A Modified Finger Vein Identification Approach based on Niching Genetic Algorithm

Abstract

Finger vein engineering is the most modern biometric software using vein recognition patterns. Since these patterns are veiled under the skin surface, they offer great privateness, and therefore provide tremendous protection and are therefore extremely difficult to shape. The detection of the finger vein has received more consideration because previous methods have Highly vulnerable such as imbalanced finger vein dataset, and the extraction of salient feature within the low-quality images. Such defect has led the optimization algorithm not to be converged or its output to be reduced due to the limitations of the static means of finger vein identification, the need for intelligent approaches to finger vein identification is thus imperative. One of such intelligent approaches is machine learning which can be seen as the acquisition of structure description from examples. The kind of descriptions found can be used for identification, explanation, and understanding. The main contribution of the work presented in this thesis is to investigate the effect of genetic algorithm in selecting optimal finger vein features vector by adding niching concept in form of Context Based Clearing (CBC) procedure to increase the heterogeneity of items within the feature vector with the aim of removing the correlation between items of features vector. The enhanced feature selection approach produces an optimal vector that able to deal efficiently with the massive intra-class variations and the diminutive inter-class similarity. Besides, it yields the concept of feature set reduction to remove redundancy without degradation of the accuracy. The performance analysis of suggested model is carried through several experiments and the results show an improvement on an average by 6% in terms of accuracy compared with some Up-to-date distinguishing finger vein mechanisms available in the literature.