Feature extraction and classification by machine learning methods for biometric recognition of face and iris

Abstract

Biometric recognition became an integral part of our living. This paper deals with machine learning methods for recognition of humans based on face and iris biometrics. The main intention of machine learning area is to reach a state when machines (computers) are able to respond without humans explicitly programming them. This area is closely related to artificial intelligence, knowledge discovery, data mining and neurocomputing. We present relevant machine learning methods with main focus on neural networks. Some aspects of theory of neural networks are addressed such as visualization of processes in neural networks, internal representations of input data as a basis for new feature extraction methods and their applications to image compression and classification. Machine learning methods can be efficiently used for feature extraction and classification and therefore are directly applicable to biometric systems. Biometrics deals with the recognition of people based on physiological and behavioral characteristics. Biometric recognition uses automated methods for recognition and this is why it is closely related to machine learning. Face recognition is discussed in this presentation — it covers the aspects of face detection, detection of facial features, classification in face recognition systems, state-of-the-art in biometric face recognition, face recognition in controlled and uncontrolled conditions and single-sample problem in face recognition. Iris recognition is analyzed from the point of view of state-of-the art in iris recognition, 2D Gabor wavelets, use of convolution kernels and possibilities for the design of new kernels. Software and hardware implementations of face and iris recognition systems are discussed and an implementation of a multimodal interface (face and iris part of a system) is presented. Also a contribution of Machine Learning Group working at FEI SUT Bratislava (http://www.uim.elf.stuba.sk/kaivt/MLgroup) to this research area is shown.