Optic disc detection using template matching based on color plane histograms

Abstract

In this paper, a robust and efficient histogram based template matching method for automatic detection of Optic Disc (OD) in retinal images to help ophthalmologists for diagnosis of retinal diseases is presented. Based on camera field of view (FOV) and image resolution, the OD size estimation algorithm for creation of specific size templates is proposed. Color plane histograms of OD region were used for template creation instead of complete OD image. Mean of color plane histograms of four templates were used for template matching to detect OD in retinal images. The proposed method is evaluated on retinal images of publically available DRIVE and STARE datasets. The success rate of 100% and 97.5% for DRIVE and STARE dataset respectively is achieved with proposed method. The performance comparison of results with previous methods shows that the proposed method is robust and efficient for the localization of OD in the retinal images with and without pathology.