基于序列图像的工业钣金件三维重建与视觉检测  

3-D reconstruction and visual inspection of industrial sheetmetal parts with image sequence

作　　者：张永军[1] 张祖勋[1] 张剑清[1] 

机构地区：[1]武汉大学遥感信息工程学院,武汉430079

出　　处：《清华大学学报（自然科学版）》2004年第4期534-537,共4页Journal of Tsinghua University(Science and Technology)

基　　金：国家自然科学基金资助项目(40301041)

摘　　要：针对目前工业制造领域面临的难题,提出利用非量测数字摄像机进行工业钣金件高精度三维重建与视觉检测。采用二维直接线性变换分解摄像机参数初值并结合光束法平差进行高精度标定;利用最小二乘模板匹配提取像面上的点、直线信息并进行混合光束法平差,从而进行钣金件的高精度三维重建及其尺寸制造误差的检测。所开发的视觉检测系统硬件成本低、自动化程度较高,实际数据的多次实验均取得了优于1/8000的相对精度,说明所论述的方法为工业零件的自动化三维检测提供了一条可行途径。A three-dimensional reconstruction and visual inspection method was developed for industrial sheetmetal parts using a non-metric CCD camera. The initial values of the camera's internal and external parameters were decomposed by a 2-D direct linear transformation of several images. The non-metric CCD camera was precisely calibrated using bundle adjustment with a planar control grid. Hybrid point-line bundle adjustment with precise points and lines from images extracted by a line segment least squares template matching is used to accurately reconstruct the industrial part shapes. The reconstruction results can be used to automatically or interactively inspect production accuracy. The visual inspection software uses inexpensive hardware and runs automatically. The relative precision is better than 1/8 000 with image sequences of several parts. The approach is effective for automatic 3-D visual inspection of industrial parts which are mainly described by points and line segments such as sheet metal parts.

关 键 词：序列图像 图像识别 三维重建 视觉检测 最小二乘模板匹配 摄像机标定 工业钣金件 

分 类 号：TH166[机械工程—机械制造及自动化] TP391.41[自动化与计算机技术—计算机应用技术]