Retinal Vessel Segmentation using Robinson Compass Mask and Fuzzy C-Means

Retinal vessel segmentation is significant for diagnosis of vascular diseases such as diabetes, hypertension etc. The paper proposed an unsupervised retinal segmentation approach based on Robinson Compass Mask and Fuzzy C Means(FCM). Vessel enhancement is an important pre-processing step in retinal vessels identification applications. In this paper, two different approaches are used in pre-processing for vessel enhancement (i) primarily inspired from the commonly used CLAHE filter and (ii) using combined response of Gaussian kernel and Gabor filter. We investigated both the approaches to pre-process the image and its effect on accuracy of vessel segmentation result. Enhanced vessels using the combination of various filters and further segmented with Robinson Compass Mask and Fuzzy C-means represent better segmentation result. The proposed approach is assessed using publically available DRIVE database. The experimental results illustrate that proposed approach using combined filtering, Robinson Compass Mask and FCM shows better results in comparison to state-of-the-art method.