单目视觉
校准
计算机科学
计算机视觉
单眼
人工智能
坐标系
光学
物理
量子力学
作者
Shuai Fu,Liyan Zhang,Nan Ye,WeiZhong Zhang,Shenglan Liu
标识
DOI:10.1088/0957-0233/25/12/125006
摘要
A monocular-vision-based coordinate measuring system (MVB-CMS) obtains the 3D coordinates of the probe tip center of a light pen by analyzing the monocular image of the target points on the light pen. The light pen calibration, including the target point calibration and the probe tip center calibration, is critical to guarantee the accuracy of the MVB-CMS. The currently used method resorts to special equipment to calibrate the feature points on the light pen in a separate offsite procedure and uses the system camera to calibrate the probe tip center onsite. Instead, a complete onsite light pen calibration method is proposed in this paper. It needs only several auxiliary target points with the same visual features of the light pen targets and two or more cone holes with known distance(s). The target point calibration and the probe tip center calibration are jointly implemented by simply taking two groups of images of the light pen with the camera of the system. The proposed method requires no extra equipment other than the system camera for the calibration, so it is easier to implement and flexible for use. It has been incorporated in a large field-of-view MVB-CMS, which uses active luminous infrared LEDs as the target points. Experimental results demonstrate the accuracy and effectiveness of the proposed method.
科研通智能强力驱动
Strongly Powered by AbleSci AI