基于LBP和HOG特征的图案织物缺陷检测

Patterned fabric defect detection based on LBP and HOG feature

为解决图案织物缺陷检测时传统人工的误检率、漏检率较高的问题,提出一种基于局部二值模式(local binary pattern,LBP)和方向梯度直方图(histogram of oriented gradient,HOG)特征相结合的检测算法。首先,将织物图像分解为多个重复单元(repeat units,RUs),提取其LBP和HOG特征,并对特征降维;其次,根据标记每个RUs特征的类别和对应在织物图像上的位置训练支持向量机(SVM);最后,利用分类器判别RUs特征中有无缺陷,并定位出RUs在织物图像中的位置。实验结果表明,与灰度共生矩阵(GLCM)作为特征矩阵的方法相比,该算法对图案织物常见的6种缺陷图像可实现提高检测效率、缩短检测时间,获取准确位置的目的。

To solve the problem of high error detection and omission rate of traditional artificial detection in patterned fabric defect detection,the patterned fabric defect detection algorithm based on LBP and HOG feature is proposed. Firstly,the fabric image is decomposed into multiple repeat units( RUs),and the features of LBP and HOG are extracted,meanwhile the feature dimension is reduced. Secondly,the labeled RUs samples that contain the class of features and the position corresponding to the fabric image are used to train a support vector machine( SVM). Finally,the classifier is used to discriminate that with or without defect in RUs features and the location of the repeating unit in the fabric image is returned. The experimental results show that compared with the method where GLCM is taken as a characteristic matrix,the proposed algorithm achieves the purpose of improving efficiency,shortening the time of measurement,and obtaining an accurate defect location when it comes to six common fabric defects.