- Abstract:
-
Most range data registration techniques are variants on the iterative closest point (ICP) algorithm, proposed by Chen and Medioni [2] and Besl and McKay [3]. That algorithm, though, is only one approach to optimizing a least-squares point correspondence sum proposed by Arun et al [1]. In its basic form ICP has many problems for example, its reliance on pre-registration by hand close to the global minimum and its tendency to converge to sub-optimal or incorrect solutions.
This paper reports on an evolutionary registration algorithm which does not require initial pre-alignment and has a very broad basin of convergence. It searches many areas of a registration parameter space in parallel and has available to it a selection of evolutionary techniques to avoid local minima which plague both ICP and its variants.
- Copyright:
- (c) 2002 by Elsevier/Academic Press
- Links To Paper
- 1st Link
- 2nd Link
- Bibtex format
- @Article{EDI-INF-RR-0406,
- author = {
Craig Robertson
and Robert Fisher
},
- title = {Parallel Evolutionary Registration of 3D Data},
- journal = {Computer Vision and Image Understanding},
- publisher = {Elsevier},
- year = 2002,
- volume = {87},
- pages = {39-50},
- doi = {10.1006/cviu.2002.0981},
- url = {http://www.sciencedirect.com/science?_ob=MImg&_imagekey=B6WCX-47CYFMR-4-1&_cdi=6750&_user=809099&_orig=browse&_coverDate=07%2F31%2F2002&_sk=999129998&view=c&wchp=dGLbVlb-zSkWb&md5=8b7dc7deda19b9d9caf5ba3a743f4613&ie=/sdarticle.pdf},
- }
|