Optical Disc Location Based on Similarity to Improved Harris Algorithm

Optic disc location is the basis of fundus image analysis and computer-aided diagnosis of retinopathy, which provides basic data for subsequent operations such as optic disc segmentation. In this paper, an improved Harris corner location algorithm is proposed. According to the characteristics of dense vessels and obvious gray changes in the retinal fundus image, the area with the most corners is the optic disc area. The main idea is to extract the region of interest by matching filter, main vessel extraction, image enhancement and other methods. The corner of the region of interest is extracted by Harris corner detection algorithm improved based on similarity. First, the corner is preliminarily screened to reduce the number of edge corners and reduce the operation time. Finally, the moving window is used to locate the disc. Some fundus medical images are selected for experiments, and the results show that the segmentation result of the algorithm is accurate and the segmentation effect is good. The design of GUI interface makes the positioning process and result more intuitive.