Title :
Using motion from orthographic projections to prune 3-D point matches
Author :
Chen, Homer H. ; Huang, Thomas S.
Author_Institution :
AT&T Bell Labs., Holmdel, NJ, USA
Abstract :
The authors presented a new method using motion constraint to prune 3-D point matches. The method discards noisy z coordinates of points and uses only x and y coordinates to compute motion. This method is specifically designed to resolve the matching ambiguity due to the use of large error tolerances in rigidity tests to account for depth errors. A least-squares solution of the motion problem is derived. Results show that this method is able to detect false matches more effectively than the traditional method which uses full 3-D coordinates
Keywords :
computer vision; computerised pattern recognition; computerised picture processing; errors; least squares approximations; 3D point matches; depth errors; error tolerances; false matches; least-squares solution; matching ambiguity; motion constraint; noisy z coordinates; orthographic projections; pruning method; rigidity tests; Approximation algorithms; Computer errors; Data mining; Error correction; Feature extraction; Layout; Motion estimation; Quaternions; Testing; Uncertainty;
Conference_Titel :
Visual Motion, 1989.,Proceedings. Workshop on
Conference_Location :
Irvine, CA
Print_ISBN :
0-8186-1903-1
DOI :
10.1109/WVM.1989.47121