GA based Feature Selection for Multimodal Biometric Authentication

Abstract

Multi-modal biometric authentication effectively replaces uni-modal biometric authentication system towards addressing a wide range of technical glitches in identity management and authentication. Legitimacy is playing a vital role in banking, military, and healthcare sectors where highly secure, strategic and confidential data transmission is involved. By integrating many independent biometric systems, one can overcome the problems of spoofing. However, there is lack of a simple, efficient and sufficient biometric authentication. Hence, the present study focuses on designing and implementing a multi-modal biometric authentication using a Genetic Algorithm (GA) based feature extraction method. The proposed research focuses on extracting human Skeleton and Human face feature using 3D Imaging technology. This modelling technique is used to capture human joints including the depth data to improve the efficiency of the system. The proposed research is subdivided into three phases. These are, image preprocessing (MinMax method), feature extraction using Heuristic Optimization Techniques (HOT), and Personnel recognition via the Artificial Neural Network (ANN). The Performance of the proposed method is evaluated based on the measure of FAR, FRR and accuracy. Finally, the performance of proposed approach is compared with existing techniques like GA, Neural network, etc. Combined Biometric is done in an unobtrusive way whereas other human recognition needs physical contact.