Ameliorated Anti-Spoofing Application for PCs with Users’ Liveness Detection Using Blink Count

Abstract

The paper proposes "Anti-spoofing application for desktop". This application uses a face recognition approach along with the use of eye-blink count to detect liveness. Main phases of application are namely, face detection and recognition, and determination of liveness status of user. Liveness detection is proven to prevent the video play-back attacks and use of printed photograph in order to compromise the security. Webcam captures the user’s image after every short interval of time. Image captured after passing authentication process is checked for liveness. In case of security breach, countermeasures are executed. This include capturing image of adversary and system logoff or exit. This paper proposes an additional functionality which uses HOG feature descriptor of user image along with passcode. It uses SVM classifier that gives performance metric of 100% accuracy. The experimental results of the ameliorated functionality show the effectiveness of the proposed approach.