A New Approach to Optic Disc Segmentation Based on Contrast Enhancement and Brightness Difference

In this paper, we propose a new method of segmenting optic discs in fundus images. Although there is a lot of research related to this, since optical discs are very diverse in shape, complicated approaches are often difficult to apply to various types of optic discs. Therefore, the proposed method uses the most obvious and simple feature to obtain segmentation results. The feature is the difference in brightness between the inside and the outside of the optic disc, a difference which is relatively large. The proposed method is classified into two steps. In the preprocessing stage, an image suitable for optic disc segmentation is generated. In the optic disc segmentation stage, many optic disc candidates are generated using an image processing technique that includes contrast enhancement, and one of the candidates is determined to be the optic disc using the feature. Experimental results show that the proposed method segments the optic disc accurately.