Anisotropic Noise Reduction Model Based On Variable Weight Function

Because of the difference in the image collector and the influence of some human factors,the image will be polluted by noise during the process of image acquisition and propagation.The multiplicative noise is a common noise when using hi-teach methods to acquire images.Multiplicative noise reduces image quality,and seriously affects the image segmentation,classification,target detection and extraction of interested regions.Therefore,study and develop the image restoration method with multiplicative noise has important theoretical and practical significance.The traditional image denoising models,such as the ideal low pass filter and the Gauss filter,are all high frequency parts of the image filtering.Because the edge details of the image are also in the high frequency region,so the traditional linear filter will also blurs the edge area and lose a lot of details.Total variation image denoising can not only effectively remove noise,but also preserve the detail edges of images,but it will produce "step effect" in processed images.In order to eliminate the impact of "step effect" in the processed images.In this paper,the PDE method is used to process the image,because it has a good inhibitory effect on the step effect in the image processing.Especially the model based on the high order derivative,it can not only inhibit the "block effect",but also preserve the image details.This paper is based on the research of scholars to make use of the concept of edge control function and edge curvature and edge index.A new adaptive TV model with edge index is proposed,and the edge weight index is used to design the weight function of the sub item,which achieves different denoising effects on the edge region and the smooth area of the image.The established model is solved by the relaxation descent method.