Optical Disc Detection in Retinal Image Based on Spatial Density of Grayscale Pixels

Retinal Images contain vital information that extraction of these features has various applications in medical diagnosis and identification applications. Optical disk (OD) is one of these features that it can detect different eye retinopathies through analyzing it. In this paper, a method is proposed for optical disk segmentation based on spatial density of grayscale pixels. In the proposed method, firstly image has been decomposed by Principle Component Analysis (PCA) and then preprocessing has been made to improve quality of images. Then, image is segmented via Otsu image segmentation. Ultimately, the segments pertaining to OD are detected and other segments are differentiated using thresholding and averaging methods. In this research, to evaluate our proposed method, DRIONS dataset consisting of 110 fundus images of retina and DRIVE dataset consisting of 40 retinal images are applied. The proposed method has been obtained accuracy rate of 94% and 73% on two datasets.