基于GMRF模型的统计特征畸变织物疵点识别

为了提高纹理模型对统计特征畸变织物疵点的识别率，本文提出了一种GMRF纹理模型自动识别不同种类的统计特征畸变织物疵点。首先，介绍了GMRF纹理模型并对GMRF模型参数进行估计；然后，利用生成的GMRF纹理模型进行仿真实验，以验证参数估计算法和纹理合成算法的正确性；最后，设计织物疵点的检测流程，并对实际疵布进行自动检测，实验证明：通过GMRF模型参数构造的距离统计量能够敏感地区分正常织物纹理和统计特征畸变疵点纹理，比较适用于统计特征畸变疵点的自动检测。

Fabric defect detection of statistic aberration feature based on GMRF model

The GMRF texture model is imported to detect fabric defect automatically in this paper. Firstly, GMRF texture model is introduced and evaluate the parameter of this model. Then, the aroused GMRF model is applied to operate simulation experiment, which test the accuracy of parameter evaluation algorithm and texture synthesize algorithm. Finally, the detecting workflow is designed for fabric defect and detected the real defect fabric automatically. It shows that GMRF model can be used to detect statistic aberration defect which existing the surface of fabric keenly.