Image De-noising Model Based On Variable Norm Of Total Variation

The mainly content of this paper is the introduce of the application in the total variation model in denoising and the improved algorithms of the total variation in denoising. With the development of digital science and technology, the digital image processing has been applied more and more extensively, including the recognition of biological characteristic, traffic control, a variety of industrial and military field etc. The digital image processing is an essential and very important tool for the development of social wisdom. Durning the digital image processing, it is inevitable to be interfered by many factors, thus creating noise pollution, which resulting in degradation of image quality or even losing the value of image. How to improve the effectiveness of image processing technology is also very important. At this stage of the image processing technology, the image processing model based on total variation not only has a good mathematical properties and efficient numerical algorithms, but constrained relatively small of image itself. Therefore, the total variation method has aroused general concern in the field of image processing.At frist, this article briefly described some classical total variation denoising model, and mainly studying the adaptive total variation denoising model. According to the characteristics of these models, we propose two new adaptive edge indicators ATV model with the edge indicators, and establishe a new regularization and diffusion coefficient in the models. By the values of the difference curvature, these models set the weight parameters of the relevant information according to the characteristics of the image itself. In the first model, the regularizing term is added to the Hessian matrix with the edge indicator to reduce the " staircase effect" of denoising image, and the Hessian matrix takes the change form of F-norm. In another model combining the denoising model of TV with fourth-order differential theory, a new adaptive ATV model with the edge indicator was obtained, and used difference equation to define variables in the model.The experimental simulation results show that this two image denoising methods not only can inherited advantages of the classical total variation, but also can reduce the noise well, preserve edge features better, reduce the influence of the "staircase effect" , avoid the "blocky effect" of the image.