Application Of Diffusion Tensor In Image Processing

In recent years, a new method based PDE theory emerge in the field of image processing and attract attention of more and more people. This paper presents models of image denoising based on Partial Differential Equation (PDE), and focus on J.Weickert model. On this basis, we study the diffuseion tensor and apply it to other image processing model. Results of simulation experiments showed that the methods are effective.First, the paper analyzes the advantages of diffusion tensor in detecting feature direction. And applies the diffusion tensor to weight function, then a method for image interpolation based on diffusion tensor is presented. Second, we study the diffusion directions of Prona-Malik model and J.weickert model, and presents two methods:Hessian method(using higher order differentiations) and Gabor method(using space- frequency analysis) to detect the oscillations. Morever, they were applied in image denoising, result of simulation experiment showed that we got a better image. The third, paper combines the J.weickert model with the regularization model, then gets a regularization model based on tensor diffusion, and makes the simulation experiment. The last, Applies the J.Weickert model in the coupled geometry-driven diffusion equations, we have a coupling model for denoising and edge detection based on diffusion tensor. And according to the main function of the equation different, Principle of selecting parameters is discussed.