Segmentation of Retinal Image Vessels with a Novel Automated Approach

Retinal image vessels can reflect different states of many diseases of the people, which is also the pre-characteristic for the registration and mosaic of retinal images. Therefore, it is very important to accurately extract blood vessels from the retinal images. In this paper, a novel automated approach for segmentation of retinal images is presented, which involves two steps. In the first step, after analyzing the feature of retinal images, gray mathematical morphology theories are employed to smooth and strengthen images in order to remove the background and enhance the brightness of retinal blood vessels. In the second step, maximal entropic thresholding algorithm is introduced to extract retinal blood vessels. The results obtained from the proposed method are compared to those of the standard manual segmentation. Experimental results indicate our approach is effective and can get higher segmentation accuracy and lower misclassify ratio