Blood vessels segmentation using thresholding approach for fundus image analysis

The principal target of blood vessels detection algorithm is to early detecting the diabetes in advanced stages by comparison of its states of retinal blood vessels. Due to relatively low contrast, potential presence of dark pathologies like microaneurysms and hemorrhages and variable size of blood vessels, segmentation of blood vessels is a nontrivial task. The dictum of work present in this paper is local thresholding approach, which can be used in computer based retinal image analysis to extract the retinal image vessels. To enhance the blood vessels and suppress the background information, we performed preprocessing operation on the retinal image. Then the enhanced image is segmented using local thresholding algorithm. The proposed approach is tested on the DRIVE dataset and is compared with alternative approaches. Experimental results obtained by the proposed approach show that it is effective when this obtained the average accuracy of 96.02% and best accuracy of 97.2%.