Eye Tracking Based on the Template Matching and the Pyramidal Lucas-Kanade Algorithm

Eyes are important facial organs and play an important role in face recognition, human-computer interactions and other areas. In order to locate and track eyes accurately, in this paper, we used the Adaboost algorithm to detect face, then located the eyes in the first frame using the Template Matching algorithm, finally combined the Pyramid algorithm and Lucas-Kanade optical flow algorithm to track eyes. At the same time, by using the OTSU algorithm and integral projection, we got the state of the eyes. The experimental results show that the algorithm has certain advantages in speed and accuracy relative to other algorithms.