Design and Implementation of Eye Pupil Movement Based PIN Authentication System

PINs (Personal Identification Systems) have been widely adopted worldwide as primary means of secure communication for user authentication and verification purposes. However, it’s not a foolproof system since it can be easily forged. Since PINs needs to be entered manually, it provides an easy opportunity for an intruder to crack it. Thus it is susceptible to various intrusions such as shoulder surfing, key logger, tap print etc. In this paper, an eye pupil movement based PIN generation system has been devised. At first, the user enters sensitive authentication input (PIN) by using eye pupil movements in various directions (i.e. Left, Middle and Right), which further is internally mapped into various pattern of digits from 0 to 9. Thus eavesdropping by a malicious observer becomes practically impossible. It utilizes Haar-Cascade classifier for face and eye detection followed by combined approach of HOG features integrated with SVM classifier for eye blink detection. For pupil detection, canny operator is employed followed by fitting a circle to pupil using circular Hough Transform. Tracking the position of eye pupil is achieved using projection function algorithm. The accuracy of eye detection, eye blink detection and eye tracking is 98%, 92.51 % and 96.25 % respectively. The contribution of this paper is outlined along with a comparative study between proposed approach and traditional authentication systems like gaze, gaze – touch, eye movement CAPTCHA and such graphical image based authentication methodologies. Our devised system is simple, user friendly and works under low light conditions without involving any significant dependencies on the intricacies of the system.