To avoid the time⁃consuming match of multiple feature points and the difficulty in getting the full pictures of Long Shaft Part(LSP) in the process of measuring its 3D pose using non⁃orthogonal binocular stereovision, a method for measuring 3D pose of assembly end of LSP based on orthogonal binocular stereovision is presented. The pictures of assembly end of LSP are taken individually at the same time by two cameras whose optical axes are perpendicular to each other and processed by method of sub⁃pixel respectively, and then the position and posture of assembly end of LSP in the image coordinate systems of two cameras can be obtained. The 3D pose of assembly end of LSP can be acquired from the information fusion of position and posture of LSP in two image coordinate systems according to the model of orthogonal binocular stereovision. Experimental result shows that the 3D pose of assembly end of LSP can be measured rapidly and precisely by orthogonal binocular stereovision consisting of two CMOS cameras. Since only one feature point needs to be matched in this method with simplicity and low calculation cost, it can be applied to the rapid assembly of precise LSP with robots guided by machine vision.%为解决非正交双目立体视觉在长轴类零件三维位姿检测中需要多个特征点进行匹配，计算量大以及视野中难以获取整个零件的图像等问题，提出一种基于正交双目立体视觉检测长轴类零件装配端三维位姿方法。采用光轴正交的两个相机同时采集长轴类零件装配端的图像，分别采用亚像素法对所采得的图像进行处理，可以得到被检测长轴类零件装配端在两个相机图像坐标系内的位置和姿态，再依据正交双目立体视觉模型对两组位置信息进行融合得到其三维位姿。实验结果表明：利用两个CMOS相机构成的正交双目立体视觉能够快速、精确检测长轴类零件装配端的三维位姿。该方法只需要一个特征点进行匹配，计算量小、简单易行，适用于通过机器视觉引导工业机器人进行精密长轴类零件的快速装配。
展开▼
