View-Based Biometrics

Abstract

Biometric recognition systems are considered to be one of the most secured means of authentication. In this context several biometrics have been proposed but the view based biometrics such as face, iris etc remain the most natural choice. In the paradigm of face recognition, it is generally assumed that major information contents lie in the lower frequency region of an image and therefore little effort has been made in sys tematic exploration of the detail images. Although some wrapper-based approaches have been proposed in the literature, they are primarily based on experimental eval uation of a specific classifier on various subbands. Therefore there is a dire need of a framework for automatic selection of the most significant subbands based on the underlying statistics of the data. In this thesis, the problem of identifying the most dis criminant subbands based on information theoretic measures is addressed. Essentially the face images are transformed into textures using the linear binary pattern (LBP) ap proach, these texturized-faces undergo the wavelet packet decomposition resulting in several subband images. We propose to use the energy features to effectively represent these subband images. The underlying statistical patterns of the data are harnessed in form of information-theoretic metrics to select the most discriminant subbands. The proposed algorithms are extensively evaluated on several standard databases and are shown to always pick the most significant subbands resulting in better performance. The proposed algorithms are entirely generic and do not depend on the validation re sults for specific classifiers. Noting that localized features are often more useful than theholisticapproaches, wehavealsotargetedtheproblemofirisrecognitionproposing the concept of class-specific dictionaries. Essentially, the query image is represented as a linear combination of training images from each class. The well-conditioned inverse problem is solved using least squares regression and the decision is ruled in favor of the class with the most precise estimation. An enhanced modular approach is further proposed to counter noise due to imperfect segmentation of the iris region. As such iris images are partitioned and individual decisions of all sectors are fused using an efficient fusion algorithm. The proposed algorithm is compared to the state-of-the-art Sparse Representation Classification (SRC) with Bayesian fusion for multiple sectors. The proposed approach has shown to comprehensively outperform the SRC algorithm on standard databases. Complexity analysis of the proposed algorithm shows decisive superiority of the proposed approach.