智能三坐标测量机的零件位姿识别

提高三坐标测量机对零件检测的智能化，实现零件的位姿识别是非常必要的。提出了一种新的视觉识别零件及位姿的方法，该方法首先比较全面地提取实际零件的三维特征信息，然后根据零件三维特征在虚拟空间中寻找匹配组合，并根据匹配组合计算旋转平移矩阵，最后根据实际和虚拟的零件成像轮廓的相似度进行位姿识别。实验表明新方法能够对零件及位姿进行识别，并且最大误差不超过1mm。

Part Position and Orientation Recognitions on Intelligent CMM

In order to improve the intelligence of CMM for part detection, it was necessary to recognize positions and orientations of parts. A new visual recognition method for positions and orientations of parts was proposed. The new method firstly extracted actual 3D feature informations of parts more comprehensively, and then matched combinations of the 3D features were found in virtual space. Rotation matrix and translation matrix were computed by matched combinations. Finally, the positions and orientations were determined according to part contour matching similarity between real and virtual imaging. Experiments show that the positions and orientations may be recognized by the new method.