摘要
为提高纹理模型对统计特征畸变织物疵点的识别率,提出一种GMRF纹理模型自动识别不同种类的统计特征畸变织物疵点。介绍了GMRF模型的纹理合成。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。设计了织物疵点的检测流程,并对实际疵布进行自动检测。结果证明: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。
To improve fabric defect detection of statistic aberration feature by using fabric texture model, the GMRF texture model is proposed to detect fabric defects automatically. Firstly, the synthesis of fabric texture of GMRF model is introduced. Then, the generated GMRF texture model is applied to operate simulation experiment for testifying the correctness of the parameter evaluation algorithm and texture synthesis algorithm. And finally, the detection workflow of fabric defects is designed and automatic detection of real fabric defects is conducted. The experimental results reveal that by using the distance statistic of the parameter structure of the GMRF model, the normal fabric texture can be keenly differentiated from the statistic aberration defect texture. The model is comparatively suitable for automatic detection of fabric defect of statistic aberration feature.
出处
《纺织学报》
EI
CAS
CSCD
北大核心
2013年第4期137-142,共6页
Journal of Textile Research
关键词
GMRF模型
纹理合成算法
特征提取
疵点检测
GMRF model
texture synthesis algorithm
feature extraction
defect detection
