ETRI-Knowledge Sharing Plaform

KOREAN
논문 검색
Type SCI
Year ~ Keyword

Detail

Conference Paper Multi-Hypothesis Global Point Cloud Registration using Pseudorandom Transformation Generator
Cited - time in scopus Share share facebook twitter linkedin kakaostory
Authors
Hyukmin Kwon
Issue Date
2023-06
Citation
International Conference on Ubiquitous Robots (UR) 2023, pp.1-4
Publisher
IEEE
Language
English
Type
Conference Paper
Abstract
Point cloud registration is an essential step when acquiring 3D data from sensors in multiple poses. Global registration is required when the difference in initial poses is large. Global registration aims to find a global optimal pose between two point sets, but it suffers from falling into a local optimum. In this work, we present a method of generating multiple local optimums with the possibility of a global optimum and selecting the global optimum among them. The main contributions are listed as follows: 1) an algorithm for generating multiple hypotheses using a pseudorandom transformation generator, 2) a criterion for determining that the generated hypotheses include the global optimum, and 3) a metric for selecting the optimal hypothesis corresponding to the global optimum.
KSP Keywords
3D data, Global optimal, Global optimum, Global registration, Local optimum, Multi-hypothesis, Point sets, multiple hypotheses, point cloud registration