摘要
针对拉链布边纹理丰富的特点,提出了一种通过LBP算法提取纹理特征,然后由BP神经网络模型和SVM模型检测缺陷的方法。该方法首先采用LBP算子提取训练集样本图像的纹理特征;然后训练BP神经网络模型和SVM模型;最后用同一LBP算子提取测试集样本图像的纹理特征,用上述两种模型进行缺陷检测。实验结果表明,该方法能准确地检测出拉链布边的缺陷。
According to the zipper selvedge is rich of texture,an approach is proposed which through LBP algorithm ex- tracting texture feature and then using the BP neural network model and the SVM model to detect defects.This approach firstly uses LBP operator to extract the texture feature of training set sample images,then training the BP neural network model and the SVM model;finally the same LBP operator is adopted to extract the texture feature of test set sample im- ages,and the feature will be passed to the two model above to detect defect respectively.
出处
《工业控制计算机》
2017年第4期85-87,共3页
Industrial Control Computer
