Performance Comparisons of Fuzzy Logic, Back Propagation Neural Network and Graylevel Co Occurrence Matrix Texture Properties in Identification of Exudates in Diabetic Retinopathy Images

This paper presents the implementation of back propagation algorithm (BPA), Fuzzy logic (FL) and graylevel co-occurrence matrix (GLCM) in identifying the exudates in diabetic retinopathy (DR) images. Human eyes are affected due to malnutrition and other present day exposure of eyes to different environments as continued work on the computer, watching television, watching small screen sized mobile phones. The eyes are strained in one form or other and damage to the nerves of the eyes occur which can be called DR, glaucoma and many other types. Representative features are obtained from the image. They are used for training the implemented algorithms. The performance of the three algorithms in identifying the exudates are presented.