Morphologic Adaptive Fuzzy Threshold Based Retinal Segmentation for Automatic Eye Screening
Strength in similarity is considered to be a major task in terms of image segmentation particularly in segmenting retinal images. Rather than focussing on a solitary algorithm for the identification and extraction of multiple anatomical structures is treated as a challenging task during segmentation of retinal image because of the drastic variations their anatomical as well as non-anatomical retinal structure. An innovative, strong and stand-alone algorithm formed by the hybrid integration of adaptive fuzzy thresholding along with mathematical morphology is utilised for the extraction of multiple objects of anatomical nature consisting of extensive structures and features found in retinal images is proposed. For the purposes of solid analysis and overall clinical insight, the proposed work detects such anatomical structure in a single phase with an added advantage of high accuracy even in the case of compulsive images. Extraction of optic disk, exudate lesions as well as retinal vessels is carried out without analysing the texture. Moreover, the algorithm works better for the extraction of optic disc and vessels from compulsive images.
Multiple anatomical structures, Adaptive fuzzy thresholding, Mathematical morphology, Retinal vessels, Optic disk, Exudate lesions.