Colour Image Segmentation of Fundus Blood Vessels for the Detection of Hypertensive Retinopathy

Abstract

Retina is the major part of the human eye. Most of the diabetic and hypertensive problems will initially affect the retinal blood vessels of the eye. Using normal segmentation techniques, we can identify only the blood vessels. Therefore, color image segmentation is the best way to identify the retinal problems, because using color image segmentation we can differentiate the retinal veins and arteries. We have used masking techniques to segment the images of arteries and vein. The proposed technique is evaluated with data from ten patients diagnosed with hypertensive retinopathy. For this process, histogram equalization is performed on the complemented green plane image. From the result, the optic region is removed by morphological operation, after which it is converted into binary image. This output is multiplied with R, G, & B channels of input data and combined to get required segmented region of retinal blood vessels. All this vessels extraction has been done by using image processing techniques.