LAUSR

A grating projection shape measurement system has been a commonly used method in the field of three-dimensional (3D) reconstruction in recent years, and global point cloud registration is a key step in this method. However, in the registration process, a large amount of low-precision overlapping redundant data (ORD) is generated between adjacent camera stations, which will seriously affect the speed and accuracy of later modeling. Therefore, how to eliminate these low-precision ORD is a major problem to be solved at present. Determining all overlapping 3D point pairs between two adjacent stations and deleting the points with low precision in the point pairs is the key to solving this problem. Therefore, based on an omnidirectional rotation measurement system, combined with the constraint relationships between the projection space and the acquisition space in the global registration process and the stereo-matching method of space conversion, an elimination algorithm for ORD with a two-dimensional (2D) phase constraint and a 2D pixel constraint is proposed. The experimental results show that the proposed algorithm can faster locate overlapping 3D point pairs between adjacent stations, with a higher elimination rate, and the accuracy of the overall point cloud is higher after the redundancy elimination.