Morphology Approach for Segmentation of Blood Vessels in Retinal Images

Segmentation is a challenging and important part of image analysis. Various kinds of eye diseases are identified through the retinal blood vessels, so retinal image segmentation plays an important role in medical image analysis. Even though it's been a lot of algorithms for extraction and retinal vessel processing, there are still obstacles to producing the best output. Some of the obstacles include poor contrast and retinal images that lack details, especially on the thin blood vessels in the retina. This paper proposes a morphological approach for retinal blood vessel segmentation using edge detection first, we apply contrast enhancement to the retinal image to get better segmentation results, and then we apply top hat morphology using multi-structure elements to detect blood vessel edges using a 7×7 pixel structure element with direction intervals of 15°. finally, eliminate false edges using the aperture operator. Based on the results of tests carried out on the DRIVE dataset, the proposed model provides an average accuracy of 96.25%.