关键词: 高斯噪声, 偏微分方程, 各项异性扩散算法, 扩散系数

Abstract: The paper researches partial differential equations that removes the noise in an image.There is a theoretical analysis on the mechanism of reduction noise.A new diffusion coefficient in partial differential equations（PDE） is built with the advantages of two existing diffusion coefficients and the proposed scheme is incorporated with the nonlinear time-dependent cooling technique for gradient threshold.The new algorithm is compared with traditional anisotropic diffusion algorithm.The results show the noises in the images are removed and the image edgedetails are reserved.The method can reduce the noise and improve the quality of the images effectively.

Key words: Gaussian noise, partial differential equations, gnisotropic diffusion algorithm, diffusion coefficient