A Survey on Rigid Point-Sets Registration

Junfen Chen*, Ming Zhang

Key Lab of Machine Learning and Computational Intelligence, College of Mathematics and Information Sciences, Hebei University, Baoding, 071032, China

Abstract: Registration of two point-sets involves finding meaningful correspondences among points, ｏｒ reco-vering the underlying spatial transformation. There already exist several different registration methods, which can be grouped into rigid registration and non-rigid registration. Rigid registration is considered be easy but quite necessary, compared to non-rigid registration. This paper summarizes rigid point-sets registration me-thods such as Iterative Closest Point (ICP), Principle Component Analysis (PCA) and Singular Value Decom-position (SVD), in where point-sets are the 2D ｏｒ 3D coordinates usually of either a surface of an object, ｏｒ the points occupied by the object. More specifically, this paper provides a comprehensive overview of rigid point-sets registration methods.

Keywords: Rigid registration; Point-set; Iterative closest point (ICP); Transformation; principal component analysis (PCA)