Irregular pupil localization using connected component analysis

Abstract

Irregular pupil boundary localization is proposed in this paper by designing a new approach that extracts the irregular pupil boundary points from the edge map using 4-neighborhood connected component analysis. Initially, Canny's edge detection is adopted to extract the edge map in the eye image. Then the proposed boundary extraction technique is employed to extract the pupil boundary points perfectly and it is independent of the specific eye image characteristics such as pupil deformation, poor contrast, poor brightness, etc., The proposed approach is adaptable to all monochrome databases and exhibits encouraging results when compared with the existing techniques.