A novel statistical method for 3D range data registration
based on Lie group framework
Yaxin Peng
a
, Wei Lin
a
, Chaomin Shen
b
and Shihui Ying
a
a
Department of Mathematics, Shanghai University, Shanghai, China;
b
Department of Computer Science, East China Normal University, Shanghai, China
ABSTRACT
Registration of 3D range data is to find the transformation that best maps one data set to the other. In this
paper, Lie group parametric representation is combined with the Expectation Maximization (EM) method to
provide a unified framework. First, having a transformation fixed, the EM algorithm is introduced to find the
correspondence between two data sets through correspondence probability, which covers the relationship of all
points, instead of using exact correspondence such as the classical Iterative Closest Point (ICP) method. With
this type of ststistical correspondence, we could deal with the presence of the degradations such as outliers and
incomplete point sets. Second, having the updated correspondence fixed, and introducing Lie group parametric
representation, the transformation is updated by minimizing a quadratic programming. Then, an alternative
iterative strategy by the above two steps is used to approximate the desired correspondence and transformation.
The comparative experiment between our Lie-EM-ICP algorithm and Lie-ICP algorithm using point cloud is
presented. Our algorithm is demonstrated to be accurate and robust, especially in the presence of incomplete
point sets and outliers.
Keywords: Registration, Lie group, Iterative Closest Point method, Expectation Maximization
1. INTRODUCTION
Registration is to find a transformation that best aligns one data set to the other.
1
It is an important issue in
remote sensing. In this paper we focus on the registration of two 3D point data sets.
The Iterative Closest Point (ICP) algorithm
2
is one of the most popular frameworks for data set registration.
With its wide applications, a large number of ICP-based algorithms have been proposed.
On the other hand, many transformations belong to certain Lie groups. Ying et al.
3, 4
introduced Lie group
to parameterize the ICP algorithm for n-D data registration, giving a unified framework to the registration
algorithms. In this paper we further the study of Lie group in registration, focusing on affine transformation,
i.e. a linear transformation followed by a translation.
Unfortunately, most ICP-based algorithms perform unsatisfactory in the presence of noise. Thus, many
methods were developed
5–9
from a probability perspective. In particular, the Expectation Maximization (EM)-
ICP method
8
has become a well-known method for registration of data sets with noise, which greatly improves
the robustness.
In this paper, we adopt the idea of EM-ICP, thus we term our algorithm as Lie-EM-ICP. Our contributions
here is twofold. One is that we generalize the idea of Ying et al.
3, 4
to the case that noise exists; more important,
we present a unified Lie group framework to the EM-ICP algorithm.
Yaxin Peng: E-mail: yaxin.peng@shu.edu.cn
Wei Lin: E-mail: 00linwei@163.com
Chaomin Shen: E-mail: cmshen@cs.ecnu.edu.cn
Shihui Ying: Corresponding author, E-mail: shying@shu.edu.cn
Multispectral, Hyperspectral, and Ultraspectral Remote Sensing Technology, Techniques and Applications IV,
edited by Allen M. Larar, Hyo-Sang Chung, Makoto Suzuki, Jian-yu Wang, Proc. of SPIE Vol. 8527, 85270G
Proc. of SPIE Vol. 8527 85270G-1