A Novel Method of Detecting and Locating the Macula Based on Morphological Features Fitting

Color retinal images have been widely used in diagnosis and screening of ophthalmic diseases. Macula detection and fovea location in retinal images are important steps in grading and diagnosis of ophthalmic diseases. However, due to the weaker reflection of the macula, it is difficult to determine the fovea accurately by the lack of clear border in the macula. An efficient method for detecting and locating macula based on morphological features fitting is proposed. The macula area is detected by its low brightness and round shape features. And then the K-means clustering and circle fitting is applied to achieve accurate position of macular center. A series of tests show good performance in the public retinal images database. Thus the proposed method based on morphological features fitting is proved to be efficient and useful for computer-aided diagnosis of ocular diseases.