Fast Localization Algorithm of Eye Centers Based on Improved Hough Transform

Aiming at the problem of localization of eye centers in complex scenes, a method for quickly locating eye center is proposed in this paper. For the collected face images, this paper firstly uses bilateral filtering algorithm to remove the possible noise, and performs histogram equalization operation on the gray image to increase the dynamic range of the image grayscale and improve its distinguishability. Then, constructing cascaded strong classifier based on improved Ada Boost algorithm, and proposed three-layer eye detection. Finally, the method of canny operator edge detection and improved Hough circle detection is used to obtain the pupil center. The experimental results show that the algorithm can acquire the coordinates of the eye center quickly and accurately, and it is robust to eye location under illumination changes.