Research On 3D Reconstruction Of Moon Surface And Obstacle Recognition Method

Moon stones and moon pits,large or small,are scattered on the moon’s surface for complex lunar terrain environment.In carrying out scientific exploration activities,a lunar rover needs not only basic traveling,turn sampling,but also basic artificial intelligence such as obstacle identification,path planning,and obstacle avoidance.Therefore,exploring how to make the lunar rover have environmental awareness is one of the key technologies in lunar exploration.Binocular stereo vision system as a lunar rover’s main environmental awareness system can simulate the human eye to identify the surrounding environment,thereby helping the lunar rover to carry out better exploration mission.The research of this paper is based on the background of the first "Jade Rabbit" Lunar Rover in our country.The main contents are as follows:First,the binocular stereo vision calibration algorithm is studied.On the erected binocular stereoscopic vision platform,the calibration algorithm of Matlab calibration and VS + Opencv is studied.The binocular stereoscopic vision calibration interface is designed,and the calibration results of the two are compared and analyzed.Experimental results show that the above two calibration methods have achieved higher accuracy.Secondly,the method of recognizing lunar obstacles(moon rock and moon crater)is studied.In this paper,mainly studied and analyze Otsu and two-dimensional maximum inter-class variance method,and then introduce Particle Swarm Optimization(PSO)based on them to get the improved two-dimensional maximum inter-class variance method.The comparative experiments were carried out using the "jade rabbit" scientific data and the ground simulation experiment photographs respectively,and the applicability of the algorithm in the lunar environment was verified.Thirdly,the binocular stereo matching algorithm is studied.The sparse matching algorithm and adaptive weight dense matching algorithm based on SURF feature points are respectively studied.Among them,the maximum parallax and the minimum parallax were calculated by using the sparse matching algorithm,which provided the parallax search range for the subsequent improved dense matching.At the same time,the applicability of the stereo matching algorithm was verified by the ground simulation experiment images.Finally,the three-dimensional reconstruction algorithm of lunar environmentis is studied.Based on binocular stereo vision system,the construction and display of virtual environment with VS,OpenCV and OpenGL are carried out to realize the three-dimensional reconstruction of lunar environment.