基于正常纹理的木材表面缺陷检测算法  

Wood Surface Defect Detection Algorithm Based on Normal Texture

作　　者：张少奇 陆园园 陈志豪 李欣航 李绍丽[1] ZHANG Shsoqi;LU Yuanyuan;CHEN Zhihao;LI Xinhang;LI Shaoli(School of Information Science and Engineering,Shenyang University of Technology,Shenyang 110870,China)

机构地区：[1]沈阳工业大学信息科学与工程学院,沈阳110870

出　　处：《微处理机》2025年第1期47-49,54,共4页Microprocessors

基　　金：辽宁省教育厅高等学校基本科研项目(面上项目)“基于机器视觉的换热器波纹板片冲压缺陷检测方法研究”(JYTMS20231211)。

摘　　要：针对木材表面缺陷种类繁多、缺陷特征各异的问题,提出一种基于高斯混合模型参数自适应的木材表面缺陷检测算法,仅需要对木材表面无缺陷区域提取特征并通过构建算法检测模型即可完成表面缺陷识别。为进一步提高算法智能性,针对训练样本大小和数量如何选择的问题,提出了一种基于正常纹理特征的参数自适应训练样本选择算法。最终通过自适应算法选择训练样本并基于高斯混合模型实现对木材表面缺陷的定位。实验结果表明,所提算法能同时实现对木材表面多种缺陷的检测,提高了木材表面缺陷的检测精确度和准确率,具有一定的应用价值。To address the challenge of diverse defect types and varying defect characteristics on wood surfaces,a wood surface defect detection algorithm based on parameter-adaptive Gaussian Mixture Model is proposed.The algorithm only requires feature extraction from defect-free areas of the wood surface and constructs a detection model to accomplish surface defect identification.To further enhance the algorithm's intelligence,regarding the problem of how to select training sample size and quantity,a parameter-adap-tive training sample selection algorithm based on normal texture features is proposed.Finally,the training samples are selected through the adaptive algorithm,and defect localization on wood surfaces is achieved based on the Gaussian Mixture Model.Experimental results demonstrate that the proposed algorithm can simultaneously detect multiple types of wood surface defects,improving the precision and accuracy of wood surface defect detection,thus showing practical application value.

关 键 词：木材表面缺陷 高斯混合模型 参数自适应 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]