摘要
给出了一种基于LPQ特征向量的帘子布疵点识别方法。首先给出了局部相位量化(LPQ)的定义,,然后计算帘子布样本图像的LPQ特征向量,使用PCA降维处理,再利用降维后的LPQ特征向量对预先设置的BP神经网络参数进行最优选择,最后利用最优的BP神经网络作为帘子布疵点分类器。此识别算法能对断经、浆斑、劈缝、稀经和经线粘连等帘子布疵点进行有效识别。
In this paper, the method based on the Local Phase Quantization are presented to recognize the de- fects of the cord fabric. First of all, A new kind of the Local Phase Quantization is constructed, and its implemen- tation is given. Then, the Local Phase Quantization of the cord sample images are calculate, the BP neural net- work is trained by these moments. At last, it is used trained BP neural network to implement the cord fabric＇ s de- fect identification. Experiment results show that the method accurately identifies defects such as broke end, lump, split slot, broken warp and warp adhesion.
出处
《安阳工学院学报》
2015年第4期38-40,共3页
Journal of Anyang Institute of Technology
关键词
局部相位量化
主成分分析
疵点识别
神经网络
the local phase quantization
the principal component analysis
fabric defect recognition
neural network
