Improving the efficiency and effectiveness of estimating the ego motion of a stereo camera

Estimating the ego motion is a key problem for a mobile robot, its efficiency and precision are hard to be balanced. In this paper, a fast and accurate approach is proposed to estimate the ego motion of stereo camera. Firstly, two images from left and right cameras are rectified. secondly, a index is introduced to speed up the matching process to find corresponding point from right image for each point from left image. finally, a RANSAC structure based method is proposed, which eliminates the outliers before estimating the ego motion. Methods developed in this paper are validated using both real and synthetic data.