通过对断口图像FGLCM的14个特征统计量进行相关性分析，选择角二阶矩和熵等7个统计量作为特征参数，并验证了其有效性.最后，在4类典型断口图像的特征空间上，采用隐马尔夫模型（HMM）进行分类识别。-On the fracture characteristics of the 14 images FGLCM correlation analysis statistics, select the angular second moment and entropy 7 statistics as parameters, and verified its effectiveness. Finally, in four typical fracture image of the feature space , using hidden Malfoy model (HMM) for classification.