DETECTION OF OPTIC DISC IN COLOR FUNDUS IMAGES USING GREEN CHANNEL COMPONENT

Abstract

Detection of optic disc in fundus images forms an important step in the automatic screening process for eye diseases. This paper presents a novel technique for tracing of optic disc boundary using green channel component in the color fundus images. The brightness, color, and contrast attributes of optic disc are similar to those of other components such as exudates and cotton wools which are also present in the fundus images. The identification of these components helps in detecting the abnormal signs of eye diseases such as diabetic retinopathy and glaucoma. The proposed algorithm employs average filtering, contrast stretching transformation, thresholding, morphological gradient and boundary tracing techniques to detect the optic disc in the retinal image. The preprocessing steps are used for enhancing the image to make it better fit for the subsequent analysis. The morphological gradient is useful for determining the edge points. The tracing phase then tracks the border pixels along the top of the ridges and returns the accurate row and column coordinates of all the pixels lying on the border of the optic disc. The proposed method is tested on images of DRIVE database with different brightness and contrast conditions using a fixed threshold value. The results obtained confirm the accuracy and robustness of the proposed algorithm.