Digital Simulation Image Generation And Security Evaluation For Rendezvous And Docking

This paper comes from the project of the center of pattern recognition and intelligent system of Harbin Institute of Technology.In the project,the ascending spacecraft enters the orbit around the moon,and the orbiter adjusts its posture to carry out the rendezvous and docking procedure.This paper mainly simulates the process of docking(the distance of two spacecraft from 100 m to 0.3m)to generate digital simulation image sequences,and carries out feature extraction and target tracking for the simulation image sequence,and estimates the position and pose of the ascending device.Based on the two spacecraft structure,orbit data and relative pose,a digital simulation model is established.The imaging principle of the visible light camera on the simulator is studied,and the digital simulation image sequence is generated,and the simulation image database is established.For digital simulation images,background subtraction is applied to detect moving objects.The background subtraction method performs well in the simulated image,in which the simulator is far away from the camera,and obtains the foreground image containing the complete moving target.In the foreground image,the contour tracking is carried out.The contour tracking method is simple and efficient,and it can detect complete,closed and serialized contours.The contour is calculated by linear prediction method and the centroid is calculated.At the same time,the Canny operator is used to extract the edge information of the spacecraft and achieve the acquisition of contour and edge data in the simulation database.With the rising of the model approaching the visible light camera,the local features such as feature points and lines can be extracted from the simulation image.In this paper,SURF and ORB are used to extract feature points.Both SURF and ORB features have scale invariance and rotation invariance.The SURF feature has 64 dimensional descriptors,which is used as a feature point extraction method in simulation database,and is conducive to the application of machine learning method in subsequent experiments.ORB features faster computing speed.As a feature extraction method of real-time simulation,the matching speed is faster.At the same time,the probability Hough method is used to extract the straight line features and achieve the acquisition of straight line data in the simulation database.This paper selects significant ORB feature points,uses template matching and optical flow method to track feature points,and uses EPn P algorithm to calculate the rising position position.For the ORB features in the simulation image and the template,the homography matrix is obtained by RANSEC,and the salient feature points on the template are projected onto the target simulation image.Next,Lucas-Kanade is used to improve the optical flow method to track the salient features.The speed of optical flow method is faster than that of template matching.It can track the significant feature points moving along with the rising device and meet the real-time tracking requirements.According to the significant feature points,the EPn P algorithm is used to calculate the position and posture of the riser,and the real-time display on the target data 3D model is used to realize the security prediction system of rendezvous and docking based on the significant feature points.