Abstract
Retinal images show an essential role in Ophthalmology to diagnosis wide set of diseases. In this direction, using retinal images in computerized techniques increases the ability of diagnosis in fast time effectively. However, some eye diseases and capturing conditions produce low-quality retinal images, which reduces the diagnosis ability for machines and humans. To solve that, several works have been proposed to enhance retinal images. But they show a lot of negative observations, especially with color images of retina. In this paper, a novel enhancement algorithm for color retinal images is proposed. It consists of three stages; firstly, the appearance of visual details is increased by enhancing the contrast of structural details of retinal image using details enhanced and Bilateral filters. Then, a novel uneven illumination correction method is proposed to solve the uneven illumination problem adaptively. Finally, the advantages of both previous stages are combined using HSV color model to produce the final enhanced retinal images. DRIVE and STARE benchmark datasets are used to conduct experiments. The results were compared with histogram Equalized (HE), Contrast Stretching (CS), the adaptive histogram equalization (CLAHE) and Zhou's method retinal enhancement methods. In conclusion, the results show that the proposed method shows high performance compared with the corresponding enhancement methods.