Abstract
Conference Title: 2016 IEEE Canadian Conference on Electrical and Computer Engineering (CCECE) Conference Start Date: 2016, May 15 Conference End Date: 2016, May 18 Conference Location: Vancouver, BC, Canada Retinal blood vessels extraction is a primary step for detecting eye diseases including diabetic retinopathy which causes blindness. It also simplifies other image processing techniques such as classification. Since manual extraction is a long task and it requires training, many automated methods have been proposed. In this paper, an algorithm for extracting blood vessels from fundus images has been proposed. The algorithm is based on two dimensional Gabor filter, local entropy thresholding and alternative sequential filter. The proposed method has been tested on fundus images from Structured Analysis of the Retina (STARE) and Digital Retinal Images for Vessel Extraction (DRIVE) databases using MATLAB codes. The results show that this method is perfectly capable of extracting blood vessels.