Automatic Fovea Center Localization in Retinal Images Using Saliency-Guided Object Discovery and Feature Extraction

An accurate and effective detection of fovea is a significant task in computer-aided diagnosis of retinal images. Most of the existing fovea detection algorithms suffer from varying illumination, computational load and abnormal images that is, images affected by pathologies. This paper develops an automatic detection approach based on saliency-guided object discovery and feature extraction for fovea localization in retinal images. Saliency detection and feature extraction techniques are firstly used to locate the optic disc. Then, according to the anatomical prior, i.e., the fovea can be located at an approximately distance of twice the optic disc diameter away from the optic disc center, a search region is defined. Within the search region, the fovea center is located at the center of the largest blob saliency region. The performance of the proposed method is illustrated using two publicly available databases, DIARETDB0 (130 images), and DIARETDB1 (89 images). Compared to the state of the art methods, the experimental proposed method.