Research On The Measurement Technology Of Feature Points Of Imitating Lunar Surface Environment Based On Stereo Vision

When using telecontrol technology to control lunar probe robot,the operator can not get real-time feedback information from the remote end because of the time delay in the long-distance communication between the earth and the moon.Virtual reality technology is an effective means to solve the problem of large time delay,but in the virtual reality technology needs to build the robot working three-dimensional environment,three-dimensional vision based construction technology is one of the main means used.Different from the ground environment,the lunar surface environment has single texture and high background similarity,and the traditional software-based 3D reconstruction method cannot meet the requirements under the existing technical conditions.In view of the above problems,under the condition that automatic modeling of 3D environment cannot be satisfied,this paper proposes a method of manual selection of feature points based on stereo vision and interactive.Through accurate 3D measurement of feature points,this paper studies the method to realize 3D construction of the imitating lunar surface environment.The main work contents of this paper include:(1)A binocular stereo vision camera model with parallel optical axis is designed,and Zhang Zhengyou calibration method is used to realize stereo calibration of the camera.Aiming at the problem that feature points are few and difficult to be extracted in the low texture environment of lunar surface,the histogram equalization is firstly carried out to enhance the image contrast,and then the edge feature is extracted,and then the feature points are selected on the extracted edges.The method can improve the efficiency of feature points selection.After comparing and analyzing three edge detection algorithms,Roberts,Prewitt and Canny,it is found that Canny edge detection algorithm has better effect.Therefore,feature point extraction technology based on Canny edge detection algorithm is adopted in this paper.(2)For the collected images with many textures,the SIFT algorithm based on correction and RANSAC was developed to realize the matching of left and right image feature points and the elimination of mismatched point pairs.In this algorithm,in order to eliminate the mismatched feature points generated by SIFT matching,stereo correction and RANSAC algorithm are used to remove the mismatched points,so as to improve the matching accuracy of the feature points in the left and right images.According to the principle of binocular vision triangulation,the sample image is used to simulate the lunar surface environment.The test results show that the method can effectively reduce the mismatching of feature points and improve the measurement accuracy and efficiency.(3)In order to effectively solve the problems that the feature points are not easy to distinguish and the background color is single in the low-texture images of the lunar surface,and to ensure the accuracy and reliability of the 3D construction,a 3D ranging system of interactive manual selection of feature point pairs is developed.In order to verify the validity of the system,based on texture feature less corridor floor tile experiment test,and verify the low effectiveness of texture,the background of a single environment selection,by verifying measure of feature points on the three dimensional coordinates,and comparing with actual measured value,found that measurement error is less than 1%,prove the validity of the development system and the related algorithm.