基于序列图像的工业零件尺寸检测技术

提出利用像面上的点、直线信息进行混合光束法平差，对工业零件进行高精度三维重建并检测其尺寸误差。介绍了零件坐标系与物方坐标系之间的变换方法，讨论了直线摄影测量误差方程式的基本形式、基于距离的改化形式以及点线混合光束法平差的数学模型，提出在适当选取直线观测值的权值时，可以按间接平差模型解算直线空间前方交会。所开发的工业零件视觉检测系统可全自动运行，取得了约0．1mm的实验精度，为工业领域中广泛存在的以直线段为主要特征的工业零件如钣金件的自动化三维检测提供了一条有效途径。

Dimensional Inspection of Industrial Parts with Image Sequence

A novel dimensional inspection approach for industrial parts based on hybrid point-line bundle adjustment with point and line observations extracted precisely from digital image sequence was proposed. A coordinate transformation between the part coordinate system and the world one was introduced. The basic error equation of line photogrammetry and its modified form were addressed. The model of hybrid point-line bundle adjustment was also discussed. It is shown that when certain proper weight is selected, the adjustment by observation equations can be used for line photogrammetry. The developed inspection system runs automatically. Precision of 0. 1mm is obtained with true image data. It shows a promising potential in the automatic dimensional inspection of the widely existed industrial parts mainly composed of line segments such as sheetmetal parts.