Formula Based On The Value Of The Tv Model

Variational methods is a major method of image processing. The TV image diffusion model based on total variation is the foudational work in this area, and has successful applications in image denoising, optic flow computation, image inpainting, image segmentation and so on. The classic TV(Total Variation) model can be used to diffuse images, but it also can result in staircase effects beyond edges, and the PDE's computational efficiency is very low. Variational method is one of the mainstream approach to calculate optical flow, but the computational efficiency is very low. Median formula based on TV model discussed in this paper, Several aspects are introduced as follows:First, A new median formula is proposed based on traditional TV-L2 variational image diffusion model, and is used in TV-L1 variational image diffusion model, and then extended to higher order image diffusion model to show their capabilities of edge diffusion and edge preserving/enhancement and to reduce staircase effects. Second, Variational method was used to compute the optical flow, different hypotheses of the data term and the smooth term in optic flow computation model are studied. Then, median formula was studied and used to compute optic flow, which could avoid the complex computational processing and increase the efficiency. Third, propose Bregman iteration parameter, and combine the TV model and Bregman iteration parameter. And then use median formula to solve TV minimization problems. The new method yields a more accurate de-noised image, and thus, should be a more precise solution to the problem.