Gaussian Intensity Distribution Modelling of Blood Vessels in Fundus Images

Detection and extraction of images of blood vessels is an important problem in Fundus image processing. During the last two decades, different algorithms were proposed and these are evaluated using subjective evaluation criterion due to non-availability of any prior information on the blood vessels. In this paper, we propose a novel Gaussian intensity distribution (GID) model for synthesis of blood vessels in a Fundus image. The gray-level profile of the cross section of a blood vessel is approximated by a Gaussian function. Blood vessels are synthesized by concatenating piecewise linear segments. Three parameters, length, width and angle, are used to design each segment. This model is used for optimal design of matched filter for detection of blood vessels. Matched filters of fixed parameters are popularly used widely in the literature. The performance of the matched filter is evaluated with the computer simulated images in different environments, by varying the various parameters of the simulated blood vessels.