Touchless Authentication System Using Visual Fingertip Trajectories

Abstract

Biometric traces which are left on the screen of a security device or a mobile phone may cause a threat for the privacy of people when such personal data is likely to be stolen. In order to overcome this problem, a touchless authentication system is proposed which allows the usage of fingertip trajectories for creating or entering personal information such as password. First, fingertip regions are detected with Haarcascade method and these regions are tracked using Lucas-Kanade algorithm. The accurate detection of fingertip with invariance to pose, lightning and brightness is essential and crucial for security. Experimental results show that the proposed system proved its efficiency in various backgrounds where the finger is oriented in various directions. Having low computational effort, this system is suitable for low cost and real time processing.