Performance evaluation of retinal vessel segmentation using a combination of filters

Image Segmentation is the procedure that breaks down any image into different parts. This is required in order to attain the essential characteristics from the image. Segmentation finds its applications in different fields of satellite imagery, weather forecasting, traffic controlling etc. But it is mainly used in the medical field to diagnose different kinds of diseases suffered by different people. Though manual segmentation provides with good results but automated analysis of retinal images is more promising as it reduces time and effort required by the experts. Diabetic Retinopathy is such a disorder of an eye which can be mostly seen in a person suffering from diabetes. If this disease is not recognized earlier the person may suffer from complete blindness. Although various techniques have been proposed to detect the blood vessels and diagnose the disorders but the main issue of noise has not been considered greatly. In this paper, a hybrid combination of two filters is proposed which provides us with better results over the existing methods even in the case when the colored retina image is more hampered by noise. The proposed method is tested on DRIVE database.