Comparison of Automatic Blood Vessel Segmentation Methods in Retinal Images

Blood vessel segmentation can be used for automatic retinal disease screening system. In this paper, comparison of automatic blood vessel segmentation methods in retinal images is presented and discussed. Morphological operation for the automatic segmentation of blood vessels in retinal images has been proposed and the results are compared with matched filter technique using local entropy thresholding. Blood vessels in retinal images are enhanced by the application of matched filter. The gray levels are spatially distributed by using local entropy based thresholding. Label filtering is performed by connected pixel labeling to eliminate the misclassified and isolated pixels. The blood vessels are segmented by using morphological opening based on structuring element. These methods were evaluated on the publicly available DRIVE database and the database contains retinal images along with the ground truth data that has been precisely marked by the experts. Morphological operation is more suitable for blood vessel segmentation as compared to the local entropy thresholding.