Study on the coupling adaptive fidelity term in anisotropic diffusion based image denoising

A new anisotropic diffusion based image denoising method is proposed in this work. At first, three requirements of image denoising are proposed, based on which the defect of the fidelity term used in the mean curvature motion model and that in the total variation based image denoising models are investigated. After that, a new coupling adaptive fidelity term, which is obtained from image local structural information, is proposed for the diffusion based image denoising. Experimental results show that the denoising method with our new fidelity term is capable of sufficiently preserving geometric information such as edges and corners in addition to its effectiveness for image denoising.