In recent years, fusion of multiple biometric modalities for personal authentication has received considerable attention. This paper presents a feature level fusion algorithm based on texture features. The system combines fingerprint, face and off-line signature. Texture features are extracted from Curvelet transform. The Curvelet feature dimension is selected based on d-prime number. The increase in feature dimension is reduced by using template averaging, moment features and by Principal component analysis (PCA). The algorithm is tested on in-house multimodal database comprising of 3000 samples and Chimeric databases. Identification performance of the system is evaluated using SVM classifier. A maximum GAR of 97.15% is achieved with Curvelet-PCA features.
QPLC: A Novel Multimodal Biometric Score Fusion MethodCSCJournals
In biometrics authentication systems, it has been shown that fusion of more than one modality (e.g., face and finger) and fusion of more than one classifier (two different algorithms) can improve the system performance. Often a score level fusion is adopted as this approach doesn’t require the vendors to reveal much about their algorithms and features. Many score level transformations have been proposed in the literature to normalize the scores which enable fusion of more than one classifier. In this paper, we propose a novel score level transformation technique that helps in fusion of multiple classifiers. The method is based on two components: quantile transform of the genuine and impostor score distributions and a power transform which further changes the score distribution to help linear classification. After the scores are normalized using the novel quantile power transform, several linear classifiers are proposed to fuse the scores of multiple classifiers. Using the NIST BSSR-1 dataset, we have shown that the results obtained by the proposed method far exceed the results published so far in the literature.
COMPRESSION BASED FACE RECOGNITION USING TRANSFORM DOMAIN FEATURES FUSED AT M...sipij
The physiological biometric trait face images are used to identify a person effectively. In this paper, we
propose compression based face recognition using transform domain features fused at matching level. The
2D images are converted into 1-D vectors using mean to compress number of pixels. The Fast Fourier
Transform (FFT) and Discrete Wavelet Transform (DWT) are used to extract features. The low and high
frequency coefficients of DWT are concatenated to obtained final DWT features. The performance
parameters are computed by comparing database and test image features of FFT and DWT using Euclidian
Distance (ED). The performance parameters of FFT and DWT are fused at matching level to obtain better
results. It is observed that the performance of proposed method is better than the existing methods.
MultiModal Identification System in Monozygotic TwinsCSCJournals
With the increase in the number of twin births in recent decades, there is a need to develop alternate approaches that can secure the biometric system. In this paper an effective fusion scheme is presented that combines information presented by multiple domain experts based on the rank-level fusion integration method. The developed multimodal biometric system possesses a number of unique qualities, starting from utilizing Fisher’s Linear Discriminant methods for face matching, Principal Component Analysis for fingerprint matching and Local binary pattern features for iris matching and fused the information for effective recognition and authentication The importance of considering these boundary conditions, such as twins, where the possibility of errors is maximum will lead us to design a more reliable and robust security system.The proposed approach is tested on a real database consisting of 50 pair of identical twin images and shows promising results compared to other techniques. The Receiver Operating Characteristics also shows that the proposed method is superior compared to other techniques under study.
Microarray Data Classification Using Support Vector MachineCSCJournals
DNA microarrays allow biologist to measure the expression of thousands of genes simultaneously on a small chip. These microarrays generate huge amount of data and new methods are needed to analyse them. In this paper, a new classification method based on support vector machine is proposed. The proposed method is used to classify gene expression data recorded on DNA microarrays. It is found that the proposed method is faster than neural network and the classification performance is not less than neural network.
Bimodal Biometric System using Multiple Transformation Features of Fingerprin...IDES Editor
The biometric technology is used to identify
individuals effectively compared to existing traditional
methods. In this paper we propose Bimodal Biometric System
using Multiple Transformation features of Fingerprint and
Iris (BBMFI). The iris image is preprocessed to generate iris
template. The two level Discrete Wavelet Transformation
(DWT) is applied on iris template and Discrete Cosine
Transformation (DCT) is performed on second level low
frequency band to generate DCT coefficients which results in
features of iris. The fingerprint is preprocessed to obtain
Region of Interest (ROI) and segmented into four cells. Then
the DWT is applied on each cell to derive approximation band
and detailed bands. The Fast Fourier Transformation (FFT)
is applied on approximation band to compute absolute values
that results in features of fingerprint. The iris features and
fingerprint features are fused by concatenation to obtain final
set of features. The final feature vector of test and database
are compared using Euclidean distance matching. It is observed
that the values of Total Success Rate (TSR), False Rejection
Rate (FRR) and False Acceptance Rate (FAR) are improved in
the proposed system compared to existing algorithm.
QPLC: A Novel Multimodal Biometric Score Fusion MethodCSCJournals
In biometrics authentication systems, it has been shown that fusion of more than one modality (e.g., face and finger) and fusion of more than one classifier (two different algorithms) can improve the system performance. Often a score level fusion is adopted as this approach doesn’t require the vendors to reveal much about their algorithms and features. Many score level transformations have been proposed in the literature to normalize the scores which enable fusion of more than one classifier. In this paper, we propose a novel score level transformation technique that helps in fusion of multiple classifiers. The method is based on two components: quantile transform of the genuine and impostor score distributions and a power transform which further changes the score distribution to help linear classification. After the scores are normalized using the novel quantile power transform, several linear classifiers are proposed to fuse the scores of multiple classifiers. Using the NIST BSSR-1 dataset, we have shown that the results obtained by the proposed method far exceed the results published so far in the literature.
COMPRESSION BASED FACE RECOGNITION USING TRANSFORM DOMAIN FEATURES FUSED AT M...sipij
The physiological biometric trait face images are used to identify a person effectively. In this paper, we
propose compression based face recognition using transform domain features fused at matching level. The
2D images are converted into 1-D vectors using mean to compress number of pixels. The Fast Fourier
Transform (FFT) and Discrete Wavelet Transform (DWT) are used to extract features. The low and high
frequency coefficients of DWT are concatenated to obtained final DWT features. The performance
parameters are computed by comparing database and test image features of FFT and DWT using Euclidian
Distance (ED). The performance parameters of FFT and DWT are fused at matching level to obtain better
results. It is observed that the performance of proposed method is better than the existing methods.
MultiModal Identification System in Monozygotic TwinsCSCJournals
With the increase in the number of twin births in recent decades, there is a need to develop alternate approaches that can secure the biometric system. In this paper an effective fusion scheme is presented that combines information presented by multiple domain experts based on the rank-level fusion integration method. The developed multimodal biometric system possesses a number of unique qualities, starting from utilizing Fisher’s Linear Discriminant methods for face matching, Principal Component Analysis for fingerprint matching and Local binary pattern features for iris matching and fused the information for effective recognition and authentication The importance of considering these boundary conditions, such as twins, where the possibility of errors is maximum will lead us to design a more reliable and robust security system.The proposed approach is tested on a real database consisting of 50 pair of identical twin images and shows promising results compared to other techniques. The Receiver Operating Characteristics also shows that the proposed method is superior compared to other techniques under study.
Microarray Data Classification Using Support Vector MachineCSCJournals
DNA microarrays allow biologist to measure the expression of thousands of genes simultaneously on a small chip. These microarrays generate huge amount of data and new methods are needed to analyse them. In this paper, a new classification method based on support vector machine is proposed. The proposed method is used to classify gene expression data recorded on DNA microarrays. It is found that the proposed method is faster than neural network and the classification performance is not less than neural network.
Bimodal Biometric System using Multiple Transformation Features of Fingerprin...IDES Editor
The biometric technology is used to identify
individuals effectively compared to existing traditional
methods. In this paper we propose Bimodal Biometric System
using Multiple Transformation features of Fingerprint and
Iris (BBMFI). The iris image is preprocessed to generate iris
template. The two level Discrete Wavelet Transformation
(DWT) is applied on iris template and Discrete Cosine
Transformation (DCT) is performed on second level low
frequency band to generate DCT coefficients which results in
features of iris. The fingerprint is preprocessed to obtain
Region of Interest (ROI) and segmented into four cells. Then
the DWT is applied on each cell to derive approximation band
and detailed bands. The Fast Fourier Transformation (FFT)
is applied on approximation band to compute absolute values
that results in features of fingerprint. The iris features and
fingerprint features are fused by concatenation to obtain final
set of features. The final feature vector of test and database
are compared using Euclidean distance matching. It is observed
that the values of Total Success Rate (TSR), False Rejection
Rate (FRR) and False Acceptance Rate (FAR) are improved in
the proposed system compared to existing algorithm.
Efficient Small Template Iris Recognition System Using Wavelet TransformCSCJournals
Iris recognition is known as an inherently reliable biometric technique for human identification. Feature extraction is a crucial step in iris recognition, and the trend nowadays is to reduce the size of the extracted features. Special efforts have been applied in order to obtain low templates size and fast verification algorithms. These efforts are intended to enable a human authentication in small embedded systems, such as an Integrated Circuit smart card. In this paper, an effective eyelids removing method, based on masking the iris, has been applied. Moreover, an efficient iris recognition encoding algorithm has been employed. Different combination of wavelet coefficients which quantized with multiple quantization levels are used and the best wavelet coefficients and quantization levels are determined. The system is based on an empirical analysis of CASIA iris database images. Experimental results show that this algorithm is efficient and gives promising results of False Accept Ratio (FAR) = 0% and False Reject Ratio (FRR) = 1% with a template size of only 364 bits.
Personal identification using multibiometrics score level fusioneSAT Publishing House
IJRET : International Journal of Research in Engineering and Technology is an international peer reviewed, online journal published by eSAT Publishing House for the enhancement of research in various disciplines of Engineering and Technology. The aim and scope of the journal is to provide an academic medium and an important reference for the advancement and dissemination of research results that support high-level learning, teaching and research in the fields of Engineering and Technology. We bring together Scientists, Academician, Field Engineers, Scholars and Students of related fields of Engineering and Technology.
Performance Enhancement Of Multimodal Biometrics Using CryptosystemIJERA Editor
Multimodal biometrics means the unification of two or more uni modal biometrics so as to make the system more reliable and secure. Such systems promise better security. This study is a blend of iris and fingerprint recognition technique and their fusion at feature level. Our work comprises of two main sections: feature extraction of both modalities and fusing them before matching and finally application of an encryption technique to enhance the security of the fused template.
Feature Level Fusion of Multibiometric Cryptosystem in Distributed SystemIJMER
ABSTRACT: Multibiometrics is the combination of one or more biometrics (e.g., Fingerprint, Iris, and Face). Researchers
are focusing on how to provide security to the system, the template which was generated from the biometric need to be
protected. The problems of unimodal biometrics are solved by multibiometrics. The main objective is to provide a security to
the biometric template by generating a secure sketch by making use of multibiometric cryptosystem and which is stored in a
database. Once the biometric template is stolen it becomes a serious issue for the security of the system and also for user
privacy. In the existing approach, feature level fusion is used to combine the features securely with well-known biometric
cryptosystems namely fuzzy vault and fuzzy commitment. The drawbacks of existing system include accuracy of the biometric
need to be improved and the noises in the biometrics also need to be reduced. The proposed work is to enhance the security
using multibiometric cryptosystem in distributed system applications like e-commerce transactions, e-banking and ATM.
Keywords: Biometric Cryptosystem, Error correcting code, Fingerprint, Iris, Multibiometrics, Unimodal biometrics.
Robust Analysis of Multibiometric Fusion Versus Ensemble Learning Schemes: A ...CSCJournals
Identification of person using multiple biometric is very common approach used in existing user
validation of systems. Most of multibiometric system depends on fusion schemes, as much of the
fusion techniques have shown promising results in literature, due to the fact of combining multiple
biometric modalities with suitable fusion schemes. However, similar type of practices are found in
ensemble of classifiers, which increases the classification accuracy while combining different
types of classifiers. In this paper, we have evaluated comparative study of traditional fusion
methods like feature level and score level fusion with the well-known ensemble methods such as
bagging and boosting. Precisely, for our frame work experimentations, we have fused face and
palmprint modalities and we have employed probability model - Naive Bayes (NB), neural
network model - Multi Layer Perceptron (MLP), supervised machine learning algorithm - Support
Vector Machine (SVM) classifiers for our experimentation. Nevertheless, machine learning
ensemble approaches namely, Boosting and Bagging are statistically well recognized. From
experimental results, in biometric fusion the traditional method, score level fusion is highly
recommended strategy than ensemble learning techniques.
COMPARATIVE ANALYSIS OF MINUTIAE BASED FINGERPRINT MATCHING ALGORITHMSijcsit
Biometric matching involves finding similarity between fingerprint images.The accuracy and speed of the
matching algorithmdetermines its effectives. This researchaims at comparing two types of matching
algorithms namely(a) matching using global orientation features and (b) matching using minutia triangulation.The comparison is done using accuracy, time and number of similar features. The experiment is conducted on a datasets of 100 candidates using four (4) fingerprints from each candidate. The data is sampled from a mass registration conducted by a reputable organization in Kenya.Theresearch reveals that fingerprint matching based on algorithm (b) performs better in speed with an average of 38.32 milliseconds
as compared to matching based on algorithm (a) with an average of 563.76 milliseconds. On accuracy,algorithm(a) performs better with an average accuracy of 0.142433 as compared to algorithm (b) with an average accuracy score of 0.004202.
IJRET : International Journal of Research in Engineering and Technology is an international peer reviewed, online journal published by eSAT Publishing House for the enhancement of research in various disciplines of Engineering and Technology. The aim and scope of the journal is to provide an academic medium and an important reference for the advancement and dissemination of research results that support high-level learning, teaching and research in the fields of Engineering and Technology. We bring together Scientists, Academician, Field Engineers, Scholars and Students of related fields of Engineering and Technology.
The Survey of Architecture of Multi-Modal (Fingerprint and Iris Recognition) ...IJERA Editor
Biometrics based individual identification is observed as an effective technique for automatically knowing, with a high confidence a person’s identity. Multi-modal biometric systems consolidate the evidence accessible by multiple biometric sources and normally better recognition performance associate to system based on a single biometric modality.Multi biometric systems are used to overcome this issue by providing multiple pieces of indication of the same identity. This system provides effective fusion structure that combines information provided by the multiple field experts based on decision-level and score-level fusion method, thereby increasing the efficiency which is not conceivable in uni-modal system.Multi-modal biometrics can be attained through a fusion of two or more images, where the subsequent fused image will be more protected. This paper discusses various fusion techniques, architecture of multi-modal biometric authentication and working of biometric fusion i.e. Iris and Fingerprint recognition that are used in multi-modal biometrics
Face Recognition Based Intelligent Door Control Systemijtsrd
This paper presents the intelligent door control system based on face detection and recognition. This system can avoid the need to control by persons with the use of keys, security cards, password or pattern to open the door. The main objective is to develop a simple and fast recognition system for personal identification and face recognition to provide the security system. Face is a complex multidimensional structure and needs good computing techniques for recognition. The system is composed of two main parts face recognition and automatic door access control. It needs to detect the face before recognizing the face of the person. In face detection step, Viola Jones face detection algorithm is applied to detect the human face. Face recognition is implemented by using the Principal Component Analysis PCA and Neural Network. Image processing toolbox which is in MATLAB 2013a is used for the recognition process in this research. The PIC microcontroller is used to automatic door access control system by programming MikroC language. The door is opened automatically for the known person according to the result of verification in the MATLAB. On the other hand, the door remains closed for the unknown person. San San Naing | Thiri Oo Kywe | Ni Ni San Hlaing ""Face Recognition Based Intelligent Door Control System"" Published in International Journal of Trend in Scientific Research and Development (ijtsrd), ISSN: 2456-6470, Volume-3 | Issue-4 , June 2019, URL: https://www.ijtsrd.com/papers/ijtsrd23893.pdf
Paper URL: https://www.ijtsrd.com/engineering/electrical-engineering/23893/face-recognition-based-intelligent-door-control-system/san-san-naing
Multibiometric Secure Index Value Code Generation for Authentication and Retr...ijsrd.com
The use of multiple biometric sources for human recognition, referred to as multibiometrics, mitigates some of the limitations of unimodal biometric systems by increasing recognition accuracy, improving population coverage, imparting fault-tolerance, and enhancing security. In a biometric identification system, the identity corresponding to the input data (probe) is typically determined by comparing it against the templates of all identities in a database (gallery). An alternative e approach is to limit the number of identities against which matching is performed based on criteria that are fast to evaluate. We propose a method for generating fixed-length codes for indexing biometric databases. An index code is constructed by computing match scores between a biometric image and a fixed set of reference images. Candidate identities are retrieved based on the similarity between the index code of the probe image and those of the identities in the database. The number of multibiometric systems deployed on a national scale is increasing and the sizes of the underlying databases are growing. These databases are used extensively, thereby requiring efficient ways for searching and retrieving relevant identities. Searching a biometric database for an identity is usually done by comparing the probe image against every enrolled identity in the database and generating a ranked list of candidate identities. Depending on the nature of the matching algorithm, the matching speed in some systems can be slow. The proposed technique can be easily extended to retrieve pertinent identities from multimodal databases. Experiments on a chimeric face and fingerprint bimodal database resulted in an 84% average reduction in the search space at a hit rate of 100%. These results suggest that the proposed indexing scheme has the potential to substantially reduce the response time without compromising the accuracy of identification. New representation schemes that allow for faster search and, therefore, shorter response time are needed.
IJRET : International Journal of Research in Engineering and Technology is an international peer reviewed, online journal published by eSAT Publishing House for the enhancement of research in various disciplines of Engineering and Technology. The aim and scope of the journal is to provide an academic medium and an important reference for the advancement and dissemination of research results that support high-level learning, teaching and research in the fields of Engineering and Technology. We bring together Scientists, Academician, Field Engineers, Scholars and Students of related fields of Engineering and Technology.
An Indexing Technique Based on Feature Level Fusion of Fingerprint FeaturesIDES Editor
Personal identification system based on pass word
and other entities are ineffective. Nowadays biometric based
systems are used for human identification in almost many
real time applications. The current state-of-art biometric
identification focuses on accuracy and hence a good
performance result in terms of response time on small scale
database is achieved. But in today’s real life scenario biometric
database are huge and without any intelligent scheme the
response time should be high, but the existing algorithms
requires an exhaustive search on the database which increases
proportionally when the size of the database grows. This paper
addresses the problem of biometric indexing in the context of
fingerprint. Indexing is a technique to reduce the number of
candidate identities to be considered by the identification
algorithm. The fingerprint indexing methodology projected
in this work is based on a combination of Level 1, Level 2 and
Level-3 fingerprint features. The result shows the fusion of
level 1, level 2 and level 3 features gives better performance
and good indexing rate than with any one level of fingerprint
feature.
Scale Invariant Feature Transform Based Face Recognition from a Single Sample...ijceronline
International Journal of Computational Engineering Research (IJCER) is dedicated to protecting personal information and will make every reasonable effort to handle collected information appropriately. All information collected, as well as related requests, will be handled as carefully and efficiently as possible in accordance with IJCER standards for integrity and objectivity.
Paper multi-modal biometric system using fingerprint , face and speechAalaa Khattab
Biometric system is often not able to meet the desired performance requirements.
In order to enable a biometric system to operate effectively in different applications and environments, a multimodal biometric system is preferred.
In this paper introduce a multimodal biometric system which integrates fingerprint verification , face recognition and speaker verification.
Efficient Small Template Iris Recognition System Using Wavelet TransformCSCJournals
Iris recognition is known as an inherently reliable biometric technique for human identification. Feature extraction is a crucial step in iris recognition, and the trend nowadays is to reduce the size of the extracted features. Special efforts have been applied in order to obtain low templates size and fast verification algorithms. These efforts are intended to enable a human authentication in small embedded systems, such as an Integrated Circuit smart card. In this paper, an effective eyelids removing method, based on masking the iris, has been applied. Moreover, an efficient iris recognition encoding algorithm has been employed. Different combination of wavelet coefficients which quantized with multiple quantization levels are used and the best wavelet coefficients and quantization levels are determined. The system is based on an empirical analysis of CASIA iris database images. Experimental results show that this algorithm is efficient and gives promising results of False Accept Ratio (FAR) = 0% and False Reject Ratio (FRR) = 1% with a template size of only 364 bits.
Personal identification using multibiometrics score level fusioneSAT Publishing House
IJRET : International Journal of Research in Engineering and Technology is an international peer reviewed, online journal published by eSAT Publishing House for the enhancement of research in various disciplines of Engineering and Technology. The aim and scope of the journal is to provide an academic medium and an important reference for the advancement and dissemination of research results that support high-level learning, teaching and research in the fields of Engineering and Technology. We bring together Scientists, Academician, Field Engineers, Scholars and Students of related fields of Engineering and Technology.
Performance Enhancement Of Multimodal Biometrics Using CryptosystemIJERA Editor
Multimodal biometrics means the unification of two or more uni modal biometrics so as to make the system more reliable and secure. Such systems promise better security. This study is a blend of iris and fingerprint recognition technique and their fusion at feature level. Our work comprises of two main sections: feature extraction of both modalities and fusing them before matching and finally application of an encryption technique to enhance the security of the fused template.
Feature Level Fusion of Multibiometric Cryptosystem in Distributed SystemIJMER
ABSTRACT: Multibiometrics is the combination of one or more biometrics (e.g., Fingerprint, Iris, and Face). Researchers
are focusing on how to provide security to the system, the template which was generated from the biometric need to be
protected. The problems of unimodal biometrics are solved by multibiometrics. The main objective is to provide a security to
the biometric template by generating a secure sketch by making use of multibiometric cryptosystem and which is stored in a
database. Once the biometric template is stolen it becomes a serious issue for the security of the system and also for user
privacy. In the existing approach, feature level fusion is used to combine the features securely with well-known biometric
cryptosystems namely fuzzy vault and fuzzy commitment. The drawbacks of existing system include accuracy of the biometric
need to be improved and the noises in the biometrics also need to be reduced. The proposed work is to enhance the security
using multibiometric cryptosystem in distributed system applications like e-commerce transactions, e-banking and ATM.
Keywords: Biometric Cryptosystem, Error correcting code, Fingerprint, Iris, Multibiometrics, Unimodal biometrics.
Robust Analysis of Multibiometric Fusion Versus Ensemble Learning Schemes: A ...CSCJournals
Identification of person using multiple biometric is very common approach used in existing user
validation of systems. Most of multibiometric system depends on fusion schemes, as much of the
fusion techniques have shown promising results in literature, due to the fact of combining multiple
biometric modalities with suitable fusion schemes. However, similar type of practices are found in
ensemble of classifiers, which increases the classification accuracy while combining different
types of classifiers. In this paper, we have evaluated comparative study of traditional fusion
methods like feature level and score level fusion with the well-known ensemble methods such as
bagging and boosting. Precisely, for our frame work experimentations, we have fused face and
palmprint modalities and we have employed probability model - Naive Bayes (NB), neural
network model - Multi Layer Perceptron (MLP), supervised machine learning algorithm - Support
Vector Machine (SVM) classifiers for our experimentation. Nevertheless, machine learning
ensemble approaches namely, Boosting and Bagging are statistically well recognized. From
experimental results, in biometric fusion the traditional method, score level fusion is highly
recommended strategy than ensemble learning techniques.
COMPARATIVE ANALYSIS OF MINUTIAE BASED FINGERPRINT MATCHING ALGORITHMSijcsit
Biometric matching involves finding similarity between fingerprint images.The accuracy and speed of the
matching algorithmdetermines its effectives. This researchaims at comparing two types of matching
algorithms namely(a) matching using global orientation features and (b) matching using minutia triangulation.The comparison is done using accuracy, time and number of similar features. The experiment is conducted on a datasets of 100 candidates using four (4) fingerprints from each candidate. The data is sampled from a mass registration conducted by a reputable organization in Kenya.Theresearch reveals that fingerprint matching based on algorithm (b) performs better in speed with an average of 38.32 milliseconds
as compared to matching based on algorithm (a) with an average of 563.76 milliseconds. On accuracy,algorithm(a) performs better with an average accuracy of 0.142433 as compared to algorithm (b) with an average accuracy score of 0.004202.
IJRET : International Journal of Research in Engineering and Technology is an international peer reviewed, online journal published by eSAT Publishing House for the enhancement of research in various disciplines of Engineering and Technology. The aim and scope of the journal is to provide an academic medium and an important reference for the advancement and dissemination of research results that support high-level learning, teaching and research in the fields of Engineering and Technology. We bring together Scientists, Academician, Field Engineers, Scholars and Students of related fields of Engineering and Technology.
The Survey of Architecture of Multi-Modal (Fingerprint and Iris Recognition) ...IJERA Editor
Biometrics based individual identification is observed as an effective technique for automatically knowing, with a high confidence a person’s identity. Multi-modal biometric systems consolidate the evidence accessible by multiple biometric sources and normally better recognition performance associate to system based on a single biometric modality.Multi biometric systems are used to overcome this issue by providing multiple pieces of indication of the same identity. This system provides effective fusion structure that combines information provided by the multiple field experts based on decision-level and score-level fusion method, thereby increasing the efficiency which is not conceivable in uni-modal system.Multi-modal biometrics can be attained through a fusion of two or more images, where the subsequent fused image will be more protected. This paper discusses various fusion techniques, architecture of multi-modal biometric authentication and working of biometric fusion i.e. Iris and Fingerprint recognition that are used in multi-modal biometrics
Face Recognition Based Intelligent Door Control Systemijtsrd
This paper presents the intelligent door control system based on face detection and recognition. This system can avoid the need to control by persons with the use of keys, security cards, password or pattern to open the door. The main objective is to develop a simple and fast recognition system for personal identification and face recognition to provide the security system. Face is a complex multidimensional structure and needs good computing techniques for recognition. The system is composed of two main parts face recognition and automatic door access control. It needs to detect the face before recognizing the face of the person. In face detection step, Viola Jones face detection algorithm is applied to detect the human face. Face recognition is implemented by using the Principal Component Analysis PCA and Neural Network. Image processing toolbox which is in MATLAB 2013a is used for the recognition process in this research. The PIC microcontroller is used to automatic door access control system by programming MikroC language. The door is opened automatically for the known person according to the result of verification in the MATLAB. On the other hand, the door remains closed for the unknown person. San San Naing | Thiri Oo Kywe | Ni Ni San Hlaing ""Face Recognition Based Intelligent Door Control System"" Published in International Journal of Trend in Scientific Research and Development (ijtsrd), ISSN: 2456-6470, Volume-3 | Issue-4 , June 2019, URL: https://www.ijtsrd.com/papers/ijtsrd23893.pdf
Paper URL: https://www.ijtsrd.com/engineering/electrical-engineering/23893/face-recognition-based-intelligent-door-control-system/san-san-naing
Multibiometric Secure Index Value Code Generation for Authentication and Retr...ijsrd.com
The use of multiple biometric sources for human recognition, referred to as multibiometrics, mitigates some of the limitations of unimodal biometric systems by increasing recognition accuracy, improving population coverage, imparting fault-tolerance, and enhancing security. In a biometric identification system, the identity corresponding to the input data (probe) is typically determined by comparing it against the templates of all identities in a database (gallery). An alternative e approach is to limit the number of identities against which matching is performed based on criteria that are fast to evaluate. We propose a method for generating fixed-length codes for indexing biometric databases. An index code is constructed by computing match scores between a biometric image and a fixed set of reference images. Candidate identities are retrieved based on the similarity between the index code of the probe image and those of the identities in the database. The number of multibiometric systems deployed on a national scale is increasing and the sizes of the underlying databases are growing. These databases are used extensively, thereby requiring efficient ways for searching and retrieving relevant identities. Searching a biometric database for an identity is usually done by comparing the probe image against every enrolled identity in the database and generating a ranked list of candidate identities. Depending on the nature of the matching algorithm, the matching speed in some systems can be slow. The proposed technique can be easily extended to retrieve pertinent identities from multimodal databases. Experiments on a chimeric face and fingerprint bimodal database resulted in an 84% average reduction in the search space at a hit rate of 100%. These results suggest that the proposed indexing scheme has the potential to substantially reduce the response time without compromising the accuracy of identification. New representation schemes that allow for faster search and, therefore, shorter response time are needed.
IJRET : International Journal of Research in Engineering and Technology is an international peer reviewed, online journal published by eSAT Publishing House for the enhancement of research in various disciplines of Engineering and Technology. The aim and scope of the journal is to provide an academic medium and an important reference for the advancement and dissemination of research results that support high-level learning, teaching and research in the fields of Engineering and Technology. We bring together Scientists, Academician, Field Engineers, Scholars and Students of related fields of Engineering and Technology.
An Indexing Technique Based on Feature Level Fusion of Fingerprint FeaturesIDES Editor
Personal identification system based on pass word
and other entities are ineffective. Nowadays biometric based
systems are used for human identification in almost many
real time applications. The current state-of-art biometric
identification focuses on accuracy and hence a good
performance result in terms of response time on small scale
database is achieved. But in today’s real life scenario biometric
database are huge and without any intelligent scheme the
response time should be high, but the existing algorithms
requires an exhaustive search on the database which increases
proportionally when the size of the database grows. This paper
addresses the problem of biometric indexing in the context of
fingerprint. Indexing is a technique to reduce the number of
candidate identities to be considered by the identification
algorithm. The fingerprint indexing methodology projected
in this work is based on a combination of Level 1, Level 2 and
Level-3 fingerprint features. The result shows the fusion of
level 1, level 2 and level 3 features gives better performance
and good indexing rate than with any one level of fingerprint
feature.
Scale Invariant Feature Transform Based Face Recognition from a Single Sample...ijceronline
International Journal of Computational Engineering Research (IJCER) is dedicated to protecting personal information and will make every reasonable effort to handle collected information appropriately. All information collected, as well as related requests, will be handled as carefully and efficiently as possible in accordance with IJCER standards for integrity and objectivity.
Paper multi-modal biometric system using fingerprint , face and speechAalaa Khattab
Biometric system is often not able to meet the desired performance requirements.
In order to enable a biometric system to operate effectively in different applications and environments, a multimodal biometric system is preferred.
In this paper introduce a multimodal biometric system which integrates fingerprint verification , face recognition and speaker verification.
Multimodal biometric systems are those that utilize more than one physical or behavioural characteristic for enrolment , verification, or identification.
Feature Extraction using Sparse SVD for Biometric Fusion in Multimodal Authen...IJNSA Journal
Token based security (ID Cards) have been used to restrict access to the Secured systems. The purpose of
Biometrics is to identify / verify the correctness of an individual by using certain physiological or
behavioural traits associated with the person. Current biometric systems make use of face, fingerprints,
iris, hand geometry, retina, signature, palm print, voiceprint and so on to establish a person’s identity.
Biometrics is one of the primary key concepts of real application domains such as aadhar card, passport,
pan card, etc. In this paper, we consider face and fingerprint patterns for identification/verification. Using
this data we proposed a novel model for authentication in multimodal biometrics often called ContextSensitive Exponent Associative Memory Model (CSEAM). It provides different stages of security for
biometrics fusion patterns. In stage 1, fusion of face and finger patterns using Principal Component
Analysis (PCA), in stage 2 by applying Sparse SVD decomposition to extract the feature patterns from the
fusion data and face pattern and then in stage 3, using CSEAM model, the extracted feature vectors can be
encoded. The final key will be stored in the smart cards as Associative Memory (M), which is often called
Context-Sensitive Associative Memory (CSAM). In CSEAM model, the CSEAM will be computed using
exponential kronecker product for encoding and verification of the chosen samples from the users. The
exponential of matrix can be computed in various ways such as Taylor Series, Pade Approximation and
also using Ordinary Differential Equations (O.D.E.). Among these approaches we considered first two
methods for computing exponential of a feature space. The result analysis of SVD and Sparse SVD for
feature extraction process and also authentication/verification process of the proposed system in terms of
performance measures as Mean square error rates will be presented.
Integrating Fusion levels for Biometric Authentication SystemIOSRJECE
— Recently a lot of works are presented in the literature for the multimodal biometric authentication. And such biometric systems have been widely accepted with increasing accuracy rates and population coverage, reducing the vulnerability to spoofing. This paper descripts about the proposed multimodal biometric system that combines the feature extraction level and the score level fusion of iris and face unimodal biometric systems in order to take advantage of both fusion techniques. The experimental results shows the performance of the multimodal and multilevel fusion techniques which are analysed using TRR and TAR to study the recognition behaviour of the approach system. From the ROC Curve plotted, the performance of the proposed system is better compared to the individual fusion techniques.
A NOVEL BINNING AND INDEXING APPROACH USING HAND GEOMETRY AND PALM PRINT TO E...ijcsa
This paper proposes a Bio metric identification system for person identification using two bio metric traits
hand geometry and palm print. The hand image captured from digital camera is preprocessed to identify
key points on palm region of hand. Identified key points are used to find hand geometry feature and palm
print Region of interest (ROI). The discriminative palm print features are extracted by applying local
binary descriptor on palm print ROI. In a bio metric identification system the identity corresponding to the
input image (probe) is determined by comparing probe template with the templates of all identities enrolled
in biometric system (gallery). Response time to establish the identity of an individual increases in proportion to the number of enrollees. One way to reduce the response time is to retrieve a smaller set of candidate identity templates from the database for explicit comparison. In this paper we propose a coarseto-fine hierarchical approach to retrieve a smaller set of candidate identities called as candidate set to reduce the response time. The proposed approach is tested on the database collected at our institute.Proposed approach is of significance since hand geometry and palm print features can be extracted from the palm region of the hand. Also performance of identification system is enhanced by reducing the response time without compromising the identification accuracy.
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.
Ijaems apr-2016-1 Multibiometric Authentication System Processed by the Use o...INFOGAIN PUBLICATION
The present day authentication system is mostly uni-model i.e having only single authentication method which can be either finger print, iris , palm veins ,etc. Thus these models have to contend with a variety of problems such as absurd or unusual data, non-versatility; un authorized attempts, and huge amount of error rates. Some of these limitations can be reduced or stopped by the use of multimodal biometric systems that integrate the evidence presented by several sources of information. This paper converses a multi biometric based authentication system based on Fusion algorithm using a key. Our work mainly focuses on the fusion algorithm, i.e fusion of finger and palm print out of which the greatest features from the above two traits are taken into account. With minimum possible features the fusion of the both the traits is carried out. Then some key is generated for multi biometric authentication. By processing the test image of a person, the identity of the person is displayed with his/her own image. By the fusion algorithm, it is found that it has less computation time compared to the existing algorithms. By matching results, we validate and authenticate the particular individual.
Performance of Hasty and Consistent Multi Spectral Iris Segmentation using De...ijtsrd
The recognition system is composed of seven phases acquisition, preprocessing, segmentation, normalization, feature extraction, feature selection, and classification. In the acquisition phase, iris images are captured, followed by preprocessing to enhance the quality of the images. The segmentation phase involves separating the iris region from the background, and the normalized iris region is shaped into a rectangle in the normalization phase. Iris segmentation is a critical step in iris recognition systems and has a direct impact on authentication and recognition results. However, standard segmentation techniques may not perform well in noisy iris databases captured under challenging conditions. Moreover, the lack of large iris databases hinders the performance improvement of convolution neural networks. The proposed method addresses these challenges by effectively handling irregular iris images captured under visible light. The iris region is processed and evaluated to generate a unique feature vector, which is then used for person identification. VGG16, a well known deep learning model, is employed for image classification, and the feature vector is fed into VGG16 for classification purposes. Ram Niwas Sharma | Ankit Kumar Navalakha | Neha Sharma "Performance of Hasty and Consistent Multi Spectral Iris Segmentation using Deep Learning" Published in International Journal of Trend in Scientific Research and Development (ijtsrd), ISSN: 2456-6470, Volume-7 | Issue-5 , October 2023, URL: https://www.ijtsrd.com/papers/ijtsrd59853.pdf Paper Url: https://www.ijtsrd.com/engineering/computer-engineering/59853/performance-of-hasty-and-consistent-multi-spectral-iris-segmentation-using-deep-learning/ram-niwas-sharma
Face recognition is a widely used biometric approach. Face recognition technology has developed rapidly
in recent years and it is more direct, user friendly and convenient compared to other methods. But face
recognition systems are vulnerable to spoof attacks made by non-real faces. It is an easy way to spoof face
recognition systems by facial pictures such as portrait photographs. A secure system needs Liveness
detection in order to guard against such spoofing. In this work, face liveness detection approaches are
categorized based on the various types techniques used for liveness detection. This categorization helps
understanding different spoof attacks scenarios and their relation to the developed solutions. A review of
the latest works regarding face liveness detection works is presented. The main aim is to provide a simple
path for the future development of novel and more secured face liveness detection approach.
Verification or Authentication systems use a single biometric sensor which having higher error rate due to single evidence of identity (voice can be change due to cold, face can be changed due facial hairs, cosmetics, fingerprint can be change due to scar etc.). To enhance the performance of single biometric systems in these situations may not be effective because of these problems. Multi-biometric systems overcome some of these limitations by providing multiple proofs of any identity. This paper presents an effective multimodal biometric system which can be used to reduce the above mentioned drawbacks of unimodal systems.
BIOMETRIC BASED AUTHENTICATION SYSTEM TECHNOLOGY
Multimodal authentication is one of the prime concepts in current applications of real scenario. Various
approaches have been proposed in this aspect. In this paper, an intuitive strategy is proposed as a
framework for providing more secure key in biometric security aspect. Initially the features will be
extracted through PCA by SVD from the chosen biometric patterns, then using LU factorization technique
key components will be extracted, then selected with different key sizes and then combined the selected key
components using convolution kernel method (Exponential Kronecker Product - eKP) as Context-Sensitive
Exponent Associative Memory model (CSEAM). In the similar way, the verification process will be done
and then verified with the measure MSE. This model would give better outcome when compared with SVD
factorization[1] as feature selection. The process will be computed for different key sizes and the results
will be presented.
Unit 8 - Information and Communication Technology (Paper I).pdfThiyagu K
This slides describes the basic concepts of ICT, basics of Email, Emerging Technology and Digital Initiatives in Education. This presentations aligns with the UGC Paper I syllabus.
Operation “Blue Star” is the only event in the history of Independent India where the state went into war with its own people. Even after about 40 years it is not clear if it was culmination of states anger over people of the region, a political game of power or start of dictatorial chapter in the democratic setup.
The people of Punjab felt alienated from main stream due to denial of their just demands during a long democratic struggle since independence. As it happen all over the word, it led to militant struggle with great loss of lives of military, police and civilian personnel. Killing of Indira Gandhi and massacre of innocent Sikhs in Delhi and other India cities was also associated with this movement.
Welcome to TechSoup New Member Orientation and Q&A (May 2024).pdfTechSoup
In this webinar you will learn how your organization can access TechSoup's wide variety of product discount and donation programs. From hardware to software, we'll give you a tour of the tools available to help your nonprofit with productivity, collaboration, financial management, donor tracking, security, and more.
How to Split Bills in the Odoo 17 POS ModuleCeline George
Bills have a main role in point of sale procedure. It will help to track sales, handling payments and giving receipts to customers. Bill splitting also has an important role in POS. For example, If some friends come together for dinner and if they want to divide the bill then it is possible by POS bill splitting. This slide will show how to split bills in odoo 17 POS.
The Art Pastor's Guide to Sabbath | Steve ThomasonSteve Thomason
What is the purpose of the Sabbath Law in the Torah. It is interesting to compare how the context of the law shifts from Exodus to Deuteronomy. Who gets to rest, and why?
How to Make a Field invisible in Odoo 17Celine George
It is possible to hide or invisible some fields in odoo. Commonly using “invisible” attribute in the field definition to invisible the fields. This slide will show how to make a field invisible in odoo 17.
Synthetic Fiber Construction in lab .pptxPavel ( NSTU)
Synthetic fiber production is a fascinating and complex field that blends chemistry, engineering, and environmental science. By understanding these aspects, students can gain a comprehensive view of synthetic fiber production, its impact on society and the environment, and the potential for future innovations. Synthetic fibers play a crucial role in modern society, impacting various aspects of daily life, industry, and the environment. ynthetic fibers are integral to modern life, offering a range of benefits from cost-effectiveness and versatility to innovative applications and performance characteristics. While they pose environmental challenges, ongoing research and development aim to create more sustainable and eco-friendly alternatives. Understanding the importance of synthetic fibers helps in appreciating their role in the economy, industry, and daily life, while also emphasizing the need for sustainable practices and innovation.
Multimodal Biometrics at Feature Level Fusion using Texture Features
1. Maya V. Karki & Dr. S. Sethu Selvi
International Journal of Biometrics and Bioinformatics (IJBB), Volume (7) : Issue (1) : 2013 58
Multimodal Biometrics at Feature Level Fusion using Texture
Features
Maya V. Karki mayavkarki@msrit.edu
Faculty, MSRIT, Dept. of E&C MSRIT
Bangalore-54, INDIA
Dr. S. Sethu Selvi selvi_selvan@yahoo.com
Faculty, MSRIT, Dept. of E&C MSRIT
Bangalore-54, INDIA
Abstract
In recent years, fusion of multiple biometric modalities for personal authentication has received
considerable attention. This paper presents a feature level fusion algorithm based on texture
features. The system combines fingerprint, face and off-line signature. Texture features are
extracted from Curvelet transform. The Curvelet feature dimension is selected based on d-prime
number. The increase in feature dimension is reduced by using template averaging, moment
features and by Principal component analysis (PCA). The algorithm is tested on in-house
multimodal database comprising of 3000 samples and Chimeric databases. Identification
performance of the system is evaluated using SVM classifier. A maximum GAR of 97.15% is
achieved with Curvelet-PCA features.
Keywords: Multimodal Biometrics, Feature Level, Curvelet Transform, Template Averaging, PCA
Features and SVM Classifier.
1. INTRODUCTION
Personal authentication systems built upon only one of the biometric traits are not fulfilling the
requirements of demanding applications in terms of universality, uniqueness, permanence,
collectability, performance, acceptability and circumvention. This has motivated the current
interest in multimodal biometrics [1] in which multiple biometric traits are simultaneously used in
order to make an identification decision. Depending on the number of traits, sensors and feature
sets used, a variety of scenarios are possible in a multimodal biometric system. They include
single biometric with multiple sensors, multiple biometric traits, single biometric with multi-
instances, single biometric with multiple representations and single biometric with multiple
matchers. Among all these scenarios, system with multiple biometric traits is gaining importance
and this method itself is known as multimodal biometric system. Based on the type of information
available in a certain module, different levels of fusion are defined [2]. Levels of fusion are broadly
classified into two categories: fusion before matching also called as pre-classification which
includes sensor level and feature level. Fusion after matching also called as post classification
which includes match score level and decision level. Amongst these, fusion at feature level is
gaining much research interest.
Most of the existing multimodal systems are based on either score level or decision level fusion
[3]. Match score is a measure of the similarity between the input and template biometric feature
vector. In match score level, scores are generated by multiple classifiers pertaining to different
biometric traits and combined [4]. In order to map score of different classifiers into a single
domain, where they possess a common meaning in terms of biometric performance,
normalization technique is applied to the output of classifier before score fusion. Gupta [5]
developed a multimodal system based on fingerprint, face, iris and signature with score level
2. Maya V. Karki & Dr. S. Sethu Selvi
International Journal of Biometrics and Bioinformatics (IJBB), Volume (7) : Issue (1) : 2013 59
fusion. In all these systems texture features are extracted and score level and decision level
fusion are compared using SVM classifier. The most promising recent research is certainly the
information fusion at the matching score level invoking user specific weights and threshold levels.
Though a few multimodal systems developed are considered to be very accurate, still they are
not validated since, systems are tested on a medium size database.
Fusion at feature level involves integration of feature sets corresponding to multiple biometric
traits. Feature set contains rich information about biometric data than the match score or final
decision. Therefore integration at this level is expected to give improved recognition performance.
Due to the constraints of feature level fusion, very few researchers have studied integration at
feature level. Chetty [6] combined face and voice using visual and acoustic features with artificial
neural network as a recognizer and obtained an Equal Error Rate (EER) of 2.5%. Nandakumar
[7] concatenated fingerprint and iris code at feature level using fuzzy vault classifier and showed
that uncorrelated features when combined gives best possible EER. Ferrer [8] proposed fusion of
features extracted from hand geometry, palmprint and fingerprint. It is imperative that an
appropriate feature selection scheme is used when combining information at the feature level.
This paper proposes a multimodal identification system that combines fingerprint, face and
signature biometric traits. These three traits are considered due to their wide acceptance by users
and also the data acquisition cost involved in these three traits are much less compared to other
biometrics. Texture features are extracted from each modality independently and fusion at feature
level is performed. Texture features are extracted from Curvelet Transform. Section 2 describes
the proposed multimodal biometric Identification system based on feature level fusion. Section
3 describes database collection protocol and pre-processing. Section 4 describes feature
extraction and dimension reduction techniques. Section 5 summarizes experimental results and
section 6 gives comparisons with similar work. Section 7 concludes proposed system.
2. PROPOSED MULTIMODAL IDENTIFICATION SYSTEM
The schematic of the multimodal system at feature level fusion is shown in Figure 1. The
multimodal system has two phases of operation: enrolment and identification. The enrolment
module registers a person and then the three biometric traits are acquired and representation of
these three traits are stored in a database. The proposed system is designed to operate in
parallel mode. Fingerprint, face and signature of a person are acquired separately in data
acquisition module. Required pre-processing techniques are applied on every biometric trait and
features are extracted simultaneously. Features from all three biometric traits are concatenated
and a feature vector is formed and stored as template in a database. Single matcher is used to
evaluate the performance. SVM classifier is used for matching. During authentication, feature
vector extracted from the test person is compared with the template stored in the database.
Matching is performed using a recognizer which compares query feature vector with the
template in the database and generates a match score. To prevent impostor from being
identified, the match score from matcher is compared with a predefined threshold in the decision
module. This module makes a decision as either person under test is recognized or not
recognized.
3. Maya V. Karki & Dr. S. Sethu Selvi
International Journal of Biometrics and Bioinformatics (IJBB), Volume (7) : Issue (1) : 2013 60
FIGURE 1: Block Diagram of Proposed Multimodal Biometric System Based on Feature Level Fusion.
2. DATABASE COLLECTION and PRE-PROCESSING
A multimodal database including fingerprint, signature and face samples are essential to test the
performance of the proposed system. Since there is no standard database freely available to
meet the requirement of the proposed algorithm, ECMSRIT multimodal database and Chimeric
databases have been formed. ECMSRIT database is collected from fingerprint, off-line signature
and face samples of 100 users. Collection of these unimodal traits are described below:
Nitgen fingerprint scanner is used to collect fingerprints. It is an optical sensor with ultra-precise
500dpi resolution. To locate centre point of fingerprint, it is divided into non-overlapping blocks.
Gradient in x and y direction at each pixel in a block is obtained. A 2D Gaussian filter is applied to
smooth the gradient. A slope perpendicular to direction of gradient in each block is computed.
Blocks with slope values ranging from 0 to pi/2 are considered. In each block a path is traced
down until a slope that is not ranging from 0 to pi/2 and that path is marked. Block with highest
number of marks gives slope in the negative y direction. This provides the centre point of
fingerprint. Region of interest around the centre point is cropped and normalized in size to 128
*128 pixels. Figure 2 represents centre point detection and cropping of fingerprint. Figure 2 (a)
represents scanned fingerprint, (b) shows orientation of fingerprint, (c) represents maximum
curvature points, (d) shows centre point and (e) shows cropped fingerprint and (f) shows third
level LL subband of Curvelet transformed fingerprint.
Still face images are collected using digital camera LifeCam Nx-6000 with a 2.0 mega pixels
sensor. The 2D colour face image is converted to a gray scale image. Canny edge detection
mask with suitable threshold value is applied on image with a uniform background to extract
outer curvature of the face. From this only foreground face image of size 128 * 128 is cropped
Figure 3 (a) shows edge detection and (b) represents cropped face and (d) represents third level
LL subband of Curvlet transformed face.
4. Maya V. Karki & Dr. S. Sethu Selvi
International Journal of Biometrics and Bioinformatics (IJBB), Volume (7) : Issue (1) : 2013 61
The signatures were taken on a A-4 size white paper. These were scanned using an 8-bit, 300
dpi resolution scanner. The scanned signatures were cut out from the scanned page in their
original orientation and size using an image editor. The scanned signature is binarized. Since the
signature consists of black pixels on a white background, the image is then complimented to
make it a white signature on a black background. When a signature is scanned, the image
obtained may contain some noise components like the background noise pixels and these noise
pixels are removed by employing median filter. To avoid inter-personal and intra-personal size
variations of signatures, size is normalized to 128 * 256. Figure 4 (a) shows input signature, (b)
noise removed, (c) complemented and (d) represents normalized signature sample.
The size of MSRIT database is 10 x 3 x 100= 3000. Chimeric database-I is formed by FVC2002-
DB3 fingerprint, ECMSRIT signature and ORL face databases. As ORL face database has only
40 users, Chimeric database-I is formed by considering only 40 users from fingerprint and
signature databases. Chimeric database-I consists of 8 samples of each person for each trait with
total of 8 x 3 x 40 = 960. Chimeric database-II is formed by FVC2004-DB3 fingerprint, CEDAR
signature and Faces-94 face databases. As the CEDAR signature database has only 55 users,
Chimeric database-II is formed by considering only 55 users from fingerprint and face databases.
FVC2004-DB3 has only 8 samples per user and Chimeric database-II consists of 8 samples of
each person for each modality with total of 8 x 3 x 55=1320 samples.
a) Input fingerprint (b) Orientation (c) Maximum curvature points
(d)Fingerprint with centre (e) Cropped fingerprint
FIGURE 2: Result of Pre-Processing of Fingerprint.
5. Maya V. Karki & Dr. S. Sethu Selvi
International Journal of Biometrics and Bioinformatics (IJBB), Volume (7) : Issue (1) : 2013 62
(a) Edge Detection. (b) Cropped Face.
FIGURE 3: Result of Pre-Processing of Face.
a) Input Signature (b) Noise Removed (c) Complemented (d) Normalised
FIGURE 4: Result of Pre-Processing of Signature.
4. FEATURE EXTRACTION
4.1 Extraction of Texture Features using Curvelet Transform
Texture features are extracted by applying Curvelet transform on each trait. Curvelet transform
[10,11,12] is based on multi-scale ridgelet transform [12] combined with spatial bandpass filtering
operation at different scales. It is better for representing point discontinuities. Figure 5 shows the
flow graph of Curvelet transform. The transform involves following steps: (1) The subbands of
input trait is obtained using DB4 wavelet transform. (2) The 2D fast Fourier transform of the LL
subband is obtained. (3) Using interpolation scheme, the samples of the Fourier transform
obtained on the square lattice is substituted with sampled values on a polar lattice. (4) Inverse
fast Fourier transform is computed on each radial line. (5) 1D Wavelet transform is computed at
each radial line using DB4 filter and approximate coefficients are used as features.
Steps 2 through 5 form Ridgelet transform [12] and Steps 2 to 4 represent finite Radon
transform[13] for digital data. Figure 6 (a), (b) and (c) show third level LL subband of Curvelet
transformed fingerprint, face and signature respectively. For example, consider a normalized
signature of size 128 x 256. Following the steps described above, third level Curvelet transformed
LL subband coefficients of size 25 x 20 will give feature dimension of 500.
Curvelet feature dimension is decided based on d-prime number (d’). Performance of the
biometric system has been predicted by calculating d’ value [3]. It measures separation between
the means of the genuine and impostor probability distributions in standard deviation unit. To
evaluate d’, genuine and impostor match scores are calculated. A match score is found to be
genuine if it is a result of matching two samples of the same user and is known as impostor score
if two samples of different users are compared. During training period all samples in the database
are considered to find genuine and impostor score. Let P be the number of persons enrolled in
6. Maya V. Karki & Dr. S. Sethu Selvi
International Journal of Biometrics and Bioinformatics (IJBB), Volume (7) : Issue (1) : 2013 63
the system and let S be the number of samples of the trait obtained from each person, then
number of genuine scores Gscore and number of impostor scores Iscore are given by
FIGURE 5: Flow Graph of Curvelet Transform.
( 1)
2
score
S
G P S= × − × (1)
2 ( 1)
2
score
P
I P S
−
= × × (2)
From these genuine and impostor scores mean and standard deviation are calculated to evaluate
d’. The d’ value is given by
2 2
' 2 genuine impostor
genuine impostor
d
µ µ
σ σ
−
=
+
(3)
Where µ and σ are the mean and standard deviation of genuine and impostor scores. For each
trait, different dimension of Curvelet features are evaluated and corresponding d’ value has been
calculated. Figure 7 shows the variation of d’ value for different feature dimensions. From the
graph it is observed that as feature dimension increases, d’ value increases and higher the value
of d’ better is performance. The d’ value remains constant for a feature dimension of 500 and
above. Hence, for each trait a maximum feature dimension of 504 has been considered to
evaluate recognition performance of the system. In feature level fusion the features from each
trait are concatenated. With concatenation feature dimension increases and to reduce the
dimension few reduction techniques [14] are adapted.
7. Maya V. Karki & Dr. S. Sethu Selvi
International Journal of Biometrics and Bioinformatics (IJBB), Volume (7) : Issue (1) : 2013 64
FIGURE 6: Third Level LL Subband of (a) Fingerprint (b) Face and (c) Signature.
FIGURE 7: Variation of d’ Value for Different Feature Dimension.
4.2 Dimension Reduction Techniques
Let FP, FS and FF be three feature vectors extracted by applying Curvelet transform on
fingerprint, signature and face respectively. Let fp, fs and ff be the dimension of each trait.
Feature vectors of three traits are represented by
[ ]1 2, ......FP P P fp=
[ ]1 2, ......,FS S S fs=
[ ]1 2, .....,FF F F Ff= (4)
All three feature vectors are concatenated to form a new feature vector Fc where
cF FP FS FF= + +
The dimension of Fc is equal to fp+fs+ff. In the proposed algorithm approximate coefficient
features of dimension 504 is extracted from each of the trait and concatenated, resulting in
dimension 504+504+504= 1512. Fc is stored as new template in the database for
matching.
(1) Template Averaging: The concatenated feature vector shows increase in size but
homogeneous in nature. Therefore the size of concatenated feature vector Fc is reduced by
8. Maya V. Karki & Dr. S. Sethu Selvi
International Journal of Biometrics and Bioinformatics (IJBB), Volume (7) : Issue (1) : 2013 65
applying averaging and this method is known as template averaging [3] and this method is very
simple. Average of feature vector calculated from three traits is
3
a
FP FS FF
F
+ +
= (5)
(2) PCA Features: In this method, to reduce the dimension of concatenated feature vector,
Principal Component Analysis (PCA) [15] is applied to Fc. PCA [16] transforms a number of
correlated variables into number of uncorrelated variables referred as principal components. PCA
reduces dimension without loss of information and preserves the global structure. Two
dimensional PCA (2DPCA) [14] has been applied on the concatenated features from three traits.
Algorithmic steps involved in applying 2DPCA are (1) Subband coefficients from Curvelet
transform are extracted from each trait and concatenated. Let A be the concatenated matrix.
Covariance of matrix of A is calculated. (2) Eigen values and eigen vectors of covariance matrix
are calculated. Eigen vector with highest eigen value is called as principal component of the
matrix A. Choosing first v eigen values from covariance matrix A, a transformed matrix B is
obtained as B=A x P where P=[P1,P2 …Pv ] is the projection matrix whose columns are the eigen
vectors of covariance matrix in the decreasing order of the eigen values. B is the required feature
matrix which is stored as PCA feature vector Fpca in the database for matching. By applying
Curvelet transform for fingerprint, face and signature, subband matrix of size 18x28, 18x28 and
20x25 are extracted and concatenated to form a matrix A of size 60x 25. 2DPCA is applied on A
and by considering first eight eigen values, the transformed matrix 60x8 gives a feature
dimension of 480.
(3) PCA Features without Fusion: To reduce the dimension at feature level fusion, 2DPCA is
applied on the subbands of Curvelet transform of each trait independently. The subband PCA
features are concatenated to form a feature vector Fp and stored as templates in the database.
2DPCA is applied on Curvelet subband obtained from each trait independently and selecting nine
largest eigen vectors from fingerprint, face and eight eigen vectors from signature, PCA feature of
dimension 162, 162 and 160 is obtained and concatenated to form feature dimension of 484.
(4) Statistical Moment Features without Fusion: Moment features are extracted from
subbands of Curvelet transform [18] from each trait and calculated as described below
Mean of each subband is calculated. Let µk be the mean value of k
th
subband. Second order
moment or variance of each subband σk is calculated using
( )( )
2
1 1
1
,
M N
k k k
i j
W i j
M N
σ µ
= =
= −
×
∑∑ (6)
Where Wk indicates the subband coefficients of k
th
ban and MxN be the size of subband.
Third order moment is calculated as
( )( )
3
1 1
3 ,
1
∑∑= =
−
×
=
M
i
N
j
kkk jiW
NM
µµ (7)
Fourth order moment is calculated as
( )( )
4
1 1
4 ,
1
∑∑= =
−
×
=
M
i
N
j
kkk jiW
NM
µµ (8)
9. Maya V. Karki & Dr. S. Sethu Selvi
International Journal of Biometrics and Bioinformatics (IJBB), Volume (7) : Issue (1) : 2013 66
Energy of each subband is calculated as
( )∑∑= =
=
M
i
N
j
kk jiWE
1 1
2
, (9)
The resulting feature vector of k
th
subband is given by
Fmk = [ µk σk µ3k µ4k Ek ]
These moment features from each trait are concatenated to form a new feature vector Fm and
stored in a database. First, second and third level Curvelet decompositions are applied on each
trait and results into 12 subbands. From these 12 subbands moment features of size 60 are
calculated and when concatenated from three traits give a feature dimension of 180.
(5) Feature Concatenation by extracting Significant Coefficients: Curvelet subband
coefficients are extracted from each trait and sorted. Significant coefficients from each trait are
selected and concatenated to form a feature vector Fr. Dimension of Fr is made comparable to
feature dimension of unimodal system. For example, by applying Curvelet transform on each trait
504 subband coefficient features are sorted and only first 168 features from each trait are
concatenated to form a feature vector of dimension 168+168+168= 504.
5. EXPERIMENTAL RESULTS
Performance of the proposed algorithm is tested for identification mode using SVM classifier. In
SVM classifier method [19, 20], each person in the database has an associated SVM. The test
samples are assigned the label of the person whose SVM gives the largest positive output. SVM
classifier with a polynomial kernel of order 2 is selected. Penalty parameter C is tuned from 2 to
10 to get better results. In this experiment, C is set to a value 2.
Samples in each of the databases are split into training and test set. Training and test samples
are selected in different ratios starting from 1:9,2:8,3:7 ....,9:1 and corresponding recognition rate
for five trials have been calculated using Euclidean distance measure. The average recognition
rate is calculated and result is compared with the results obtained from different sets of train and
test ratios and the ratio which gives maximum recognition rate is considered for performance
evaluation. In this experiment train to test ratio of 6:4 is considered and each database is
randomly split 30 times at each time performance of the system is evaluated and average of
these 30 times result is considered as final result. Genuine and impostor scores are calculated
for the six different feature vectors. Figure 7 shows histogram plots evaluated on ECMSRIT
database for six feature sets.
From figure8 it is seen that separation of genuine and impostor scores are more in Fpca and Fa
features compared to other features. Based on these distribution curves, the threshold is varied
to calculate FAR and FRR from which EER has been calculated for each of the feature vector.
Figure 9 shows threshold vs FAR and threshold vs FRR for all feature vectors evaluated. Figure
9 indicates FAR and FRR varies for each algorithm and each feature vector, resulting to variation
in EER. Table 1 indicates a minimum EER of 5.32% is obtained for Fpca and a maximum EER of
22.35% is obtained for Fr features.
10. Maya V. Karki & Dr. S. Sethu Selvi
International Journal of Biometrics and Bioinformatics (IJBB), Volume (7) : Issue (1) : 2013 67
Features Feature
Dimension
Optimal
Threshold
FAR (%) FRR (%) EER (%)
Fc 1512 0.46 19.85 18.54 19.31
Fa 504 0.45 12.07 11.78 12.00
Fpca 480 0.38 5.34 4.35 5.32
Fp 484 0.31 16.46 15.53 15.33
Fm 180 0.14 20.54 17.78 20.54
Fr 504 0.35 25.25 22.34 22.35
TABLE 1: EER (%) for Different Features.
FIGURE 8: Histogram Plots for Genuine and Impostor Scores.
11. Maya V. Karki & Dr. S. Sethu Selvi
International Journal of Biometrics and Bioinformatics (IJBB), Volume (7) : Issue (1) : 2013 68
FIGURE 9: FAR and FRR vs Threshold for Different Feature Sets.
Biometric Traits Feature Vector Feature Dimension GAR (%)
Fingerprint+ Signature
Fingerprint +Face
Face +Signature
Fc 504+504=1008
94.52
94.36
93.14
Fingerprint+ Signature
Fingerprint+ Face
Face+ Signature
Fa 504
93.52
95.26
92.24
Fingerprint+ Signature
Fingerprint+ Face
Face+ Signature
Fpca 480
93.82
96.64
96.16
Fingerprint+ Signature
Fingerprint+ Face
Face+ Signature
Fp 484 94.08
94.86
93.04
Fingerprint+
Signature
Fingerprint+ Face
Face+ Signature
Fm 120 89.05
90.05
88.26
Fingerprint+ Signature
Fingerprint+ Face
Face+ Signature
Fr 504 89.65
90.45
89.26
TABLE 2: Performance of Identification based on Curvelet Feature Vectors by Combining Two Traits in
ECMSRIT Multimodal Database.
12. Maya V. Karki & Dr. S. Sethu Selvi
International Journal of Biometrics and Bioinformatics (IJBB), Volume (7) : Issue (1) : 2013 69
Multimodal Database Feature
Vector
Feature Dimension GAR(%)
ECMSRIT
Chimeric Database-I
Chimeric Database-II
Fc 504+504+504=1512 96.82
96.04
96.84
ECMSRIT
Chimeric Database-I
Chimeric Database-II
Fa 504 96.92
95.22
95.02
ECMSRIT
Chimeric Database-I
Chimeric Database-II
Fpca 97.15
96.32
96.14
ECMSRIT
Chimeric Database-I
Chimeric Database-II
Fp 96.08
94.93
94.82
ECMSRIT
Chimeric Database-I
Chimeric Database-II
Fm 92.67
92.35
92.86
ECMSRIT
Chimeric Database-I
Chimeric Database-II
Fr 92.35
91.25
91.89
TABLE 3: Performance of Identification based on Curvelet Feature Vectors by Combining Three Traits.
Table 2 shows the performance of the system at feature level fusion considering two traits at a
time. Results show that GAR obtained from fingerprint and face is more compared to other two
combinations and maximum GAR of 96.64% is obtained for Fpca features. Table 3 shows the
performance at feature level fusion from all three biometric traits and a maximum GAR of 97.15%
is obtained for Fpca features. Results show that GAR is better in ECMSRIT database compared to
Chimeric databases. This is because in Chimeric databases the three biometric traits are not from
the same person. The performance of the proposed algorithm is better for correlated traits
compared to non-correlated traits. The test samples are rotated in steps of 2
o
to verify rotation
invariance for Fc, Fpca and Fa. Feature sets as these feature sets give better GAR compared to
other three feature sets. The results show that the GAR for rotated samples also remains almost
same and confirms that Curvelet transform is rotation invariant.
13. Maya V. Karki & Dr. S. Sethu Selvi
International Journal of Biometrics and Bioinformatics (IJBB), Volume (7) : Issue (1) : 2013 70
Feature
Vector
Feature
Dimension
Rotation in Degrees
0
o
2
o
4
o
6
o
8
o
10
o
GAR (%) for ECMSRIT Multimodal Database
Fc 1512 96.82 96.45 95.20 95.04 94.86 94.32
Fa 504 95.92 95.22 94.74 94.22 93.75 93.12
Fpca 480 96.78 96.22 95.74 94.22 93.75 93.12
GAR (%) for Chimeric Database-I
Fc 1512 96.04 96.04 95.86 94.75 93.25 92.89
Fa 504 95.22 95.22 94.74 94.22 93.75 93.12
Fpca 480 96.32 95.02 94.78 93.21 92.56 92.12
GAR (%) for Chimeric Database-I I
Fc 1512 96.84 95.05 94.54 93.84 92.14 91.85
Fa 504 95.02 94.89 94.74 94.22 93.75 93.12
Fpca 480 96.14 95.42 94.02 93.65 92.28 91.64
TABLE 4: Recognition Results by Applying Rotation to Test Samples for three Feature Sets.
Figures 9 shows the comparison between GAR obtained from unimodal identification system with
subband coefficient features and multimodal identification system with Fpca features. Fpca gives
maximum GAR and its dimension remains same as that of unimodal traits. When fingerprint is
combined with face and signature an improvement in GAR of 9.09% and when signature is
combined with face and fingerprint improvement has been 12.67%. Chart also show that
improvement in GAR is less when performance of two and three traits are compared.
FIGURE 10: Comparison between GAR Obtained from Unimodal and Feature Level Identification for
Curvelet Features.
6 COMPARISONS WITH SIMILAR WORK
Currently, as there are few multimodal system combining fingerprint, signature and face with
curvelet features at feature level unimodal recognition system is considered for comparison.
Table 5 gives comparison of fingerprint, face and signature recognition system based on
Curvelet features. In [21] author applied Curvelet transform on a fingerprint of size 64 x 64 which
was divided into four blocks. Each block is divided into 8 angular directions and standard
deviation from each direction was concatenated to form feature vector. Proposed Curvelet
algorithm with moment feature of dimension 40 is used for comparison and GAR obtained for 15
users is higher. In [22], author applied sixth level Curvelt transform decomposition and used 160
subband coefficients as features. Performance is evaluated using SVM classifier. The proposed
algorithm with 120 subband coefficients as features has similar performance. In [23], author
applied Curvelt transform on face at different scales subband coefficients are used as features.
Few dimension reduction techniques are applied and performance is evaluated. GAR obtained
14. Maya V. Karki & Dr. S. Sethu Selvi
International Journal of Biometrics and Bioinformatics (IJBB), Volume (7) : Issue (1) : 2013 71
with the subband feature dimension of 1258 is similar to the GAR obtained from proposed
algorithm with the feature dimension of 504.
Unimodal
Recognition
System
Author Database
used
No. of
users
Feature
Dimension
Classifier Performance
Fingerprint
Verification
A.Mujumadar[21]
Proposed
Algorithm
FVC2004-
DB1
FVC2004-
DB3
15
15
32
40
Fuzzy-
KNN
Euclidean
GAR=91.7%
GAR=95.02%
Signature
Verification
M.Fakhlai [22]
Proposed
Algorithm
Own
CEDAR
39
55
160
120
SVM
Euclidean
GAR=89.87%
GAR=88.87%
Face
Recognition Tanaya G. [23]
Proposed
Algorithm
ORL
ORL
40
40
1258
504
Euclidean
SVM
GAR=94.54%
GAR=95.04%
Table 5: Comparison of Unimodal Recognition System based on Curvelet Features.
7. CONCLUSION
The proposed multimodal system comprises of fingerprint, off-line signature and face traits.
Performance of the system is evaluated based on Curvelet transform features with SVM
classifier. The algorithm is tested by combining two traits at a time and three traits together. The
increase in dimension at feature level fusion is reduced by using template averaging, PCA and
statistical moment features. Six different feature vectors from these algorithms have been tested
at feature level fusion. Fpca and Fa features are obtained from concatenated feature vector while
Fp, Fm and Fr feature vectors are obtained without any fusion. The dimension of reduced feature
vectors are comparable with those of unimodal traits. The feature dimensions of unimodal traits
are decided based on d’ values. Fusion algorithm is tested on in-house created ECMSRIT
multimodal and Chimeric databases. Though the size of databases are small, the performance
obtained from these databases are low compared to that of ECMSRIT database. Results indicate
that the proposed algorithm performs better on correlated data than uncorrelated data. From
simulation results it can be summarized that when three traits are combined performance of the
system increases compared to either unimodal system or by combining two traits.
8. ACKNOWLEDGEMENTS
This work was supported in part by Research Grants from AICTE, New Delhi under Research
Promotion Scheme, grant no. 8020/RID/BOR/RPS-42/2005-2006.
9. REFERENCES
[1] Anil K. Jain and Arun Ross. Multimodal Biometrics an Overview. Communications of ACM,
pages 1221{1224, September 2004.
[2] K. Nanadakumar. Multibiometric System: Fusion Strategies and Tem plate Security. PhD
thesis, MSU, 2008.
15. Maya V. Karki & Dr. S. Sethu Selvi
International Journal of Biometrics and Bioinformatics (IJBB), Volume (7) : Issue (1) : 2013 72
[3] Arun Ross, Nandakumar, and A.K.Jain. Handbook of Multibiometrics. Springer Verilag New
York, 1 edition, 2004.
[4] R. Snelick, U. Uludag, A. Mink, M. Indovina, and A. K. Jain. Large Scale Evaluation of
Multimodal Biometric Authentication using State of the art Systems. IEEE Transactions On
Pattern
[5] Phalguni Gupta, Ajita Rattani, aHunny Mehrotra, and Anil Kumar Kaushik. Multimodal
Biometrics System for Efficient Human Recognition. In Proceedings of SPIE, 2006.Analysis and
Machine Intelligence, 27(3):450{455, 2005.
[6] Girija Chetty and Wagner M. Investigating Feature Level Fusion for Checking Liveness in
Face-Voice Authentication. In Proceedings of the Eighth International Symposium on Signal
Processing and Applications, ISSPA-2005, pages 66{69, August 2005.
[7] K. Nandakumar and A. K. Jain. Multibiometric Template Security using Fuzzy Vault. In
Proceedings of IEEE Second International Conference on Biometrics: Theory, Applications and
Systems, September 2008.
[8] Ferrer, Miguel A.and Travieso, Carlos M.and Alonso, and Jesus B. Multimodal Biometric
System based on Hand geometry and Palmprint Texture. In Proceedings 40th Annual IEEE
International Carnahan Confrence on Security, pages 92{97,2006.
[9] M. Husken, S M. Brauckmann, K. Okada Gehlen, and C. V. Malsburg. Evaluation of Implicit
3D Modeling for Pose-invariant Face Recognition, 2004.
[10] D..Donoho and M.R.Duncun. Digital Curvelet Transform: Strategy, Implementation and
Experiments. Technical report, Stanford University, 1999.
[11]Guillaume, Joutel, Eglin, and Bres Emptoz. Curvelet based Feature Extraction of Handwritten
Shapes for Ancient Manuscripts Classi_cation. In Procedings Of SPIE Electronic Imaging, volume
6500, pages 1-12, 2007.
[12]Jianwei Ma and Gerlind Plonka. A Review of Curvelets and Recent Applications. IEEE Signal
Processing Magazine, 27(2):118{133, 2010.
[13]Miroslaw Miciak. Radon Transformation and Principal Component Analysis, Method Applied
in Postal Address Recognition Task. International Journal of Computer Science and Applications,
7(3):33{34, 2010.
[14] Cheng Lu and Liu. Multimodal Biometrics Recognition by Dimensionality Reduction
Methods. In Second International Symposium on Electronics Commerce and Security, pages
113{116, 2009.
[15] M. Turk and A. Penteland. Eigenfaces for Recognition. Cognitive Neuroscience, 3(1):71{86,
1991.
[16] G.C.Feng, P.C.Yuen, and D.Q.Dai. Human Face Recognition using Wavelet Subband.
Journal of Electronic Imaging, 2(2):226{233, 2000.
[18] F.Murtagh and J.L.Starck. Wavelet and Curvelet Moments for Image Classification:
Application toAggregate Mixture Grading. Journal on Computer Vision and Pattern Recognition,
29:1557{1564, 2008.
[19] V. Vapnik. The Nature of Statistical Learning Theory. Springer-Verlag, New York, NY, 1995.
16. Maya V. Karki & Dr. S. Sethu Selvi
International Journal of Biometrics and Bioinformatics (IJBB), Volume (7) : Issue (1) : 2013 73
[20] Chih-Chung Chang and Chih-Jen Lin. LIBSVM: A library for support vector machines. ACM
Transactions on Intelligent Systems and Technology, vol 2 Issue 3, article no. 7:27, 2011.
Software available at http://www.csie.ntu.edu.tw/~cjlin/libsvm.
[21] A.Mujumadar and R.K.Ward. Fingerprint Recognition with Curvelet Features and Fuzzy KNN
Classi_er. In Proceedings of Signal and Image Processing, 2008.
[22] M.Fakhlai and H.Pourreza. O_-line Signature Recognition based on Wavelet, Curvelet and
Contourlet transforms. In International Conference on Document Analysis and Recognition
(ICDAR), pages 734{738, September 2007.
[23] Tanaya Guha, Q.M. Jonathan Wub and Yuan Yuan” Curvelet based face recognition via
dimension reduction” Signal Processing Volume 89, Issue 12, page no. 2345-2353