Segmentation of Retinal Blood Vessel Using an Algorithm-Based Gamma Distribution of Matched Filter

Abstract

Retinal image contains information about the vascular structures of the retina to predict the diseases such as diabetes, hypertension, obesity and glaucoma. However, the retinal image information is also used for diagnose the diseases. The retinal image contains the high resolution of blood vessel and close to the background of the retinal image. However, the detection of retinal vessel is very difficult and challenging problem. Therefore, we proposed Gamma probability distribution function based Matched filtered approach for retinal blood vessel segmentation having less computation time. The proposed algorithm is divided into three-stage which is divided into three stages: preprocessing, Gamma-matched filter and post-processing. Each stage computes the computation time to the process of each step of the algorithm. In the preprocessing step, conversion of retinal color images into grayscale image using PCA further enhances the grayscale image using the CLAHE method. Gamma-matched filter is used to generate the MFR images. In post-processing step, extraction of retinal blood vessel by using the entropy-based optimal thresholding technique is to generate the segmented image and compute the total time to process of segmentation of blood vessel in retinal images. The evolution of the proposed approach is tested in 20 normal images of DRIVE and STARE databases to measure in terms of true positive rate, false positive rate, average accuracy and computation time. The results obtained are 95.47%, 39.11%, 91.04% and computation time 5 min, respectively.