International Journal of Biometrics (10 papers in press)
An Approach for Facial Expression Classification
by Ali Muhamed Ali, Hanqi Zhuang, Ali Ibrahim
Abstract: Human facial expression classification has attracted great attention in the field of computer vision and pattern recognition over the past decades. This is partly due to peoples curiosity in exploring this fascinating research area and partly due to its potential applications. In this paper, a new method for facial expression classification is proposed. The method uses the Histograms of Oriented Gradients (HOG) algorithm to extract facial expression features and the Sparse Representation Classifier (SRC) to classify facial expressions with a large variation of poses. The Histograms of Oriented Gradients algorithm was selected due to its effectiveness in picking up both local and global facial expression features in different orientations and scales, and the Sparse Representation Classifier was chosen because of its proven effectiveness in face recognition. A novelty of the proposed approach is that given a facial image for classification, its pose is determined first in order to select a pose-dependent dictionary for the SRC procedure. The paper also discusses in detail how to select parameters to improve the effectiveness of the HOG algorithm. The proposed method was applied to two multi-pose facial expression databases: KDFE and RaFD, and satisfactory results were obtained for a majority of facial expressions under various poses.
Keywords: Facial expression; expression classification; Histograms of Oriented Gradients; emotion detection; Sparse Representation Classifier.
Fingerprint Representation and Matching for Secure Smartcard Authentication
by Ibrahim El-leithy, Gouda Salama, Tarek Mahmoud
Abstract: In this paper, a light weight fingerprint matching algorithm (4 KB) is proposed. This algorithm is based on matching features that are invariant to major transformations like translation and rotation. The algorithm can be executed on devices with low computing power and limited memory size. Thus, the matching algorithm is implemented on smartcard over the Java Card TM platform. The algorithm has an asymmetric terminating behavior. Therefore, the execution time varies depending on correct positive matches (similar fingerprint) and correct negative matches (dissimilar fingerprint). In this paper, two fingerprint authentication methods are implemented. The first fingerprint authentication method is applied using one reference and one candidate fingerprints. However, the second method is performed by the aid of a fusion based fingerprint authentication manner using two reference and one candidate fingerprints. The performance of the methods in terms of authentication accuracy is tested on some standard databases from the Fingerprint Verification Competition 2002 (FVC2002).
Keywords: Biometric Authentication ; Fingerprint Matching ; Fusion at the Decision Level ; Match on Card ; Smartcard.
Personal authentication based on finger knuckle print using quantum computing
by Ali Altaher, Saleem M. R. Taha
Abstract: The finger knuckle print (FKP) images are used for personal authentication. The proposed model consists of preprocessing of the FKP image, and then feature extraction algorithm is applied to extract coefficients that will be used in the matching process. In the classification process, improved versions of neural networks (Quantum Neural Network (QNN), Wavelet Neural Network (WNN) and Quantum Wavelet Neural Network (QWNN)) are used to approach better accuracy and speed of convergence. This paper has precedence in implementation of the Quantum Computing (QC) in the structure of the FKP recognition system. It has advantages of low inexactness and high speed of execution by using the quantum superposition state ideology. A database gathered from 165 volunteers by Hong Kong Polytechnic University (Poly U), and the proposed authentication model performance is tested upon it. Compared with other existing FKP recognition systems, the proposed one has merits of more secure as well as high accuracy and speed.
Keywords: Finger knuckle print (FKP); Personal authentication; Quantum computing; QNN; QWNN; WNN.
AN EFFICIENT AND REDUCED MEMORY INDEXING APPROACH BASED ON PRIORITY RANK SPECTRAL HASHING FOR MULTIBIOMETRIC DATABASE
by Revathi Balasundaram, Sudha Gnanou Florence
Abstract: Fast retrieval of data from a multibiometric data base is a challenging task as the size of the databases have increased considerably. For the retrieval of data to be faster, the search space of the database has to be narrowed down by forming a smaller set comprising of nearest neighbors of the query. In order to achieve this, an appropriate data structure is to be built in accordance with the nature of the dataset. In this work, a novel priority rank based Spectral Hashing algorithm is implemented to enhance the efficiency of indexing and retrieving from a multibiometric database comprising of iris and palmprints. To improve the matching accuracy, the biometric images are represented by GIST features and the features are fused together by weighted feature level fusion. The proposed Spectral Hashing with dynamic rank based priority queue indexing algorithm aims to overcome the limitations of the existing systems by enhancing the Hit rate with reduced storage and computational costs. From the experimental results, it is concluded that the proposed indexing algorithm has reduced storage cost by 85%, along with reduced Penetration rate, False Acceptance Rate and False Rejection Rate. In addition, Hit rate has improved by 25% compared to the existing kd tree technique.
Keywords: Indexing; Multibiometrics; GIST; spectral hashing.
Simplified and Efficient Face Recognition System on real image set and synthesized data
by Donato Barbuzzi, Angelo Galiano, Alessandro Massaro, Valeria Vitti, Leonardo Pellicani
Abstract: This paper presents experimental results related to a simplified and efficient face recognition system using a basic webcam. More specifically, the purpose of this work is twofold: (1) to detect constantly the individuals face in front of the computer, in an uncontrolled environment, and (2) to send an alert to the system manager if another individuals face is recognized.
Sixteen distances based on fourteen points of the face between eyes, nose and mouth are considered. Experimental results carried out on two different training sets are presented. The first database has been constructed on 500 real face images of 10 individuals (50 faces each one), while the second database has been created on the same 500 previous images and 500 new synthetic data obtained through a crossover operation.
For experimental evaluation, a 5-NN classifier is used. Finally, results show the performance in terms of a new correlation score for the recognition task.
Keywords: Face Recognition; Image Processing; Synthetic Data; k-NN; Correlation Score.
Incremental Robust Principal Component Analysis for Face Recognition Using Ridge Regression
by Haïfa Nakouri, Mohamed Limam
Abstract: Face recognition efficiency is extremely challenged by image corruption, noise, shadowing and variant face expressions. In this paper, we propose a reliable incremental face recognition algorithm to address this problem. The algorithm is robust to face image misalignment, occlusion, corruption and different style variations. To apply this for large face data streams, the proposed algorithm uses incremental robust principal component analysis (PCA) to regain the intrinsic data of a bunch of images regarding one subject. A novel similarity metric is defined for face recognition and classification. Five different databases and a base of four different criteria are used in the experiments to illustrate the reliability of the proposed method. Experiments point that it outperforms other existing incremental PCA approaches namely incremental singular value decomposition, add block singular value decomposition and candid covariance-free incremental PCA in terms of recognition rate under occlusions, facial expressions and image perspectives.
Keywords: Image alignment; Robust Principal Component Analysis; Incremental RPCA; Ridge regression.
An Integrated Framework for Evaluating the Performance of Age Progression Algorithms
by Andreas Lanitis, Νicolas Tsapatsoulis
Abstract: Facial age progression can play an important role in biometric authentication as it enables the long term person identification based on age progressed facial renderings. In this paper a systematic evaluation approach that can be used for assessing the performance of age progression algorithms is presented. The proposed method relies on the use of a dedicated dataset in conjunction with the development and use of machine-based performance evaluation metrics that allow the assessment of the intensity of aging effects added on a face along with an assessment of the identity preservation in age progressed images. The proposed performance evaluation framework can form the basis of implementing comprehensive comparative performance evaluation between different age progression methodologies, allowing in that way the evolution of the best algorithms.
Keywords: Age Progression; Facial Aging; Performance Evaluation.
An approach to matching fingerprints using cryptographic one-way hashes
by Qinghai Gao
Abstract: Password-based authentication systems match passwords in the formats of cryptographic one-way hashes without storing the original passwords. The method prevents someone with a copy of the password database from reversing the hashes and hacking into individual accounts easily. Biometric data cant be reproduced with 100% accuracy. Therefore, this technique has yet to be utilized directly to secure biometric data. Instead, biometric authentication system manages biometric data in two ways. The first is to store original biometric templates in encrypted formats. Real-time decryption is required upon authentication during which decrypted data can be stolen and performance could also be an issue for identification with a large database. The second is to store transformed templates obtained by transforming original templates with a similarity-preserving mechanism such that matching can be done in transformed domain. However, the requirement of similarity-preserving transformation makes the transformed templates vulnerable to reverse engineering. In this paper we propose a novel approach of transforming fingerprint templates which includes generation of random synthetic minutiae, projection of real minutiae to synthetic minutiae, and hashing of individual minutiae. The projection process can eliminate the intra-class variations of real minutiae. Matching is conducted between two templates only containing cryptographic one-way hashes. Inter-class variations are maintained with user-specific synthetic templates. This approach makes it possible to store and match fingerprint minutiae templates in the formats of cryptographic one-way hashes. Our testing results indicate its feasibility.
Keywords: Fingerprint; synthetic template; randomized minutiae tessellation; minutiae hashing; MD5 entry matching.
Immunological classifiers for accelerometer based gait identification
by Ismahene Dehache, Labiba Souici-Meslati
Abstract: Research in the field of biometrics is currently oriented towards behavioral modalities. The present study is interested in gait biometrics which occupies an important place due to its various advantages compared with other biometrics. This work suggests an identification system based on gait, using an accelerometer that allows the measurement of acceleration. The proposed approach for recognition is the immunological approach. Three types of classifiers; namely AIRS1, AIRS2 and AIRS parallel are suggested. An improvement on the three classifiers is done specially on the calculation of the affinity by integrating three types of distances: Hamming, Manhattan and Chebychev. The results are very satisfactory emphasizing the importance of gait modality and the interest of using immunological approaches in the domain of recognition of persons through behavioral biometrics.
Keywords: Gait identification; Artificial Immune Recognition System; biometrics; accelerometers; metrics.
Contact Lens Detection for Iris Spoofing Countermeasure
by Edward Tan, Anto Satriyo Nugroho, Maulahikmah Galinium
Abstract: The development of biometric authentication system should be followed by strengthening to spoofing attempts. Among various identifiers, iris has aroused many attentions due to its uniqueness and stability. Nevertheless, the use of iris for biometric authentication is accompanied by spoofing risk, for example using contact lens. In order to handle the spoofing attempts, its detection is inevitable part of a recognition system, to reduce the risk of forging system. Cosmetic contact lens is one of common spoofing material which is hard to be detected. In this study, weighted Local Binary Pattern (w-LBP) and simplified Scale Invariant Feature Transform (SIFT) descriptor were used to extract the feature of the iris, in which segmented using gradient magnitude and Fourier descriptor. Simplified SIFT descriptor is extracted at each pixel of iris image and being used to rank the Local Binary Pattern (LBP) sequence of encoding. The features were then presented to Support Vector Machine (SVM) classifier, for positive vs negative classification. Positive class means that contact lens was used by a person, and vice versa. The experimental results showed that combining SIFT and w-LBP as features for SVM yielded an accuracy of 84%.
Keywords: Biometrics; Iris Recognition; Contact Lens; Detection; Classification; Iris.