Diabetic Retinopathy Recognition using Enhanced Crow Search with Levy Flight Algorithm

This paper aims to introduce an improved model for Diabetic Recognition (DR) recognition. Accordingly, the proposed model is executed under two stages, the initial one is the blood vessel segmentation and next step is the DR recognition. Using tophat by reconstruction of red portions in the green plane image, the two thresholds binary images are obtained in vessel segmentation. The areas that are found similar to two binary images are extracted as the main vessels. Additionally, the residual pixels in both the binary images are integrated in order to form a vessel sub-image i.e. facilitated to a classification of Gaussian Mixture Model (GMM). As a result, the complete pixels in the sub-image that are classified as vessels are amalgamated with the main vessels to obtain the segmented vasculature. Moreover, from the segmented blood vessel, the extraction of GLRM and Gray-Level Co-Occurrence Matrix (GLCM) features is performed that are subsequently classified by exploiting Neural Network. To enhance the accurateness, training is performed using Enhanced Crow Search with Levy Flight (ECS-LF) algorithm, so the error among actual output and predicted must be least.