The retina is an important area for medical treatment of the disease. By observing the changes in the blood vessels in the retina lines help doctors diagnose diseases, to collect and analyze the symptoms and the development of related treatments. Consequently, improve retinal image quality is an important preprocessing step. And to improve retinal image quality several techniques have been proposed such as Histogram Equalization, Local Normalization, Adaptive Histogram Equalization Contrst Limit, Lapacian, .... but still can not provide high efficiency by persists high noise and poor image results. Therefore, in this paper, we propose a method of raising the quality of retinal images using filter change curvelet combines nonlinear diffusion and minimum Minimax algorithm. By the analysis and calculation results in picture quality parameters through experimental treatment, we will draw conclusions indicate that the proposed method improves the image quality better than previous methods.