Robust time-fractional diffusion filtering for noise removal

In this paper, we utilize a time-fractional anisotropic diffusion equation for image denoising. Theoretical results are provided thanks to Schauder fixed-point theorem. A discretization scheme by finite difference is also presented. Numerical experiments show a great performance in deleting the noise while preserving important features. The model robustness is tested, which is clear visually and quantitatively. All the obtained results conclude that our model surpasses the competitive models, such as Perona-Malik and Weickert.