在使用扩散过程平滑噪声之后引入反扩散过程来恢复边缘，结合尺度空间理论和反扩散函数对图像进行去噪处理。该方法使用最小描述长度(MDI )准则自适应地选择图像中每一点处的最优尺度对图像进行滤波。加入尺度范围限制降低了过平滑和欠平滑的影响。改进了反扩散函数模型，对降质图像中的边缘进行恢复。与经典的滤波方法以及各向异性扩散方程的结果相比。本文方法取得了较好的效果。-Diffusion process in the use of noise after smoothing the introduction of anti-diffusion process to restore the edge, combined with scale space theory and the anti-spread function of image denoising treatment. This method is the use of minimum descr iption length (MDI) criteria for adaptive selection of images every point in the Agency