Double Authentication System Based on Face Identification and Lipreading

Abstract

With the amercing new technologies more advanced security systems are developed. Security system means difficult to be attacked by any intruder, complicated design, expensive and requires additional software and hardware resources. Biometric system comes over all the simple security system. Since the biological data is used for the security purpose and its cheap, simple design and much easy to handle. But now-a-days these systems are also hacked and more advanced, secured system is to be developed. So to overcome all these problems new biometric technique of lipreading is brought into implementation as authenticating system. In this paper, double biometric authentication system is developed. This system is based on face identification as “ID” and then after identification lipreading is used as “Password”. In the process of face identification the KAZE features are used which gave 90.1923% of accuracy using the KNN classifier. After the correctly identification of face, lipreading process starts. In the lipreading part, hybrid based technique feature are extracted. By making use of multi-class SVM, this gave 80% accuracy for lipreading in MATLAB 2018a software. This developed system can be used in various areas like banking, highly secured areas and also for the locker systems.