Liveness Detection Based on Eye Flicker

Abstract

Many of the methods has been reported for detection of liveness based on eye flicker are intrusive, in our method detecting liveness using non intrusive. The proposed methodology uses eye flicker activity in an image sequences. The present generic camera will capture the face video with twenty five FPS, frame intervals difference is less than seventy milliseconds. Sequences of images are used in AdaBoost method and Viola Jones technique for the face identification. The Harr like components are figured by convolving the picture with layouts of various size and introduction. Next step is to detect eye region, the position of eye in a face is found on certain geometric dependencies known for human face. If iris is detected in eye region then eye is open. If not, it is closed. Here in this paper we are thinking about eye flicker and iris varieties as a confirmation of liveness to reject utilizing few fake. We tested around more than 20 natural videos to detect the liveness based on eye flicker and achieved 92.50% accuracy.