Research On Space Object Measurement Base On Stereo Vision

With the continuous development of space technology,there are some operational tasks such as approaching and capturing space objects in space.Measuring the attitude of a space target is a prerequisite for completing these tasks.However,objects in space are generally not cooperative and can not provide a priori information for identification.Therefore,it is more difficult to accurately measure the relative pose of such objects.At present,the use of stereo vision technology to solve this problem has become the focus of research in relevant fields at home and abroad.Based on the measurement of stereo vision in the space target object,this dissertation launches the research on related theories and technologies.First of all,through the theoretical analysis of algorithms in digital image processing,the improved GrabCut algorithm and the improved Hough mutation algorithm are proposed,and the two algorithms are combined to identify the image features of the non-cooperative target.The simulation experiment using the space object image proves that the proposed image preprocessing method can satisfy the requirement of solving the space object object relative to the pose of the space robot.Secondly,research on the calculation method for the orientation of non-cooperation goals.By establishing a plurality of spatial coordinate systems,the objects in space are linked together,and the transformation relations between the various coordinate systems are analyzed.Using the camera's imaging model and the SGM semi-global stereo matching algorithm,the 3D coordinates of the key points are calculated to complete the 3D reconstruction.At the same time,the object coordinate system of the different types of target images is established based on the detected image features of the object,and its relative position and attitude matrix in the world coordinate system are obtained.Then,the Eye-in-hand hand-eye system is selected as the experimental model,and the robot end-tool coordinate system is setted.The UR5 robotic hand-eye system for experiment is set up.The Tsai hand-eye calibration algorithm is improved to calculate the hand-eye relation matrix with higher accuracy.Finally,the experimental verification scheme of the pose measurement of the target object is designed,and the common non-cooperative spacecraft in the space is selected as the identification object,and the ground experimental platform is set up.Through multiple sets of tests,experimental data are obtained and analyzed.The results prove that the proposed scheme and algorithm have higher accuracy and robustness.