High-orbit Non-cooperative Slow-rotary Space Object Attitude And Position Measurement

With the development of space technology,space manipulation technology developed fast.It is widely used in space junk cleaning,target acquisition,spacecraft maintenance and other fields.Target position measurement is the foundation of spatial operation technology.Position and orientation measurement technology based on Cooperative markers has been widely used in various countries.But only the spacecrafts designed for docking are equipped with cooperative markers.In order to expand the scope of space operation,it is necessary to study the position and attitude measurement technology of non cooperative object.Pose measurement of non cooperative targets plays a fundamental role in docking,capturing and repairing non cooperative targets.This paper focuses on high orbit non-cooperative slow-rotary unstable space object.Based on the existing sub vision measurement technology,the pose measurement technology for non cooperative targets is studied.The camera calibration,small target detection,target pose measurement and other related technologies are emphatically studied.The specific contents are as follows:Camera off-line calibration is the process of calculating the internal parameters of camera by calibrating the corners of the template in the image.Focusing on the accuracy of corner detection,a corner location algorithm based on gradient weighted fitting line is proposed.Based on the linearity of projection transformation,the position of the corner is calculated by calculating the position of the straight line.By keeping the fitting area on the fitting result,the accuracy of the 8 bit image with lower noise is better than 0.01 pixels in a reasonable and careful design.Even with noise,it is more accurate than common algorithms.As a result,the accuracy of offline calibration is increased by more than 40%.Aiming at the application of space target attitude measurement,an online calibration scheme for orbit camera is proposed.Thus,the measurement error caused by the change of camera parameters caused by the impact of launch,air pressure and gravity changes can be solved.Based on the main structure of the cube shape of the common satellite,the focal length and the main point of the camera are calculated by the vanishing point.In order to obtain the object location from a long distance,a method of detecting non cooperative small targets in long-range space based on multi-directional reconstruction is proposed.Through multi-target reconstruction and background enhancement based on information entropy,high SNR gain can be obtained at very low computational complexity.In addition,the algorithm can also get good results from natural and infrared images.It is difficult to match the feature points with the single space environment light source and the target rotation,which uses the intersection point of the edge line instead of the feature point to measure the position and position of the target.In order to get better edge line accuracy,a linear fitting based edge detection algorithm with O(n)complexity is proposed.Thus,a higher accuracy edge line can be detected with the speed of the existing straight line detection algorithm approaching or faster.The detection time is directly proportional to the number of edge points.For the object attitude measurement of the unknown geometric structure,the existing algorithm generally calculates the rotation matrix and the translation direction of the camera through the essential matrix,and then reconstructs the target on the scale,and then uses other algorithms to calculate the position and posture.The essential matrix is the only way to calculate the relationship between camera and target location when unknown geometric information is present.But it can only calculate the direction of translation,and the distance of translation is unknown,which limits the wider reference.In this paper,we propose an algorithm to calculate the ratio of the distance between the translation and the distance from the target to the camera through the intrinsic matrix.Thus,the position and attitude of the target can be calculated directly through the intrinsic matrix.Through simulation,the measurement error of attitude is less than 1 degree,and the measurement error of pose is less than 2% of target distance.