A Texture Image Denoising Model Based on Image Frequency and Energy Minimization

Abstract

When partial differential equation (PDE) and variation approaches are used in image denoising, its diffusion mechanism only depends on the image gradient, thus, the denoising effect is easily interfered by the noise. It is a difficult problem to suppress influence of the noise, to improve the model’s anti-noise performance and to meet the case of suppressing noises while preserving edges and other detail features. This paper first gives the definition of image frequency and then proposes a novel texture image denoising model based on image frequency and energy minimization (NTIEM) on the basis of ROF total variation (TV) model. Meanwhile, in order to ensure the computational stability, we introduce a potential function into the NTIEM model. Theoretical analysis and numerical experiment have shown that compared with other existing approaches the NTIEM model has an obvious antijamming capability and can accurately and subtly describe the image edge area and smooth area. The analyses of experimental results have indicated that the NTIEM method can overcome staircase effect and over-smoothing. Especially for the images with rich texture features, it can remove the noise while preserving significant image details and important characteristics and improve the image peak signal to noise ratio (PSNR).