A Novel Image Denoising Algorithm Based on Anisotropic Diffusion Equation

This article first give the concept of the scale space, and construct the relationship between the heat conduction equation and the Gaussian scale space, which lead to the partial differential equations. Then the article introduces linear diffusion equation, nonlinear isotropic diffusion and the nonlinear an isotropic diffusion. After analyze the connection between nonlinear filters and an isotropic diffusion equations, This article bring up the heterogeneity diffusion equation for image denoising method summarized in unique form. Therefore, they formulate the new image denoising algorithm by the novel an isotropic diffusion equation. The denoising image PSNR and SNR from our algorithm is 26.0438 and 11.9716 correspondingly, which are much higher than the PSNR and SNR resulting from L. Alvarez algorithm (25.8112, 11.5682) and Chen's algorithm (25.9292, 11.7415). From the experimental images, it can be found that, the new algorithm for image denoising effect is more effective both in large smooth areas and in the edge region. The proposed method removed Gaussian white noise and strengthened the critical edge, and the detail edge maintained as well. The key information of the image is preserved very well.