Research on Binocular Stereo Vision System for Indoor Mobile Robot on Complex Environment
|Course||Mechanical and Electronic Engineering|
|Keywords||Binocular stereo vision Camera Calibration OpenCV Stereo matching Disparity map Obstacle map|
Chapter on binocular stereo vision technology Research , summarizes the main map created vision - based obstacle detection methods , and research at home and abroad binocular stereo vision for mobile robot navigation system as well as existing binocular stereo vision system products . Clarify the research significance and main content . Chapter II , the first linear pinhole model of camera imaging lens distortion : radial distortion and tangential distortion . Checkerboard flat image experiment , using OpenCV calibration parameters and structural parameters of the space inside and outside of the left and right cameras . And use the the Bouquet corrected image on a binocular epipolar rectification . Chapter III , the introduction of a stereo matching algorithm popular international criteria and standards of experimental images provide an objective basis for the judgment of the pros and cons of various stereo matching algorithm . The stereo matching algorithm is divided into two categories , local matching algorithm and global matching algorithm , winning Select local matching algorithm as well as the more popular dynamic programming method and graph cut method global matching algorithm . Correlation algorithm qualitative and quantitative assessment . Chapter IV , according to the application requirements of the mobile robot , key research area - based local matching algorithm to analyze the various factors that affect the matching results , the qualitative similarity measure the accuracy of the function and the size of matching window and quantitative analysis , in-depth study of image preprocessing and post-processing methods to improve the disparity map of stereo matching results to the box filtering multiresolution computing acceleration techniques applied to the implementation of a stereo matching algorithm , and to use OpenCV further improve the algorithm real-time. The fifth chapter analyzes the the ideal binocular stereo vision system , depth measurement error model , local map to remove surface noise to a reasonable height model . Integrated application and validation of the stereo matching algorithm used in the previous chapters and calibration results . The depth calculation results show that the measurement error can meet the actual needs of the use of the actual scene depth . The results of the three-dimensional reconstruction and raster map well to provide environmental information for mobile robots to lay the foundation for the future path planning .