Adaptive Perona–Malik model based on dynamical threshold for image multi–noise removal with details preservation

In this paper, we propose a new dynamical threshold based Perona–Malik (DTPM) model for image denoising. As one of the most famous anisotropic diffusion equations, the PM model has been widely used in noise removal, image segmentation, edge detection and image enhancement. However, the major disadvantages of the traditional PM model are the high difficulty of choosing the best threshold and the tendency of impairing details, so that the denoising is either excessive or insufficient in the whole process. These defects are more obviously when dealing with the image that polluted by noise from multiple distributions. By designing a dynamical threshold function in the edge indicator, we establish a new PM model that can change the diffusion mode and strength adaptively according to the image features. The existence and uniqueness of the weak solution are proven. After theoretical analysis, we give an experimental approach to show the efficiency of this kind of model, especially to the image containing multi–noise and multi–details.