校准
投影机
计算机科学
结构光三维扫描仪
计算机视觉
投影(关系代数)
相(物质)
人工智能
转化(遗传学)
领域(数学)
点(几何)
像素
视野
摄像机切除
算法
数学
物理
几何学
纯数学
扫描仪
化学
统计
基因
量子力学
生物化学
作者
Shaoyan Gai,Feipeng Da,Ming Tang
标识
DOI:10.1088/1361-6501/aaf5bd
摘要
Multi-view systems are an effective scheme for solving the problem of a small field-of-view and self-occlusion in optical 3D measurements. However, due to the dispersed distribution of views and the lack of a common field-of-view, the problem of global calibration is difficult and essential. A new method is proposed, by which digital fringe projection and a phase map are used to establish global calibration information. Firstly, fringes are projected to the field of the system. By using a phase unwrapping algorithm, an absolute phase map is obtained. The projector is deemed as the reverse of a camera, which can be calibrated in a similar way to that of the camera. Therefore, the parameters of the cameras, projector and the location of each calibration plane are calculated, and the transformation relationship of adjacent cameras is obtained. Secondly, the phase is used in 3D measurement as well as the global registration of views. Sub-pixel point matching and 3D reconstruction is performed according to the phase map. Global system parameters are optimized by minimizing the reconstruction errors, and precise calibration results can be achieved. With this method, a large-sized calibration target and other auxiliary calibration devices are not required. It is reliable and easy to implement multi-view calibration and reconstruction. The calibration model and calculation processing are simple. The experiment is carried out on a multi-view 3D measurement system. It is shown that high precision multi-view calibration and reconstruction results are obtained.
科研通智能强力驱动
Strongly Powered by AbleSci AI