Pupil Location Method Based on Feature Searching and C-V Extension Model

Pupil images are usually disturbed by pupil deformation, eyebrows, eyelid and eyelash occlusions, uneven illumination and light spots, which results into inaccurate pupil location. For pupil images with interference, this paper proposes a pupil location method based on the feature searching and a C-V extension model with the constraint of standard deviation-mean and convexity. Firstly, a novel contrast stretching function is used in image preprocessing; Secondly, an objective function of feature searching is constructed to locate the pupil center and estimate the pupil area, positioning pupil preliminarily; Finally, through the C-V extension model with the constraint of standard deviation-mean and convexity, the initial contour curve extends outward and locates the pupil, positioning pupil accurately. Experimental results of processing collected images show that the method proposed in this paper can locate the pupil accurately and quickly.