Multi-view 3D measurement data registration based on encoding point spatial location and match

Abstract

This paper presents a novel registration method by encoding feature point identification and spatial location to make the registration of 3D measurement easy. A new proposed decoding algorithm based on polar coordinate segmentation is first used for identification feature point, the feature points are then measured and constructed. The overlapped 3D measurement feature points within two views are used to unify coordinate system, so the feature points of each view are achieved for global spatial location. The object is finally measured with any view which only contains at least three feature points. The unconstrained 3D registration is acquired with the feature points matching between single measurement view and global spatial points. Our experiments show that the proposed method is convenient and effective, and greatly enhances the flexibility of 3D measurement applications.