
1) The document discusses automatic human face counting in digital color images. Color segmentation is used as the initial step, followed by clustering face regions and median filtering to eliminate small clusters.
2) The resulting blobs are matched against an ellipse face pattern while applying constraints to reject non-face blobs.
3) The system was implemented and validated on images with different formats, sizes, numbers of people, and background complexities.
photo detection in personal photo collectionsonalijagtap15
This document discusses methods for person recognition in personal photo collections. It describes four main face detection methods: knowledge-based, feature-based, appearance-based, and template matching. Appearance-based methods rely on machine learning algorithms trained on large datasets of face images. The document also discusses challenges with person recognition from diverse poses and attributes. It introduces the PIPA dataset for evaluating person recognition algorithms and describes using multiple cues like face, body, attributes, and scene context to improve recognition.
Multi Local Feature Selection Using Genetic Algorithm For Face IdentificationCSCJournals
This document presents a face recognition algorithm that uses a multi-local feature selection approach based on genetic algorithms and pseudo Zernike moment invariants. The algorithm involves five stages: 1) face detection using an ellipse to approximate the face region, 2) extraction of facial features (eyes, nose, mouth) within regions using genetic algorithms to locate templates with maximum edge density, 3) generation of moment invariants from the facial features using pseudo Zernike polynomials, 4) classification of facial features using radial basis function neural networks, and 5) selection of multiple local features for face identification. The algorithm was tested on over 3000 images from three databases, achieving recognition rates over 89% which is higher than global or single local feature approaches and
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.
IJERA (International journal of Engineering Research and Applications) is International online, ... peer reviewed journal. For more detail or submit your article, please visit www.ijera.com
This document summarizes a survey and analysis of various host-to-host congestion control proposals for TCP data transmission. It discusses the basic principles that underlie current host-to-host algorithms, including probing available network resources, estimating congestion through packet loss or delay, and quickly detecting packet losses. The document then analyzes specific algorithms like slow start, congestion avoidance, and fast recovery. It also examines calculating retransmission timeout and round-trip time, congestion avoidance and packet recovery techniques, and data transmission in TCP. The overall goal of these proposals is to control congestion in a distributed manner without relying on explicit network notifications.
IJERA (International journal of Engineering Research and Applications) is International online, ... peer reviewed journal. For more detail or submit your article, please visit www.ijera.com
IJERA (International journal of Engineering Research and Applications) is International online, ... peer reviewed journal. For more detail or submit your article, please visit www.ijera.com
IJERA (International journal of Engineering Research and Applications) is International online, ... peer reviewed journal. For more detail or submit your article, please visit www.ijera.com
This document summarizes a research paper that proposes a Modified Ultra Smart Counter Based Broadcast (MUSCBA) algorithm for mobile ad hoc networks (MANETs). The algorithm dynamically adjusts the counter-based broadcast threshold based on locally available neighborhood information, without requiring location devices. It improves on previous fixed counter-based approaches. Simulation results show the minimum and maximum number of neighbors observed for nodes in different network configurations and densities. The algorithm aims to reduce redundant broadcasts while maintaining high propagation, using only neighborhood data from periodic "Hello" packet exchanges.
This document summarizes a research paper that presents an area efficient 128-point Fast Fourier Transform (FFT) processor using a mixed radix 4-2 computation for OFDM applications. The processor uses a cached-memory architecture with two single-port SRAMs to increase FFT execution time. It was designed using a 0.35um CMOS process and integrates 552,000 transistors within an area of 2.8x2.8mm2. The processor can perform a 128-point, 36-bit FFT in 23.2us for 1D and 23.8ms for 2D FFT at 133MHz clock speed.
This document presents a new mesh analytical method for symbolic simulation of linear RLC circuits in the frequency domain. The method formulates the circuit equations as a set of mesh equations that account for both the steady-state impedances of the circuit elements as well as the initialization effects of non-zero initial conditions in the capacitors and inductors. It derives the mesh equations for a simple three-node, three-mesh example circuit and shows that the circuit equations can be expressed in matrix form with impedance matrices representing the steady-state and initialization impedances. The method aims to provide a simple yet robust way to simulate transient responses of linear circuits symbolically in the frequency domain.
This document summarizes different types of virtualization technologies including hardware virtualization, presentation virtualization, application virtualization, and others. It discusses how each type abstracts and decouples different computing resources to provide benefits like increased flexibility, simplified administration, improved performance and fault tolerance. Specific Microsoft virtualization technologies are also highlighted for each category. Use cases for virtualization like server consolidation, development/testing environments, and improving quality of service are outlined.
This document summarizes the performance measurements taken on a 40-channel 10G DWDM system operated by Bharath Sanchar Nigam Limited in India. The key findings were:
1. Optical power levels were measured for each channel at different points in the system. Some channels showed power levels indicating a fault in the respective fiber cable.
2. Parameters like noise figure, gain, polarization mode dispersion and dispersion were analyzed to identify limitations and suggest improvements to optimize the system.
3. Nonlinear impairments like four-wave mixing, self-phase modulation and stimulated Raman/Brillouin scattering were also evaluated to determine their impact based on number of channels, channel power, and fiber properties
IJERA (International journal of Engineering Research and Applications) is International online, ... peer reviewed journal. For more detail or submit your article, please visit www.ijera.com
This document summarizes an academic paper that proposes optimizing elliptic curve cryptography (ECC) through application-specific instruction set processor (ASIP) design. It applies pipelining techniques to the data path and uses complex instructions to reduce latency and the number of instructions needed for point multiplication. The paper describes applying different levels of pipelining to explore performance and find an optimal pipeline depth. It also develops a new combined algorithm to perform point doubling and addition using the specialized instructions. An FPGA implementation over GF(2163) is presented and shown to outperform previous work.
This study examines the use of ethanol-kerosene fuel blends in kerosene wick stoves as a more sustainable alternative fuel. Various blends of 5%, 10%, 15%, and 20% ethanol in kerosene by volume were tested in an unmodified stove and compared to pure kerosene. The thermal efficiency and fuel consumption rate were evaluated using a standard water boiling test. The maximum thermal efficiency was obtained with a 5% ethanol blend, while the minimum was with pure kerosene. The 10% ethanol blend had the highest fuel consumption rate. Overall, the performance of the blended fuels was found to be comparable to kerosene, indicating their potential as a renewable and less polluting alternative for cooking
This document summarizes key aspects of recommender systems and ranking techniques. It discusses how recommender systems typically focus on accuracy but overlook diversity. The paper explores various recommendation techniques, including content-based, collaborative filtering, knowledge-based, and hybrid approaches. It also examines different ranking methods that can increase aggregate diversity, such as popularity-based, reverse predicted rating, and parameterized ranking. The goal is to improve recommendation diversity while maintaining adequate accuracy.
IJERA (International journal of Engineering Research and Applications) is International online, ... peer reviewed journal. For more detail or submit your article, please visit www.ijera.com
MultiModal Identification System in Monozygotic TwinsCSCJournals
This document presents a multimodal biometric system for identifying identical twins using face, fingerprint, and iris recognition. It utilizes Fisher's linear discriminant analysis to extract features from faces, principal component analysis for fingerprints, and local binary pattern features for iris matching. These features are then fused for identification. The system is tested on a database of 50 pairs of identical twins and shows promising results compared to other techniques. Receiver operating characteristics also indicate the proposed method performs better than other studied techniques in distinguishing identical twins.
The International Journal of Engineering and Science (The IJES)theijes
The International Journal of Engineering & Science is aimed at providing a platform for researchers, engineers, scientists, or educators to publish their original research results, to exchange new ideas, to disseminate information in innovative designs, engineering experiences and technological skills. It is also the Journal's objective to promote engineering and technology education. All papers submitted to the Journal will be blind peer-reviewed. Only original articles will be published.
The papers for publication in The International Journal of Engineering& Science are selected through rigorous peer reviews to ensure originality, timeliness, relevance, and readability
Artículo presentado por la Universidad de Vigo durante la jornada HOIP'10 organizada por la Unidad de Sistemas de información e interacción de TECNALIA.
Más información en http://www.tecnalia.com/es/ict-european-software-institute/index.htm
1) The document discusses automatic human face counting in digital color images. Color segmentation is used as the initial step, followed by clustering face regions and median filtering to eliminate small clusters.
2) The resulting blobs are matched against an ellipse face pattern while applying constraints to reject non-face blobs.
3) The system was implemented and validated on images with different formats, sizes, numbers of people, and background complexities.
photo detection in personal photo collectionsonalijagtap15
This document discusses methods for person recognition in personal photo collections. It describes four main face detection methods: knowledge-based, feature-based, appearance-based, and template matching. Appearance-based methods rely on machine learning algorithms trained on large datasets of face images. The document also discusses challenges with person recognition from diverse poses and attributes. It introduces the PIPA dataset for evaluating person recognition algorithms and describes using multiple cues like face, body, attributes, and scene context to improve recognition.
Multi Local Feature Selection Using Genetic Algorithm For Face IdentificationCSCJournals
This document presents a face recognition algorithm that uses a multi-local feature selection approach based on genetic algorithms and pseudo Zernike moment invariants. The algorithm involves five stages: 1) face detection using an ellipse to approximate the face region, 2) extraction of facial features (eyes, nose, mouth) within regions using genetic algorithms to locate templates with maximum edge density, 3) generation of moment invariants from the facial features using pseudo Zernike polynomials, 4) classification of facial features using radial basis function neural networks, and 5) selection of multiple local features for face identification. The algorithm was tested on over 3000 images from three databases, achieving recognition rates over 89% which is higher than global or single local feature approaches and
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.
IJERA (International journal of Engineering Research and Applications) is International online, ... peer reviewed journal. For more detail or submit your article, please visit www.ijera.com
This document summarizes a survey and analysis of various host-to-host congestion control proposals for TCP data transmission. It discusses the basic principles that underlie current host-to-host algorithms, including probing available network resources, estimating congestion through packet loss or delay, and quickly detecting packet losses. The document then analyzes specific algorithms like slow start, congestion avoidance, and fast recovery. It also examines calculating retransmission timeout and round-trip time, congestion avoidance and packet recovery techniques, and data transmission in TCP. The overall goal of these proposals is to control congestion in a distributed manner without relying on explicit network notifications.
IJERA (International journal of Engineering Research and Applications) is International online, ... peer reviewed journal. For more detail or submit your article, please visit www.ijera.com
IJERA (International journal of Engineering Research and Applications) is International online, ... peer reviewed journal. For more detail or submit your article, please visit www.ijera.com
IJERA (International journal of Engineering Research and Applications) is International online, ... peer reviewed journal. For more detail or submit your article, please visit www.ijera.com
This document summarizes a research paper that proposes a Modified Ultra Smart Counter Based Broadcast (MUSCBA) algorithm for mobile ad hoc networks (MANETs). The algorithm dynamically adjusts the counter-based broadcast threshold based on locally available neighborhood information, without requiring location devices. It improves on previous fixed counter-based approaches. Simulation results show the minimum and maximum number of neighbors observed for nodes in different network configurations and densities. The algorithm aims to reduce redundant broadcasts while maintaining high propagation, using only neighborhood data from periodic "Hello" packet exchanges.
This document summarizes a research paper that presents an area efficient 128-point Fast Fourier Transform (FFT) processor using a mixed radix 4-2 computation for OFDM applications. The processor uses a cached-memory architecture with two single-port SRAMs to increase FFT execution time. It was designed using a 0.35um CMOS process and integrates 552,000 transistors within an area of 2.8x2.8mm2. The processor can perform a 128-point, 36-bit FFT in 23.2us for 1D and 23.8ms for 2D FFT at 133MHz clock speed.
This document presents a new mesh analytical method for symbolic simulation of linear RLC circuits in the frequency domain. The method formulates the circuit equations as a set of mesh equations that account for both the steady-state impedances of the circuit elements as well as the initialization effects of non-zero initial conditions in the capacitors and inductors. It derives the mesh equations for a simple three-node, three-mesh example circuit and shows that the circuit equations can be expressed in matrix form with impedance matrices representing the steady-state and initialization impedances. The method aims to provide a simple yet robust way to simulate transient responses of linear circuits symbolically in the frequency domain.
This document summarizes different types of virtualization technologies including hardware virtualization, presentation virtualization, application virtualization, and others. It discusses how each type abstracts and decouples different computing resources to provide benefits like increased flexibility, simplified administration, improved performance and fault tolerance. Specific Microsoft virtualization technologies are also highlighted for each category. Use cases for virtualization like server consolidation, development/testing environments, and improving quality of service are outlined.
This document summarizes the performance measurements taken on a 40-channel 10G DWDM system operated by Bharath Sanchar Nigam Limited in India. The key findings were:
1. Optical power levels were measured for each channel at different points in the system. Some channels showed power levels indicating a fault in the respective fiber cable.
2. Parameters like noise figure, gain, polarization mode dispersion and dispersion were analyzed to identify limitations and suggest improvements to optimize the system.
3. Nonlinear impairments like four-wave mixing, self-phase modulation and stimulated Raman/Brillouin scattering were also evaluated to determine their impact based on number of channels, channel power, and fiber properties
IJERA (International journal of Engineering Research and Applications) is International online, ... peer reviewed journal. For more detail or submit your article, please visit www.ijera.com
This document summarizes an academic paper that proposes optimizing elliptic curve cryptography (ECC) through application-specific instruction set processor (ASIP) design. It applies pipelining techniques to the data path and uses complex instructions to reduce latency and the number of instructions needed for point multiplication. The paper describes applying different levels of pipelining to explore performance and find an optimal pipeline depth. It also develops a new combined algorithm to perform point doubling and addition using the specialized instructions. An FPGA implementation over GF(2163) is presented and shown to outperform previous work.
This study examines the use of ethanol-kerosene fuel blends in kerosene wick stoves as a more sustainable alternative fuel. Various blends of 5%, 10%, 15%, and 20% ethanol in kerosene by volume were tested in an unmodified stove and compared to pure kerosene. The thermal efficiency and fuel consumption rate were evaluated using a standard water boiling test. The maximum thermal efficiency was obtained with a 5% ethanol blend, while the minimum was with pure kerosene. The 10% ethanol blend had the highest fuel consumption rate. Overall, the performance of the blended fuels was found to be comparable to kerosene, indicating their potential as a renewable and less polluting alternative for cooking
This document summarizes key aspects of recommender systems and ranking techniques. It discusses how recommender systems typically focus on accuracy but overlook diversity. The paper explores various recommendation techniques, including content-based, collaborative filtering, knowledge-based, and hybrid approaches. It also examines different ranking methods that can increase aggregate diversity, such as popularity-based, reverse predicted rating, and parameterized ranking. The goal is to improve recommendation diversity while maintaining adequate accuracy.
IJERA (International journal of Engineering Research and Applications) is International online, ... peer reviewed journal. For more detail or submit your article, please visit www.ijera.com
The document discusses the potential use of a mix of fly ash and diluted distillery effluent in agriculture. Fly ash contains nutrients that can benefit plant growth, while distillery effluent is rich in nitrogen, phosphorus, and potassium. A study was conducted on the ornamental plant Calendula officinalis grown in soil amended with these materials. Results showed increased vegetative growth, photosynthetic pigments, and earlier flowering in the amended soil. The mix of fly ash and effluent has potential as a biofertilizer to replace chemical fertilizers and boost agricultural productivity in a sustainable way.
Este documento presenta varios artículos hechos a mano como flores de papel, cajas de cerillas decoradas, cojines pintados, manteles individuales, cuadros personalizados, marcos de fotos decorados y cajitas únicas para guardar joyas. Se proporcionan detalles como colores, precios y contacto para pedidos. El objetivo es mostrar las creaciones artesanales disponibles y facilitar el proceso de pedido.
Este documento describe los esfuerzos del Instituto Nacional de Formación Docente para conectar a los docentes con las tecnologías de la información y la comunicación (TIC) a través del programa "Conectar Igualdad". El objetivo es construir un espacio de encuentro para compartir experiencias innovadoras, capacitar a los docentes en el uso pedagógico de las TIC, y asegurar que los docentes tengan acceso a las herramientas digitales.
This document proposes using gait as a biometric for human identification. It discusses extracting image features from a person's silhouette and gait cycle, including the width of their outer contour and their entire binary silhouette. Two approaches are explored: the indirect approach transforms these high-dimensional image features into lower dimensions before using an HMM for representation, while the direct approach uses the features directly with an HMM. The document outlines issues with existing identification systems and argues that gait recognition could provide a more unique, reliable means of identification than alternatives like signatures or facial recognition.
Gait Recognition for Person Identification using Statistics of SURFijtsrd
This document presents a method for human identification using gait recognition. It extracts statistical gait features from Speeded-Up Robust Features (SURF) descriptors applied to binary, roughness, and gray-scale images of a person's gait cycle. Specifically, it calculates 12 statistical features including mean, root mean square, skewness, and kurtosis from the SURF descriptors of the three image types. A support vector machine classifier is used to evaluate the discriminatory ability of the extracted statistical gait features for human identification. The method is tested on the CASIA-B gait dataset and aims to improve recognition accuracy under variations in clothing, carrying conditions, and viewing angles.
Human gait recognition using preprocessing and classification techniques IJECEIAES
Biometric recognition systems have been attracted numerous researchers since they attempt to overcome the problems and factors weakening these systems including problems of obtaining images indeed not appearing the resolution or the object completely. In this work, the object movement reliance was considered to distinguish the human through his/her gait. Some losing features probably weaken the system’s capability in recognizing the people, hence, we propose using all data recorded by the Kinect sensor with no employing the feature extraction methods based on the literature. In these studies, coordinates of 20 points are recorded for each person in various genders and ages, walking with various directions and speeds, creating 8404 constraints. Moreover, pre-processing methods are utilized to measure its influences on the system efficiency through testing on six types of classifiers. Within the proposed approach, a noteworthy recognition rate was obtained reaching 91% without examining the descriptors.
Symbolic representation and recognition of gait an approach based on lbp of ...sipij
Gait is one of the biometric techniques used to identify an individual from a distance by his/her walking
style. Gait can be recognized by studying the static and dynamic part variations of individual body contour
during walk. In this paper, an interval value based representation and recognition of gait using local
binary pattern (LBP) of split gait energy images is proposed. The gait energy image (GEI) of a subject is
split into four equal regions. LBP technique is applied to each region to extract features and the extracted
features are well organized. The proposed representation technique is capable of capturing variations in
gait due to change in cloth, carrying a bag and different instances of normal walking conditions more
effectively. Experiments are conducted on the standard and considerably large database (CASIA database
B) and newly created University of Mysore (UOM) gait dataset to study the efficacy of the proposed gait
recognition system. The proposed system being robust to handle variations has shown significant
improvement in recognition rate.
An Image Mining System for Gender Classification & Age Prediction Based on Fa...IOSR Journals
This document presents a study on developing an image mining system for gender classification and age prediction based on facial features. It first discusses previous related work on gender recognition and age estimation from facial images using various machine learning techniques. It then describes the proposed algorithm which includes steps for gender recognition, age prediction, face detection, and facial feature extraction. For gender recognition, it extracts facial features, counts male and female features, applies fuzzy rules to classify gender. For age prediction, it trains on facial images with known ages and tests by matching input faces to training faces using histogram matching. The paper presents data flow diagrams and describes the steps involved in the proposed algorithm.
BIOMETRIC AUTHORIZATION SYSTEM USING GAIT BIOMETRYIJCSEA Journal
ABSTRACT
Human gait, which is a new biometric aimed to recognize individuals by the way they walk have come to play an increasingly important role in visual surveillance applications. In this paper a novel hybrid holistic approach is proposed to show how behavioural walking characteristics can be used to recognize unauthorized and suspicious persons when they enter a surveillance area. Initially background is modelled from the input video captured from cameras deployed for security and the foreground moving object in the individual frames are segmented using the background subtraction algorithm. Then gait representing spatial, temporal and wavelet components are extracted and fused for training and testing multi class support vector machine models (SVM). The proposed system is evaluated using side view videos of NLPR database. The experimental results demonstrate that the proposed system achieves a pleasing recognition rate and also the results indicate that the classification ability of SVM with Radial Basis Function (RBF) is better than with other kernel functions.
To identify the person using gait knn based approacheSAT Journals
Abstract In human identification, the process of identifying the person by their gait is an emerging research trend in the field of visual surveillance. Gait is a new biometrics, has been recently used to recognize a person via style of his walking. While person walking variation becomes take place in different parts of body. On the basis of these variations, proposed method is evaluated on CASIA gait database by using K-nearest Neighbor classifier. Experimental results demonstrate that the proposed method has an encouraging recognition performance also the results indicate that the classification ability of KNN with correlation measure perform better than with other type of distance measure functions. Keywords: Gait biometrics, KNN, visual surveillance, CASIA, silhouette.
This document presents a human identification system using gait recognition. The system first detects moving subjects in video sequences and extracts silhouettes using background subtraction. It then calculates motion parameters like joint angles and gait velocity from the silhouettes using Speeded Up Robust Features (SURF) descriptors. These motion parameters are classified using a Meta-sample based sparse representation method, achieving an overall classification rate of 94.6782% on a test dataset. The system provides view-invariant human identification through gait analysis.
This document presents a human identification system using gait biometrics. The system first detects moving subjects in video sequences and extracts silhouettes using background subtraction. It then uses Speeded Up Robust Features (SURF) to detect keypoints and extract descriptors, which are used to calculate gait features like joint angles and velocities. These gait features are classified using Meta-sample based Sparse Representation Classification (MSRC) which achieves an overall classification rate of 94.6782% on a self-collected dataset.
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.
ROBUST HUMAN TRACKING METHOD BASED ON APPEARANCE AND GEOMETRICAL FEATURES IN ...cscpconf
This paper proposes a robust tracking method which concatenates appearance and geometrical
features to re-identify human in non-overlapping views. A uniformly-partitioning method is
proposed to extract local HSV(Hue, Saturation, Value) color features in upper and lower
portion of clothing. Then adaptive principal view selecting algorithm is presented to locate
principal view which contains maximum appearance feature dimensions captured from different
visual angles. For each appearance feature dimension in principal view, all its inner frames get
involved in training a support vector machine (SVM). In matching process, human candidate
filtering is first operated with an integrated geometrical feature which connects height estimate
with gait feature. The appearance features of the remaining human candidates are later tested
by SVMs to determine the object’s existence in new cameras. Experimental results show the
feasibility and effectiveness of this proposal and demonstrate the real-time in appearance
feature extraction and robustness to illumination and visual angle change.
Optimized Biometric System Based on Combination of Face Images and Log Transf...sipij
The biometrics are used to identify a person effectively. In this paper, we propose optimised Face
recognition system based on log transformation and combination of face image features vectors. The face
images are preprocessed using Gaussian filter to enhance the quality of an image. The log transformation
is applied on enhanced image to generate features. The feature vectors of many images of a single person
image are converted into single vector using average arithmetic addition. The Euclidian distance(ED) is
used to compare test image feature vector with database feature vectors to identify a person. It is
experimented that, the performance of proposed algorithm is better compared to existing algorithms.
This document describes a proposed method to improve image classification accuracy and speed using the bag-of-features model with spatial pooling. The proposed method has two phases: a training phase to create an image feature database, and an evaluation phase to classify new images. In the evaluation phase, spatial pooling is applied to input image features before classification with KNN. Variance-based feature selection is also used to reduce features before KNN classification. Experimental results show the proposed method improves classification accuracy up to 5% and reduces classification time by up to 50% compared to the standard bag-of-features model.
Human Face Detection and Tracking for Age Rank, Weight and Gender Estimation ...IRJET Journal
This document proposes a system for real-time human face detection, tracking, and estimation of age, weight, and gender from face images using a Raspberry Pi processor. The system uses OpenCV for face detection and extracts facial features to classify age, estimate weight, and determine gender through a probabilistic framework. The system allows for real-time detection of multiple faces with high efficiency even from low-quality images. Evaluation shows the low-cost Raspberry Pi provides fast execution speeds suitable for real-time applications.
The Extraction of Spatial Features from remotely sensed data and the use of this information as input into further decision making systems such as geographical information systems (GIS) has received considerable attention over the few decades. The successful use of GIS as a decision support tool can only be achieved, if it becomes possible to attach a quality label to the output of each spatial analysis operation. Thus the accuracy of Spatial Feature Extraction gained more attention as geographic features can hardly formulated in a certain pattern due to intra-class variation and inter-class similarity. Besides these Spatial Feature Extraction further include positional uncertainty, attribute uncertainty, topological uncertainty, inaccuracy, imprecision/inexactitude, inconsistency, incompleteness, repetition, vagueness, noisy, omittance, misinterpretation, misclassification, abnormalities and knowledge uncertainty. To control and reduce uncertainty in an acceptable degree, a Probabilistic shape model is described for Extracting Spatial Features from multi-spectral image. The advantages of this, as opposed to the conventional approaches, are greater accuracy and efficiency, and the results are in a more desirable form for most purposes.
Human Re-identification with Global and Local Siamese Convolution Neural NetworkTELKOMNIKA JOURNAL
This document proposes a global and local structure of Siamese Convolution Neural Network (SCNN) to perform human re-identification in single-shot approaches. The network extracts features from global and local parts of input images. A decision fusion technique then combines the global and local features. Experimental results on the VIPeR dataset show the proposed method achieves a normalized Area Under Curve score of 95.75% without occlusion, outperforming using local or global features alone. With occlusion, the score is 77.5%, still better than alternatives. The method performs well for re-identification including in occlusion cases by leveraging both global and local information.
Gait is the style of walking or limb movement of a
person. Gait recognition is a biometric technology that is based
on behavioral features of human. It finds applications in different
areas such as banks, military, airports, and many other areas for
threat detection and security purposes. Biometric gait recognition
is a popular area of research as it is an unobtrusive process to
recognize a person. In the current paper we review several
approaches of gait recognition, discuss their advantages and
disadvantages and then show directions for future research.
ROBUST STATISTICAL APPROACH FOR EXTRACTION OF MOVING HUMAN SILHOUETTES FROM V...ijitjournal
Human pose estimation is one of the key problems in computer visionthat has been studied in the recent
years. The significance of human pose estimation is in the higher level tasks of understanding human
actions applications such as recognition of anomalous actions present in videos and many other related
applications. The human poses can be estimated by extracting silhouettes of humans as silhouettes are
robust to variations and it gives the shape information of the human body. Some common challenges
include illumination changes, variation in environments, and variation in human appearances. Thus there
is a need for a robust method for human pose estimation. This paper presents a study and analysis of
approaches existing for silhouette extraction and proposes a robust technique for extracting human
silhouettes in video sequences. Gaussian Mixture Model (GMM) A statistical approach is combined with
HSV (Hue, Saturation and Value) color space model for a robust background model that is used for
background subtraction to produce foreground blobs, called human silhouettes. Morphological operations
are then performed on foreground blobs from background subtraction. The silhouettes obtained from this
work can be used in further tasks associated with human action interpretation and activity processes like
human action classification, human pose estimation and action recognition or action interpretation.
1. Sudheer Reddy Bandi, A Gopi Suresh , Immanuel Alphonse / International Journal of
Engineering Research and Applications (IJERA) ISSN: 2248-9622 www.ijera.com
Vol. 2, Issue 5, September- October 2012, pp.822-827
Gait Based Gender Classification Using Silhouette Image Gait
Database
Sudheer Reddy Bandi1, A Gopi Suresh2 , Immanuel Alphonse3
1,2,3
Assistant Professor, Dept of Information Technology, Tagore Engineering College, Chennai
Abstract
A realistic appearance-based Golomb et al. trained a fully connected two-layer
representation to discriminate gender from side neural network, SEXNET, to identify gender from
view gait sequence is introduced here. This gait face images. Brunelli and Poggio [2] developed
representation is based on simple features such as HyperBF networks for gender classification in which
moments extracted from orthogonal view video two competing networks, one for male and the other
silhouettes of human walking motion. The for female, are trained using 16 geometric features.
silhouette image is divided into two regions. The To sum up, some of these techniques are appearance-
first region includes from the head to the torso based methods and others are based on geometric
region whereas the second region is from torso to features. In Moghaddam and Yang’s paper [3],
the feet region and for each region features are nonlinear SVM was investigated in gender
extracted. Finally we employ different pattern classification for low-resolution thumbnail face (21-
classifiers like KNN (K- Nearest Neighbor) and by-12 pixels) on 1,755 images from the FERET
SVM (Support Vector Machine) to classify the database.
gender. The division of two regions is based on the Proceeding toward gender classification in
centroid of the silhouette image. The experimental emotional speech; in [4] Harb et al. proposed a
results show that SVM classifier gives better method in which a set of acoustic and pitch features
results when compared to other classifiers. The are used for gender identification. Concerning
classification results are expected to be more previous work on non-emotional speech, the system
reliable than those reported in previous papers. proposed by Zeng et al. [5] is based on Gaussian
The proposed system is evaluated using side view mixture models (GMMs) of pitch and spectral
videos of CASIA dataset B. perceptual linear predictive coefficients.
Nevertheless, when compared to face or
Keywords: Appearance based features, binary voice, gait can be perceived at a distance. In addition
moments, ellipse features, gait analysis, gender to these, gait has various advantages like non-contact,
classification, and human silhouette. non-invasive and in general, does not require
subject’s willingness. This particular issue has stirred
1. INTRODUCTION up the interest of the computer vision community in
Recently, biometrics has increasingly creating gait based gender recognition systems. A
attracted the attention as a key technology for number of applications can be benefitted from the
realizing a more secure and safer society. Although development of such systems. For example
most of the studies on biometrics focus on person demographic analysis of systems, access control,
authentication, namely hard bio metrics, it is also biometric systems that use gender recognition to
important to promote the recognition of properties reduce the search space to half, etc.
such as gender, age and ethnicity, namely soft Most of gait research has been addressed to
biometrics. The perception of gender recognition biometric identification, which consists of predicting
determines social interactions. Humans are very the identity of a person according to his/her way of
accurate at identifying gender from a face, voice and walk. However, few recent works have used gait
through gait. Among biometric modalities, gait has analysis for other classification tasks, such as gender
several promising properties such as availability at a recognition or age estimation. Regardless of the
distance from a camera even without the cooperation purposes two different approaches to describe gait
of the subject; hence gait based hard biometrics can be considered. Some works extract dynamic
[13][14][15] has been extensively studied with the features from subject’s movements [12, 10], while
aim of realizing wide area surveillance and assistance others take static attributes related with the
with criminal investigation. Furthermore, gait based appearance of the subject [6, 7, 9]; what implicitly
soft biometrics are also an active research area (e.g., might contain motion information.
gender classification [7][9][10], age group There have been a number of appearance
classification [16][17]). based methods to classify gender from gait. Lee and
There has been much work to classify Grimson analyzed the motion of 7 different regions of
gender from human faces. In early 1990s, various a silhouette [9]. Features for discrimination of gender
neural network techniques were employed for gender were selected using Analysis of Variation (ANOVA),
classification from a frontal face. In paper [1] and an SVM was trained to categorize gender.
822 | P a g e
2. Sudheer Reddy Bandi, A Gopi Suresh , Immanuel Alphonse / International Journal of
Engineering Research and Applications (IJERA) ISSN: 2248-9622 www.ijera.com
Vol. 2, Issue 5, September- October 2012, pp.822-827
Similar to [9] Huang and Wang [7] also use ellipse 2. METHODOLOGY OF PROPOSED WORK
features for gender classification. The difference is The approach proposed here for human gait
that Huang and Wang combine multiview features to characterization is based on the method introduced by
improve the performance. Unlike Lee’s and Huang’s Lee and Grimson [9], where the silhouette appearance
methods, in [12], the gait signature was represented of a side-view human walking sequence was
as a sequential set of 2-D stick figures. Each 2-D described by static features that were further used for
stick figure is extracted from a human silhouette and gender recognition. The architectural block diagram
contains head, neck, shoulder, waist, pelvis, knees, for proposed methodology is given in Fig 2. The
and ankles. Davis et al. presented a three-mode work proposed here can be defined in terms of the
expressive-feature model for recognizing gender from following steps.
point-light displays of walking people [10]. They
developed a trimodal nature of walkers (posture,
time, and gender) in which an efficient three-mode
PCA representation was employed.
The previously introduced methods are
based on human dynamic movement or silhouettes,
and the background and clothing texture are all
removed. Different from these methods, Cao et al.
proposed a new method which recognizes gender
from static full body images [8]. They use the
Histogram of Oriented Gradients (HOG) as the
feature, and Adaboost and Random Forest algorithms
as classifiers. Fig 2. Architectural block diagram of proposed
Head and hair, back, chest and thigh are system.
more discriminative for gender classification than
other components. Body sway, waist-hip ratio, and 2.1 Foreground segmentation
shoulder-hip ratio are also indicative of a walker’s For each frame, the foreground (silhouette)
gender. Males tend to swing their shoulders more is segmented from the background. A new binary
than their hips, and females tend to swing their hips frame with the silhouette highlighted is obtained. The
more than their shoulders. Males normally have wider algorithm used for foreground segmentation is frame
shoulders than females, and females normally have difference approach.
thin waists and wider hips. Hair style and chest are
two important body components for gender 2.2 Silhouette Extraction
classification. However, the hair component is In order to be robust to changes of clothing
and illumination it is reasonable to consider the
binary silhouette of the object, which is nothing but
outline of the moving object with featureless interior.
A reduced image defined by the bounding box that
encloses the silhouette is extracted to normalize its
size and location.
divided into head component and trunk component. 2.3 Regionalization
Fig 1. Process of Feature Extraction The bounding box is divided into two
regions obtained from the silhouette centroid and
Chest sometimes can be divided into the arm from fixed percentages of the box sizes.
component (for fat persons), and sometimes into the
trunk component (for slim persons). To study the 2.4 Feature Extraction
effectiveness of different components, it is better to For each silhouette of a gait video sequence,
set the segmentation borders in the areas where both we find the centroid and proportionally divide the
males and females are the same, such as the centre of silhouette into two regions as shown in Fig .1. The
the trunk. The upper body silhouettes convey more frontal-parallel view of the silhouette is divided into
static information than the lower body silhouettes. the front and back sections by a vertical line at the
silhouette centroid. The parts above and below the
The rest of the paper is organized as follows. centroid are equally divided in the horizontal
Methodology of the proposed work is given in direction, resulting in two regions that roughly
Section 2. Performance analysis and results are correspond to: r1, head to torso region; r2, torso to
explained in Section 3. Section 4 concludes along feet region.
with some future works. For each and every region ellipse is formed
and its parameters are considered. The fitting of an
ellipse region involves computing the mean and
823 | P a g e
3. Sudheer Reddy Bandi, A Gopi Suresh , Immanuel Alphonse / International Journal of
Engineering Research and Applications (IJERA) ISSN: 2248-9622 www.ijera.com
Vol. 2, Issue 5, September- October 2012, pp.822-827
covariance matrix for the foreground pixels in the The orientation is defined only modulo π so
region. Since images contain a great deal of it is chosen to lie in a range of π appropriate for each
information, issues of representation become region of the silhouette. The ellipse parameters
important. We show that the geometric properties of extracted from each region of the silhouette are the
interest can be computed from projections of binary centroid, the aspect ratio(l) of the major axis to minor
images. Binary images are used in many applications axis, the angle of orientation (α) of major axis to x
since they are the simplest to process, but they are axis which forms the region feature vector f(r i) as in
such an impoverished representation of the image (“6,”)
information that their use is not always possible.
However, they are useful where all the information f(ri)=(xi,yi,li,αi) (6)
we need can be provided by the silhouette of the The features extracted from each frame of a
object and when we can obtain the silhouette of that walking sequence consists of features from each of
object easily. the two regions, so the frame feature vector Fj of the
The various binary image properties are jth frame is given as
given as 0th, 1st and 2nd order moments. The area is Fj=(f(r1),f(r2)) (7)
given by 0th moment of object. The centroid is given In addition to these 8 features, we use one
by 1st order moments as (M10, M01) and 2nd order additional feature, h, the height (relative to body
moments gives the orientation of the object with the length) of the centroid of the whole silhouette to
axis. Most of these notes are adapted from Horn's describe the proportions of the torso and legs.
book on Robot Vision [11].
Let f(x, y) be the binary image in which we
want to fit an ellipse. Assume that the foreground
pixels are 1 and the background pixels are 0 then the
mean x and y of the foreground pixels or the centroid
of the region is given by,
xmean=1/N*(∑f(x,y)*x)
(1)
ymean=1/N*(∑f(x,y)*y) Fig 4. Representation of 8 features of a silhouette
(2) frame.
where N is the total number of foreground pixels:
Given the region features across a gait
N=∑I(x, y) sequence, we need a concise representation across
(3) time. We compute two types of features across time:
The binary moments of image are given as follows the mean and standard deviation of region features
across time. Specifically, the gait average appearance
Mpq=∑∑xpyq f(x,y) feature vector of a sequence s is,
Area==∑∑f(x,y)
µ00=M00 s= (meanj(Fj), stdj(Fj)) (8)
µ01= µ10=0 Where j = 1. . . last frame and s is 8-
µ11= M11-xmean* M01= M11-ymean* dimensional. This feature set is very simple to
M01 compute and robust to noisy foreground silhouettes.
µ20=M20-xmean*M10 Intuitively, the mean features describe
µ02= M02-ymean*M01 the average–looking ellipses for each of two
µ!02= µ02/ µ00= M02/ M00-ymean2 regions of the body; taken together, the two ellipses
µ!20= µ20/ µ00= M20/ M00-xmean2 describe the average shape of the body. The standard
µ!11= µ11/ µ00= M11/ M00- deviation features roughly describe the changes in the
xmean*ymean shape of each region caused by the motion of the
body, where the amount of change is affected by
λi= (µ!20+ µ!02±sqrt(4µ112+( µ!20-
factors such as how much one swings one’s arms and
µ 02)2))/2
!
legs.
Major axis=4*sqrt(λmax)
However, the mean and standard deviation
Minor axis=4*sqrt(λmin) representation has the additional advantage that they
are not seriously affected by the occasional frame
skip that a foreground segmentation algorithm may
Fig 3. Binary moments of an image. undergo. Hence, the feature vector we compute for
From the binary moments the aspect ratio and angle each gait sequence consists of the mean and standard
of orientation are given as follows. deviations across time of each of the parameters for
Aspect ratio=major axis/minor axis (4) each region. The feature vector for one silhouette
Angle=0.5atan (2 µ!11/( µ!20- µ!02)) (5) image is given in Table I.
824 | P a g e
4. Sudheer Reddy Bandi, A Gopi Suresh , Immanuel Alphonse / International Journal of
Engineering Research and Applications (IJERA) ISSN: 2248-9622 www.ijera.com
Vol. 2, Issue 5, September- October 2012, pp.822-827
TABLE I compared to the sequences in the reference data base.
FEATURE VALUES FOR SINGLE SILHOUETTE IMAGE Train data and test data is formed from the CASIA
database B.
We used an implementation of support-
vector machine and experimented with the linear,
Multilayer Perceptron, Polynomial, Quadratic and
Radial basis function kernels. The SVM’s are trained
using the 8 features of a single video and under the
random-person vs. random-sequence conditions. The
results for these tests conditions are listed in Table II.
These features are also trained using KNN classifier.
Four normal forms of walking are used to test the
data. The average performance of all the data is
Where P=Parameters, R=Regions X= x coordinate taken and their results are tabulated in Table III.
of the centroid, Y= y coordinate of the centroid The precision rate of SVM and kNN models
are summarized in Table IV, from which it can be
3. EXPERIMENTAL RESULTS seen that the recognition performance based on SVM
To evaluate the performance of the proposed classifier is better than kNN. Precision rate of the
algorithm the experiments are conducted in the classifiers are calculated using “(9,)”.
CASIA database B and various classifiers are used to
compare the results which are explained in the
Precision= ×100
subsections given below.
(9)
TABLE III
CLASSIFICATION RATES OF DIFFERENT SVM KERNEL
TYPES
SVM kernel Type Recognition Rate
Linear 93.33%
Polynomial 81.67%
Fig 5. Male (left) and female (right) images from
the CASIA Database. Quadratic 80.00%
3.1 Data Acquisition Radial basis function 75.83%
In any pattern recognition system, the database
used for evaluation is necessary. We choose the
In this work apart from recognition rate,
CASIA Gait Database (Dataset B) for our
other measures such as True Positive rate (TPr), True
experiments as shown in Fig 5. There are 124
Negative rate (TNr), Geometric Mean and Area
subjects (93 males and 31 females) in the CASIA
Under Classification (AUC) which are more
Database (Dataset B), and most subjects are Asians.
appropriate for imbalanced problems are also
All the 31 females and 31 randomly selected males
considered and the values are tabulated in Table IV.
are used in the experiments. There are six video
The experiments for the proposed approach were
sequences for each subjects; altogether there are 372
conducted on a personal computer with an Intel Core
sequences. All selected data were captured from the
2 Duo processor (2.19 GHz) and 1 GB RAM
side view with normal clothes and without any bag as
configured with Microsoft Windows XP and Matlab
shown in Fig. 5.
7.5 software with image processing toolbox and bio
informatics toolbox. The comparison chart of all the
3.2 Classification Results and Analysis specified performance measures is given in Fig. 6.
For each side view videos of CASIA gait The formulae for the measures are given below.
database, we first generate silhouette images using
background subtraction algorithm and then silhouette
image features are extracted as explained in above TPr= ×100(10)
section. Then we trained two classification models
SVM and kNN with the 8 features and the genders
TNr= ×100 (11)
are classified by the trained model set. We report
results in terms of cumulative match scores. To
calculate these scores, we conduct multiple tests GM= (12)
using multiple test sequences. Each test sequence is
825 | P a g e
5. Sudheer Reddy Bandi, A Gopi Suresh , Immanuel Alphonse / International Journal of
Engineering Research and Applications (IJERA) ISSN: 2248-9622 www.ijera.com
Vol. 2, Issue 5, September- October 2012, pp.822-827
AUC=(TPr+TNr)/2 (13) Various pattern classifiers like KNN and SVM are
used for classification and the SVM gives better
TABLE IV recognition results. This view and appearance
RECOGNITION RATES OF DIFFERENT CLASSIFIERS dependent model of gait can be further extended to
accommodate a multiple appearance model of a
Classifier Type Recognition Rate or person and in conjunction with other recognition
Precision rate modalities.
SVM 93.33% REFERENCES
[1] B. A. Golomb, D. T. Lawrence and T. J.
KNN 83.33% Sejnowski. “Sexnet: A neural network
identifies sex from human faces,” In
International Conference on Advances in
TABLE V Neural Information Processing Systems,
COMPARISON OF DIFFERENT PERFORMANCE 1991.
MEASURES [2] R. Brunelli and T. Poggio, “Hyperbf
networks for gender classification,” In
Performance SVM KNN DRAPA Image Understanding Workshop,
Measures 1992.
TPr 96.43% 86.67% [3] B. Moghaddam and M. Yang. “Learning
Gender with Support Faces,” IEEE Trans.
TNr 90.00% 80.00% Pattern Analysis and Machine Intelligence,
24(5):707–711, May 2002.
[4] H. Harb and L. Chen, “Voice-based gender
GM 93.27% 83.27%
identification in multimedia applications,” J.
Intelligent Information Systems, 24(2):179–
AUC 93.34% 83.34% 198, 2005.
[5] Y. Zeng, Z. Wu, T. Falk, and W. Y. Chan
(2006). “Robust GMM based gender
classification using pitch and RASTA-PLP
parameters of speech,” In Proc. 5th. IEEE
Int. Conf Machine Learning and
Cybernetics, pages 3376–3379. China.
[6] X. Li, S. J. Maybank, S. Yan, D. Tao, and D.
Xu, “Gait components and their application
to gender recognition,” IEEE Trans. Syst.,
Man, Cybern. C, Appl. Rev., vol. 38, no. 2,
pp. 145–155, 2008.
[7] G. Huang and Y. Wang, “Gender
classification based on fusion of multi-view
gait sequences,” in Proc. 8th Asian Conf.
Computer Vision, 2007, pp. 462–471.
[8] L. Cao, M. Dikmen, Y. Fu, and T. S. Huang,
Performance Measures “Gender recognition from body,” in Proc.
16th ACM Int. Conf. Multimedia, 2008, pp.
Fig. 6 Comparison Charts of Performance 725–728.
Measures [9] L. Lee and W. E. L. Grimson, “Gait analysis
for recognition and classification,” in Proc.
4. CONCLUSION 5th IEEE Int. Conf. Automatic Face and
Gait-based gender classification is a new and Gesture Recognition, Washington, DC, May
2002, pp. 155–162.
interesting topic. This paper introduces a realistic
appearance based representation of gait sequences for
automatic gender recognition. An exhaustive study [10] J. W. Davis and H. Gao, “Gender
designed to evaluate the capacity of the CASIA Gait recognition from walking movements using
adaptive three-mode PCA,” in Proc. Conf.
Database for gender recognition tasks was carried
Computer Vision and Pattern Recognition
out. This dataset contains gait samples from 124
subjects, distributed in an unbalanced way with 31 Workshop, Washington, DC, 2004, vol. 1.
[11] Berthold Klaus Paul Horn, “Robot Vision,”
women and 93 men. Our representation is rich
enough to show promising results in these tasks. MIT Press, 1986.
826 | P a g e
6. Sudheer Reddy Bandi, A Gopi Suresh , Immanuel Alphonse / International Journal of
Engineering Research and Applications (IJERA) ISSN: 2248-9622 www.ijera.com
Vol. 2, Issue 5, September- October 2012, pp.822-827
[12] J.-H.Yoo, D.Hwang, and M. S. Nixon,
“Gender classification in human gait using
support vector machine,” in Proc. Advanced
Concepts for Intelligent Vision Systems,
Antwerp, Belgium, Sep. 2005, pp. 138–145.
[13] M. S. Nixon, T. N. Tan, and R. Chellappa.
Human Identification Based on Gait.
International Series on Biometrics. Springer-
Verlag, Dec. 2005.
[14] S. Niyogi and E. Adelson. Analyzing and
recognizing walking figures in xyt. In Proc.
of IEEE Conf. on Computer Vision and
Pattern Recognition, pages 469–474, 1994.
[15] J. Han and B. Bhanu. Individual recognition
using gait energy image. Trans. on Pattern
Analysis and Machine Intelligence,
28(2):316– 322, 2006.
[16] R. Begg. Support vector machines for
automated gait classification. IEEE Trans.
on Biomedical Engineering, 52(5):828–838,
May 2005.
[17] J. Davis. Visual categorization of children
and adult walking styles. In Proc. of the Int.
Conf. on Audio- and Video-based Biometric
Person Authentication, pages 295–300, Jun.
2001.
827 | P a g e