Novel World Representation for Stereopsis Correspondence

R. Mayoral, M.J. Pérez-Ilzarbe, and G. Lera (Spain)

Keywords

Computer vision, stereo vision, correspondence problem,world representation

Abstract

We present a novel world representation that allows us to find solutions to the stereopsis correspondence problem as a minimization of a defined correspondence error, without having to add penalizing terms to the function to be opti mized. Instead of this, restrictions arising from supposi tions done about the world are embedded into a model of the solutions and the structure of the algorithm. Using this representation, we prove that it is possible to obtain suitable solutions using exhaustive search to minimize a correspon dence error. We test the algorithm and evaluate the results using several publicly available test scenes.

Important Links:



Go Back