The Pupil Location Based on the OTSU Method and Hough Transform

摘要

In this paper we present a pupil location method based on the OTSU method and Hough transform. Firstly, it finds a pint in the facula by gray summing operator for ensuring the probable area to find the center of the pupil, then it chooses one channel from R,G and B these three channels by distribution of histograms of three channels respectively for reducing the computing complexity. Thirdly it ensures an automatic threshold value by OTSU method and gets binary image. Finally in the binary image it uses Hough transform in certain area to fit the pupil edge. 384 eye images comprising an eye image database are located, our algorithm reaches 100% at accuracy. What is more, it ensures the robustness of location.