Abstract: This paper presents an anisotropic diffusion model which can enhance the edge of image.This model combines the P-M diffusion model and reverses thermal diffusion model．The model can remove the image noise together with enhancing the edge of the image.It also prevents the blurring effect on the P-M diffusion model for the edge of image，and compared with reverse thermal diffusion mode，it overcomes the disadvantage of false edge’s appearance.The experimental results show that the anisotropic diffusion model can greatly remove the noise and enhance the edge of image.Peak Signal Noise Ratio（PSNR） is 1 dB higher than that of P-M model，which is under strong noise.

Key words: P-M diffusion model, reverse thermal diffusion mode, image noise removal, image enhance, Peak Signal Noise Ratio（PSNR）

摘要： 提出了一个能增强图像边缘的异性扩散模型，结合P-M扩散模型和反热扩散模型各自的优点，能在去除图像噪声的同时增强图像的边缘，一定程度上克服了P-M扩散模型对图像边缘的模糊效应和反热扩散模型容易产生虚假边缘的缺点。实验结果表明：提出的模型有很好的去噪和增强图像边缘的效果，其峰值信噪比（Peak Signal Noise Ratio，PSNR）在强噪声水平下，较P-M扩散模型大约提高1 dB。

关键词: P-M扩散模型, 反热扩散模型, 图像去噪, 图像增强, 峰值信噪比