A Dynamic Circular Hough Transform Based Iris Segmentation

Iris segmentation in the case of eye diseases is a very challenging task in the computer vision field. The current study introduces a novel iris segmentation method based on an adaptive illumination correction and a modified circular Hough transform algorithm. Some morphological operations are used to create the illumination correction function, while the pupil localization process is used to detect the Hough circle radius range to obtain the iris circle. Warsaw BioBase-Disease-Iris dataset V1.0 contains 684 images is used to verify the proposed methodology. The proposed approach results show that the true segmentation rate is 90.5%, and the main problem of iris segmentation is the absence of complete or part of the iris in some diseases like blindness, rubeosis, synechiae and retinal detachment.