Research Of3D Reconstruction Of Indoor Scenes Based On Laser And Vision Of Mobile Robot

3D reconstruction is the process of getting the depth information of the scene using the depth sensors and then building the3D model of the scene. Usually we equip depth sensors on a mobile robot, and when the robot roaming in the scene, the3D information of the scene is obtained, and finally we can build the3D model. The laser has advantages of accuracy and is affected small by the external environmental changes. But it has sparse data points and the disadvantage of a low scanning frequency. The visual images have rich details and large amount of information, but it is sensitive to changes in ambient light and difficult to match the weak texture. Laser and visual integration can make up their own shortcomings. So it has important theoretical research value and broad application prospects.This paper researches the key technologies of3D reconstruction based on the two-dimensional laser sensor and stereo camera and conducts experimental verification. In this paper, the main research content includes:(1)The paper mainly analyzes the research status of three-dimensional reconstruction of mobile robot at home and abroad, points out the significance of this study, then proposes the overall program of the three-dimensional reconstruction based on laser and binocular stereo vision, and build a mobile robot platform.(2)The paper researches the localization algorithm of the mobile robot based on binocular stereo vision, and briefly introduces the feature point extraction, the feature point matching and motion estimation algorithm.(3)The paper proposes a new disregard field overlapping joint laser vision sensor calibration method. The two have the same field of view by the plane mirror reflection. By changing the position and gesture of the triangle calibration plate, we select a sufficient number of observations, using the Gauss-Newton iterative method to estimate the transformation matrix between the two-dimensional laser scanner and stereo camera.(4)We obtain the model of three-dimensional reconstruction of the scene based on binocular vision. Next three-dimensional grid reconstruction uses Delaunay triangulation method, and obtain a realistic three-dimensional model of the indoor environment.The above algorithms are realized on the platform ubuntu10.04, OpenCV, QT4.0and OpenGL, test on the mobile car platform, and finally proved valid by experimental results.