Bidirectional Diffusion Algorithm for Image Enhancement with Local Feature

Abstract

To deal with the problem that the partial differential equation on image enhancement causes the edge blurring, block effect and ringing effect and so on, a new bidirectional diffusion model based on local features, which smooths the flat region and sharpens the edge in image, is presented. According to the edge features, the presented model chooses the different diffusion methods, that is to say, the non-edge is denoised by the forward diffusion in normal direction and the forward diffusion in tangential direction, while the edge is sharpened by the backward diffusion in normal direction and the forward diffusion in tangential direction. Through the simulation experiments, the model can not only better preserve the image feature, but also more obviously improve the defect contrast.