ICPとリージョングローイングを組み合わせた機械部品計測メッシュのユークリッド対称性認識手法

概要

論文の詳細を見る
Recently scanned meshes from mechanical objects can be easily acquired using high-energy industrial X-ray CT scanning technologies, and these meshes are widely used in various fields of product developments. Mechanical objects often exhibit symmetries for their functionality. They are originally defined by creating a copy of the solid and then by placing it with a certain transformation including an arbitrary set of translations, rotations, and reflections. Therefore it is desired to detect such symmetries from the scanned meshes for their effective use in mesh applications, such as reconstruction of CAD models with compact data representations or CAE mesh enhancement using the symmetries. In this paper, we propose a method that can detects multiple Euclidean symmetries from scanned meshes of mechanical objects based on a combination of ICP and region growing algorithms. We define Euclidean symmetries detection as to detect both pairs of regions that match under certain transformations and their transformations including arbitrary sets of translations, rotations, and reflections. We also propose a method that derives all possible sequences for reconstructing an entire model from the original mesh using planar-reflective symmetries among all the detected symmetries, and represents them in an AND/OR graph in order to compactly represent the mesh. Our proposed method contains several advantages. It can robustly detect symmetries from noisy scanned meshes that include many planar regions. Our method also can detect both global and local Euclidean symmetries from meshes. Moreover it can detect maximally large pairs of regions under user specified tolerances. We demonstrate the effectiveness of the proposed method from various experiments using X-ray CT scanned meshes of mechanical objects.

ICPとリージョングローイングを組み合わせた機械部品計測メッシュのユークリッド対称性認識手法

スポンサーリンク

概要

著者

関連論文

スポンサーリンク