The Obstacle Detection Technology Based On Binocular Stero Vision Research And Implementation

With the rapid development of robot technology,autonomous mobile robot has been widely used,and obstacle detection is the basis of autonomous navigation of autonomous mobile robot.In order to realize obstacle detection,this paper designs an obstacle detection algorithm based on binocular stereo vision system.This paper first introduces the basic technology of binocular stereo vision system and camera calibration method,and calibrates the camera used in this paper.This paper designs a stereo matching algorithm based on the new description of matching cost,and further designs a fusion algorithm for obstacle detection to improve the accuracy of obstacle detection.In this paper,the hardware and software experimental platform of binocular stereo vision system is built,and the experiment and result analysis of related algorithms proposed in this paper are completed on the experimental platform.Firstly,the principle of binocular stereo vision and camera calibration algorithm are introduced.This paper introduces the binocular stereo vision imaging model and various reference coordinate systems.By introducing the camera imaging model and binocular stereo camera measurement principle,the camera calibration algorithm,Zhang's plane calibration principle and Brown camera distortion model are further introduced.The camera calibration experiment platform was built.Aiming at the binocular stereo camera used in this paper,the camera calibration algorithm was used to complete the calibration of binocular camera,and the calibration parameters of the camera were obtained.Secondly,binocular vision stereo matching algorithm is studied.Before applying stereo matching algorithm,it is necessary to correct the pole-line of the images taken by the left and right cameras according to the camera calibration parameters.This paper introduces a common pole-line correction algorithm to complete the pole-line correction of left and right camera images.In view of the poor matching effect at depth discontinuities of existing stereo matching algorithms and the fact that feature extraction is easily affected by occlusion,light and repeated texture,this paper designs a local stereo matching algorithm based on the new matching cost description.The local stereo matching algorithm designed in this paper uses a variety of image feature information description in the construction of matching cost to realize the complementary advantages of various image features,thus reducing the mismatching degree and improving the accuracy of the algorithm.Finally,this paper studies the segmentation of obstacles using the three-dimensional spatial information of images obtained by binocular stereo vision system.Firstly,according to the obstacle detection environment,this paper designs a rule of obstacle detection using three-dimensional space information,and conducts obstacle detection.To rely on the image segmentation result in three dimensional space information for obstacles of the segmentation accuracy is not high,the obstacles in this paper,further design a segmentation algorithm using ultra pixel and image three dimensional space information fusion detection algorithm to detect obstacles,the accuracy of this method has greatly increased the obstacle detection.