Iris detection using hybrid models

Face and iris detection are great of interests to researchers in the field of video processing and computer vision. In recent years, great advances in robotics and artificial intelligence have further increased the importance of face recognition and iris detection. In this study, iris of the human eye is detected through hybrid models. These models first detect the face, then the eyes, and finally the iris. Two classifiers, the Artificial Neural Network (ANN) and the Decision Tree (DT), are trained to distinguish face images from other. They are also used to extract face images using the sliding window technique. After extracting the face from the given image, the same steps are applied to detect the eyes from the face. It has been observed that the ANN performs better than the DT with respect to detection rate of faces and eyes. The eyes that are finally detected are used to find circles that indicate the iris. Two methods are discussed to detect the iris: Hough Transform (HT) and the mean of gradients (MoG). In comparison with HT, MoG yields higher detection rate.