基于图像单应矩阵及边缘优化的自动平面度检测方法  

Automatic Flatness Detection Method Based on Image Homography Matrix and Edge Optimization

作　　者：谢文成 陈金友 XIE Wen-cheng;CHEN Jin-you(Hunan Financial&Industrial Vocational-technical College,Hengyang 421002,China;Tianjin University of Technology and Education,Tianjin 300222,China)

机构地区：[1]湖南财经工业职业技术学院,湖南衡阳421002 [2]天津职业技术师范大学,天津300222

出　　处：《仪表技术与传感器》2023年第3期88-93,共6页Instrument Technique and Sensor

基　　金：湖南省自然科学基金(2021JJ60012)。

摘　　要：零件平面形状公差是体现零件平面凹凸高度的重要指标,为实现在线平面度检测,提出一种基于图像单应矩阵和边缘优化的自动平面度检测算法。首先利用平面单应矩阵约束尺度不变特征描述子对待检平面特征进行粗定位。然后,利用霍夫直线变换检测零件边缘直线轮廓,构建边角结构体搜索轮廓包围的平面区域,获得优化的图像平面。根据视觉成像与激光三角重建原理求解对应的三维点云平面特征。实验表明:该方法有效实现了图像与三维点云的平面自动提取,自动平面度检测时间3~5 s,提升了平面度检测效率及自动化水平。Part plane shape tolerance is an important index to reflect the concave convex height of part plane.In order to realize on-line flatness detection,an automatic flatness detection algorithm based on homography matrix and edge optimization was proposed.Firstly,the plane features to be detected were roughly located by using the scale invariant feature descriptor constrained by the plane homography matrix.Then,the Hough line transform was used to detect the linear contour of the edge of the part,the edge and corner structure was constructed,and the plane area surrounded by the contour was searched to obtain the optimized image plane.According to the principle of visual imaging and laser triangulation,the corresponding plane features of 3D point cloud were solved.Experiments show that this method effectively realizes the automatic plane extraction of image and 3D point cloud,and the automatic flatness detection time is 3~5 s,thus improving the flatness detection efficiency and automation level.

关 键 词：平面度检测 机器视觉 激光测量 平面单应性 自动化 

分 类 号：TP23[自动化与计算机技术—检测技术与自动化装置]