A Method For Image Smoothing Based On PDEs

In the course of filming and transmission, images are often mixed with a lot of noise. Image noise not only seriously affects the image of visual effects, but also brings certain difficulties to the subsequent image analysis and understanding in the image processing. So image smoothing is a very important step for the follow-up treatment.Traditional image smoothing algorithms such as mean filter, median filter and gauss filter belong to the isotropic diffusion. They remove the noise of the image while undermine the character of the image. The image smoothing technology based on partial differential equations can solve this problem. They remove the noise while enhance the edge of image. Smoothing image should balance the removing noises and preserving features. Image smoothing technology based on PDE is an adaptive smoothing method. During the course of smoothing, image features and their directions are extracted. There are less smoothes in the locations with strong image features, more smoothes in the locations with weak image feature;There are minimal smoothes in the directions across the image features, and maximal smoothes in the directions along the image features. The smoothing can make noises removed while preserve edge well. And it is a quite well smoothing method. Perona-Malik models represented by such models are already widely applied in marginal detection, image enhancement and so on.In this paper, an anisotropic nonlinear diffusion equation for image smoothing is presented. The mode combines the total variation model and ALM model. The experimental results show the effective performance of the combination of these two procedures in image smoothing.