A Fast Image Restoration Algorithm Based on Variational Model

This paper studies a fast image repair algorithm based on the TV model, which combines two diffusion methods, anisotropic and isotropic, to determine whether the local region of the pixel to be repaired contains edge texture information based on the degree of gray fluctuation between the effective pixels in the local region of the pixel to be repaired. The TV repair model with anisotropic diffusion is used to iteratively repair the points to be repaired that contain edges in the local region, Effectively protecting the edge texture information of the image; For points to be repaired that do not contain edges in local regions, using isotropic diffusion iterative convolution repair can effectively avoid the ladder effect introduced by the TV repair model in smooth or flat regions, while accelerating the diffusion speed of effective information and reducing the repair time.