Comparison of Stereo Matching Algorithms for Mobile Robots

1 Abstract For navigation and obstacle avoidance in the field of mobile robots, perception and identification of the surrounding environment is necessary. Depth maps therefore provide an essential description of the world seen through cameras. This thesis will investigate with different types of stereo matching algorithms for calculating depth maps. The task is to build and to implement a system for generating depth maps on a mobile robot. Many stereo algorithms make use of the epipolar constraint, meaning that for a pixel in the left image the corresponding point in the right image lies on the same horizontal line, the epipolar line. This strong constraint is used to reduce the search space of the correspondence algorithms that calculates depth maps. To make use of the epipolar constraint the camera system needs to be calibrated first, to get the intrinsic and extrinsic camera parameters, in order to rectify images according to these parameters. The first chapters therefore deal with the necessary pre-processing steps to calibrate cameras and to rectify images.