Stereo vision based floor plane extraction and camera pose estimation

Abstract

In this paper, we present a unified approach for floor plane extraction and camera pose estimation. A histogram based method is performed on the reconstructed 3-D points obtained by an onboard video-rate stereo vision system to extract candidate points of the floor plane and determine the pose of vision system simultaneously. The obstacle area is easily localized given the floor plane region. In order to improve reliability and accuracy of camera pose estimation results, the Least Median of Squares (LMedS) based fitting method is applied to estimate the floor plane parameters with the extracted candidate points. The precise pose of the onboard stereo vision system is directly acquired related to the floor plane parameters. Experimental results in real indoor environments are discussed and show the good performance. © 2009 Springer-Verlag Berlin Heidelberg.