Significant Enhancement of Segmentation Efficiency of Retinal Images Using Texture-Based Gabor Filter Approach Followed by Optimization Algorithm

Considering Retinal image as textured image, its texture based segmentation is required to identify the presence of retinal diseases. This pre-processing is important in automatic detection system for recognizing the abnormality present in the retinal images. Likewise, the proposed system mainly focused on diabetic retinopathy disease caused into eye –retina, generally leads to eye-blindness. Inspired from robust human's texture based segmentation capability, a mathematical model of the eye was formulated. A texture based Gabor filter was applied to get the output feature helping in detecting the abnormality and deriving statistical properties, further used in segmentation and classification. This work deals with the better separation of various clusters of Gabor filter output features, in order to get better segmentation efficiency. This was also followed by formalizing an objective function to tune filter parameters with Gradient descent and further Genetic Algorithm. This paper showed both qualitative and quantitative segmentation results with improved efficiency.