Abstract: Analysis of metal fracture images can provide important information for research such as performance，action of metal materials.Denoising preprocessing of metal fracture images is vital for morphology analysis of metal fracture.Firstly images are divided into noise sub-image，inner region sub-image and edge sub-image.Then adaptive anisotropic diffusion filter is designed according to the fracture images with different noise.The adaptive anisotropic diffusion filter made parameters determination and the termination of iteration automatically.The filter has better adaptation for different noise，more accurate of smoothing selection，furthermore，protects edges and low contrast regions better.The experimental results show that the improved method has better performance of anti-noise and detail preservation，which are essential to the accuracy of pattern analysis and quantity computation.

摘要： 分析金属断口图像可以为金属材料的性能及行为等许多方面的研究提供重要信息，准确分析金属断口的形貌需要对金属断口图像进行去噪等预处理。首先将图像划分为噪声、区域内部和边缘三类子图，然后针对受不同类型噪声污染的断口图像设计了自适应确定参数和中止条件的各向异性扩散的去噪方法，使滤波对不同噪声类型的适应性更好，对噪声的选择性平滑更准确，并能保护边缘和低对比度区域。实验结果表明该方法对噪声污染的金属断口图像有更好的抗噪效果和细节保持能力，有利于提高后续模式分析和定量计算的准确性。