Comprehensive System for Detection of Pterygium from Healthy Eye

Unwanted external growth which carries toward the cornea of the eye is the reason for pterygium. It is stimulating for medical practitioner to identify and assesse the encroachment of the retina by the thick flesh in the early stages. This paper elucidates the improvement of ample system, which can ascertain the eye illness for a given input image. This ample system considered the external eye image and check weather eye is healthy or any flesh growth towards the cornea of the eye i.e pterygium. The images are preprocessed for the red channel in RGB image and tried nonlinear filters, the decision based median filter DBMF is found to be better compared to other filters with high PSNR and high entropy. Then preprocessed image is offered for feature mining using Histogram oriented gradient HOG method. The different gradient features are used to identify and classify the images with pterygium growth and the healthy eye using the dichotomous classifier Support vector machine SVM. The developed model performed well in classifying the healthy images and pterygium images with an accuracy of 97.24%.