作者：徐平, 沙从术

作者单位：河南工程学院土木工程学院, 河南 郑州 451191

关键词：表面缺陷;结构光;检测;设计

摘要：

为能够对产品表面的孔/裂纹缺陷进行检测，提出基于结构光的表面缺陷检测系统的设计与研究。在获得检测目标点云数据后，基于坐标值的不同区分面和缺陷点云集合，进而利用k-means函数获取孔和裂纹数量，并基于KDtree函数和knnsearch函数得到对应数量的孔和裂纹点云集合，然后利用minBoundingBox函数将点云集合外轮廓拟合为外接四边形，并根据孔的外接四边形长度基本相等的特点区分孔和裂纹点云，最后根据pdist和minBoundingBox函数提取到孔和裂纹的特征数据。试验结果表明：该系统可以实现表面孔/裂纹缺陷的检测，孔径测量的最大误差为3.03%，孔深度测量的最大误差为2.76%，裂纹的最大宽度误差为3.63%。表明测量的可靠性良好，该系统可为产品缺陷检测提供一种有效的方法。

Design and development of product surface defect detection system
XU Ping, SHA Congshu
College of Civil Engineering, Henan University of Engineering, Zhengzhou 451191, China
Abstract: In order to be able to detect hole/crack defects on the product surface, a surface defect detection system based on structural light is proposed. Firstly, the structural light detection system was introduced. After obtaining the detection target point cloud data, the surface and defect point cloud can be divided based on different coordinate value. Then the number of holes and cracks is calculated using the k-means function and the holes and cracks point cloud obtained using the KDtree and knnsearch function. Therefore, the holes and cracks point are separated by minBoundingBox function based on the characteristics of hole with equal length of the circumscribed quadrilateral. Finally, the characteristic of the hole and crack are extracted using pdist and minBoundingBox function separately. The test results show that the system can realize the detection of surface hole/crack defects and the maximum error of the hole diameter is 3.03%, while the maximum error of the hole depth is 2.76%. Moreover, the maximum error of the crack is 3.63%. It shows that the measurement reliability is good and the system can provide an effective method for product defect detection.
Keywords: surface defect;structure light;dectection;design
2020, 46(6):34-38  收稿日期: 2019-12-26;收到修改稿日期: 2020-03-11
基金项目: 2018年河南省产学研合作项目（182107000005）
作者简介: 徐平(1980-),女,山东济南市人,讲师,硕士,主要研究方向为检测系统设计与开发
参考文献
[1] 李昱坤, 石照耀, 于渤, 等. 线结构光传感器的静态性能测试研究[J]. 机电工程, 2019, 36(9): 925-930
[2] 徐志祥, 王铮恭, 王雨, 等. 激光激发超声波检测带涂层铝板的表面缺陷[J]. 机械工程与自动化, 2019(5): 162-163,166
[3] 刘亚楠. 基于机器视觉的钢材表面缺陷识别方法分析[J]. 计量与测试技术, 2018, 45(12): 97-98,102
[4] SOUKUP D, HUBER R. Convolutional neural networks for steel surface defect detection from photometric stereo images[J]. International Symposium on Visual Computing, 2014, 8887: 668-677
[5] WIN M, BUSHROA A, HASSAN M, et al. A contrast adjustment thresholding method for surface defect detection based on mesoscopy[J]. IEEE Trans. Industrial Informatics, 2015, 11(3): 642-649
[6] 王中旺, 蒋全胜, 李华. 一种基于三坐标测量机的齿形参数精密测量方法[J]. 机床与液压, 2019, 47(12): 33-38
[7] YU Q, WANG K S. 3D vision based quality inspection with computational intelligence[J]. Assembly Automation, 2015, 33(3): 240-246
[8] 阴雷, 王向军, 秦贯宇. 基于视觉结构光的大型曲面动态测量方法研究[J/OL]. 激光与光电子学进展: 1-17[2019-10-18]. http://kns.cnki.net/kcms/detail/31.1690.TN.20191015.0934.016.html.
[9] 马宗正, 刘杰, 潘高杰, 等. 基于结构光表面缺陷检测系统设计与开发[J]. 现代制造工程, 2018(5): 139-143
[10] 吴付峰, 李筠, 杨海马, 等. 基于线结构光路面车辙检测系统研究[J]. 电子测量技术, 2019, 42(23): 132-136
[11] 史博伟, 闫志鸿, 卢振洋. 基于图形与图像特征的铝合金焊缝成形检测[J]. 热加工工艺, 2020, 49(5): 131-136