Performance of Image Enhancement Methods for Diabetic Retinopathy based on Retinal Fundus Image

In Malaysia, Non-mydriatic fundus camera become a primary tool for Diabetic Retinopathy screening protocols due to user friendly and cost effective procedure. However, the quality of fundus image produces often suffer from uneven illumination, color distortion, blur, and low contrast. Therefore, the need for image enhancement become crucial to be implemented as a pre-processing technique in image processing funnel. This paper presents six general basic methods that commonly applied for image enhancement which includes histogram equalization, contrast stretching, image negative, brightness enhances, low light image and gray level slicing. The performance evaluation of each method compared based on human interpretations and quantitative measurement using MSE, PSNR and entropy. Retinal fundus images collected from Ophthalmology Clinic, Hospital Universiti Sains Malaysia were used as the input images. Quantitative and qualitative result shows that CS method become the preferred method to be used for image enhancement of retinal fundus image in Diabetic Retinopathy.