A simple approach of blood vessels detection in retinal images using MATLAB

Abstract

This paper proposed a simple approach of blood vessels detection and extraction in retinal images using MATLAB software. There are 60 retinal image samples from STARE and DRIVE database. The implementation process begins with changing a colour retinal image to grey-scale image followed with enhancement techniques. These enhancement techniques involving opening by reconstruction and top-hat transform to enhance the performance of conventional blood vessels extraction method. The morphology process performed by opening by reconstruction and top-hat transform technique are significant for the next extraction process. The morphology process is followed by local entropy thresholding and filtering process. The extraction process of proposed method took less than 1 minute to extract the blood vessels with an average accuracy and sensitivity of 0.89 and 0.71.