Blood vessel segmentation from color retinal images using K-means clustering and 2D gabor wavelet

Abstract

This paper presents a new unsupervised method for segmenting blood vessels in digital retinal images. The proposed method uses K-means clustering to binarize grayscale vessel-enhanced images derived from green channel image and Gabor wavelet feature image. The binary images are then combined using logical OR to produce segmented vessels. The method was evaluated on the publicly available DRIVE database and the results compared to published literature. The method proved to have comparable performance to other published unsupervised methods while being simple and fast to implement. In the future, the proposed method can be further improved to be applied in real clinical setting to assist the physicians in diagnosing ocular diseases through an automated screening system.