Enhancing retinal fundus images based on MPA optimization algorithm

Abstract

The retinal fundus image serves as a crucial tool in diagnosing numerous prevalent diseases, given that many significant conditions manifest on the retina. Nevertheless, images obtained in real clinical environments often lack quality for accurate diagnosis due to issues such as uneven lighting, blurriness, and low contrast. In this article, we present a simple yet effective approach to enhance retinal fundus images. Initially, the input images are transformed from the RGB color space to the YUV color space, yielding three components: Y, U, and V. The Y component is enhanced using the Contrast-limited Adaptive Histogram Equalization (CLAHE) method. Subsequently, it is decomposed into three layers: structure layer (L_S), texture layer (L_T), and noise layer (L_N) using the Relative Total Variation (RTV) method and Anisotropic Diffusion Filter (ADF). Next, the structure component is enhanced by the BLLI (Brighten low-light image) method, resulting in the L_B component. The MPA optimization algorithm is employed to find optimal parameters for the L_B, L_S, and L_C components. The components are then multiplied by the optimal parameters and summed to obtain the enhanced image. Experimental results demonstrate the effectiveness of the proposed method in improving the quality of retinal fundus images. Keywords: Relative total variation (RTV) method, anisotropic diffusion filter (ADF), marine predators algorithm (MPA)