Hybrid method of iris detection based on face localization

Abstract

This paper presents a hybrid method of iris detection system based on edge detection and Hough Transform, with the help of skin segmentation in face detection algorithm, Golden Face Ratio and geometric definition. The algorithm starts with skin detection using Gaussian mixture model to find the bounding area of the face. Next, the facial area is used to mask the area for eye segmentation process. After the potential eye regions have been obtained, the centroids of the potential eyes and the edge detection process is performed. Then, using the obtained information, the coordinates of the irises are determined through the circular Hough Transform formula. For verification, several geometrical tests are performed onto the centroids to find out the best pair to be the irises of the eyes. The output image is displayed with the face bounded and its irises marked.