Vessel extraction in retinal images using morphological filters

This work presents a simple method for blood vessel detection from color image of a retina. After RGB to grayscale conversion and contrast enhancement, morphological operations are applied to remove the optic disc from the image. Then the largest connected component of the binary image is identified to detect the blood vessels. Extracted vessel image is denoised by median filtering. The method is simple, easy to implement and produces good results. It is effective and robust with different image conditions. The performance of the proposed method is evaluated on Drive database. A comparative study on performance measure is also presented.