Automated segmentation of blood vasculature from retinal images

In this paper an algorithm is proposed for blood vessel extraction from an eye's fundus image. Blood vessels removal and detection is an important step to find features or abnormalities like red lesions, optic nerve and fovea used for retinal health diagnosis. The proposed method uses a strategic combination of green and L channel to develop the final vessel structure which increases the accuracy. A combination of morphological operators and intensity based thresholding are used which creates a method which is computationally efficient and less complex. A set of public DRIVE data of fundus image of an eye is used to test the proposed algorithm. The results show a better comprehensive performance of vessel extraction and computationally efficient method.