Abstract
Scanned meshes of engineering objects have been easily acquired by 3D laser or high energy industrial X-ray CT scanning systems, which are then widely used in product development. When utilizing the meshes in object inspection, re-design, and simulation, it is very important to use them to reconstruct 3D CAD models. Engineering objects often exhibit 3D periodicities for their functionality. Therefore, it is essential to recognize such 3D periodicities when reconstructing 3D CAD models that contain compact data representations. So far, we have proposed a method for recognizing 3D periodicity patterns based on the indexed-ICP algorithm. The method robustly extracts a pattern type, a set of periodically displaced regions and the parameters defining the periodicities, such as translational basis vectors or a rotational axis and a basis angle, from scanned meshes. In this paper, we propose a new method that can extract 3D periodicity patterns where periodical displacement among regions can be expressed by 3D generalized coordinates.
