An Improved Image Denoising Method Based on Variable-order Fractional-Order Anisotropic Diffusion
Xiaohu Shu, Jialing Han, Yapeng Zhan, Zelong Wang · 2022 7th International Conference on Signal and Image Processing (ICSIP) · 2022
In recent years, the fractional partial differential equations based method has been widely studied in the image denoising field. Usually, constant-order fractional differential order is employed to fractional partial differential equations based method, which fails to simultaneously achieve the optimization of denoising effect and structure preserve ability. In this paper, we proposed a variable-order fractional-order anisotropic diffusion model. The method adaptively divides the image into the homogeneous region and the non-homogeneous region by local variance. In the homogeneous region, a relatively small fractional differential order is employed to improve the denoising effect and avoid the staircase effect. In the non-homogeneous region (including edge and texture), a relatively large fractional differential order is employed to improve structure-preserving ability. Experimental results show that the proposed method has a better denoising effect and structure-preserving ability than the traditional.