Efficient multi feature selection and recognition methods for multi biometrics recognition system

Abstract

In modern era protecting data in a unique manner is an inevitable requirement. Biometric systems utilizing personal physiological or behavioural characteristics which become widely popular in providing security for information technology and entry to sensitive location like airports, governments, health care, military, and any other type of business that connected by a network. In order to enable a biometric system to operate effectively in different applications and environments, a multimodal biometric system which makes a personal identification based on multiple physiological or behavioural characteristics is preferred. Multi-modal biometric authentication system could be used to further strengthen the security by overcoming the limitations imposed by unimodal biometric authentication system. In this proposed work, in multimodal biometrics methods, particularly face, fingerprint and iris based algorithm is described. In this first work, introduced multi-feature based programmed face identification under unstable lighting conditions. In this proposed work, face image is taken as an input. To achieve good performance under illumination changes, methods based either on normalization or illumination have been proposed. Then the Gabor features, LBP features and phase congruency features are extracted from pre-processed face images. The extracted features are combined by using Z score level fusion and perform key generation. According to the fused keys the face images are recognized by using ANFIS classifier. The experimental results show that the proposed system achieves better performance compared with the existing system. However, it does not produce an efficient recognition results for colour face images.To solve this problem, the proposed system designed an efficient template matching model for color face recognition using Hidden Markov Model and Particle Swarm Optimization (HMM-PSO). In order to support color model RGB space of the input face image is converted into L*a*b Color Model. Then the colour face image is pre-processed by using Alpha-trimmed mean filter. After the completion of facial feature extraction, Singular Value Decomposition (SVD) method is used to select optimal features. After the completion of optimal feature selection, compute the score values and the score level fusion approach is utilized to fuse the scores. Finally the face based novel template key pattern matching is performed with the help of HMM-PSO. The experimental results show that the proposed system achieves better performance compared with the existing system in terms of accuracy, precision, recall and F-measure. However unimodal biometric systems have variety of problems such as intra-class variations, restricted degree of freedom, non-universality, spoof attacks, and unacceptable error rates. newline