基于Krawtchouk矩和小波支持向量机的纸张缺陷识别方案设计  

Design of Paper Defect Recognition Scheme Based on Krawtchouk Moment and Wavelet Support Vector Machine

作　　者：魏彬 李璐 WEI Bin;LI Lu(Shaanxi Railway Institute,Weinan 714000,China)

机构地区：[1]陕西铁路工程职业技术学院,陕西渭南714000

出　　处：《造纸科学与技术》2022年第5期73-76,共4页Paper Science & Technology

基　　金：渭南市科技局科研项目(2019-ZDYF-JCYJ-146)。

摘　　要：为实现对于纸张缺陷的高精度识别,提高纸张成品的生产质量,通过Krawtchouk矩不变量来提取纸张缺陷图像的特征向量。在此基础上,运用所获取的特征向量对小波支持向量机进行训练,并通过粒子群优化算法来获取惩罚因子与核函数参数,使小波支持向量机的分类性能达到最优。最后对提出的纸张缺陷识别算法的识别率进行试验,经统计分析发现,访算法的识别总精度高达98.33%。In order to realize high-precision recognition of paper defects and improve the production quality of paper products, this study uses Krawtchouk moment invariants to extract the feature vectors of paper defect images. On this basis, the wavelet support vector machine is trained by using the obtained feature vectors, and the penalty factor and kernel function parameters are obtained by particle swarm optimization algorithm, so that the classification performance of wavelet support vector machine is optimized. Finally, the recognition rate of the paper defect recognition algorithm proposed in this study is tested. Through statistical analysis, it is found that the total recognition accuracy of the algorithm is as high as 98.33%.

关 键 词：小波支持向量机 KRAWTCHOUK矩 纸张缺陷识别 混沌粒子群优化算法 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]