机械臂视觉相机除了有视觉监视作用外,还可以对观测目标进行三维位姿测试。在相机整星安装阶段,需要将相机坐标系引入到相机安装坐标系中。由于相机坐标系的原点为针孔模型的投影中心,为一虚拟位置,无法通过直接测量的方式给出其具体位置。提出了一种基于自标定技术引出相机坐标系的方法,该方法通过相机对目标拍照,根据多幅图像间的对应匹配特征点,可得出相机的方位参数,进而引出相机坐标系。该方法简单直接,引出精度可达±0.117mm,能够满足测试需求和测试精度,对摄影测量相机的装调具有指导意义。 In addition to the role of visual surveillance robotic arm camera, but also to observe the target for three-dimensional pose test. In the camera installation phase of the whole star, the camera coordinate system needs to be introduced into the camera installation coordinate system. Since the origin of the camera coordinate system is the projection center of the pinhole model, it is a virtual position and its specific position can not be given by direct measurement. A method based on self-calibration technique is proposed. The method takes the camera’s image of the target and matches the feature points according to the correspondence between the multiple images to obtain the camera’s azimuth parameter, which leads to the camera coordinate system. The method is simple and straightforward, and leads to a precision of ± 0.117mm, which can meet the test requirements and testing accuracy, and is of guiding significance for the installation of photogrammetric cameras.