Research On Key Technologies Of Visual Pose Measurement System With Large Vision Range For Non-cooperative Target

In on-orbit servicing, offensive and defensive combat, as well as space garbage disposal tasks, it need to measure the relative pose between two objects. Because measurement distance from far and near span is large, measurement target cannot be installed corresponding cooperative template, therefore, it need pose measurement with large vision range for non-cooperative target. Vision pose measurement system gets wide attention because of small system size, low power consumption, and real-time image. However, because Non-cooperative targets prior knowledge is few, pose measurement accuracy becomes low when measuring distance is farther, vision measurement system cannot meet the practical requirements. These problems become research difficulties, that is, in case of less prior knowledge to achieve pose measurement and when measuring distance is farther to improve measurement accuracy. For these difficulties, this paper mainly study the following aspects. Details are as follows:For non-cooperative target pose measurement, this paper proposes binocular vision method of measuring pose based on perpendicular lines. This algorithm only needs to know the vertical relation of two lines on the measured object, and solve the relative pose of the measured object using the vertical relation and the four element method. Simulation results show that under the influence of 0.1 and 0.5 pixel standard deviation of the image noise, the pose measurement algorithm has good measurement accuracy in 0.3~25 meters.To make the pose measurement accuracy to meet the requirements of large visual range, it need to accurately calibrate the camera’s parameters. In the wide-angle lens distortion correction, this paper proposes vanishing point collinear constraint to improve accuracy of distortion correction based on cross ratio invariance. Firstly, this algorithm uses the central region image with lower distortion to calculate other ideal image point based on collinear cross ration invariability, then, uses vanishing points collinear constrain algorithm to optimize the calculation of the image points, and ultimately improve the accuracy of distortion correction.Camera’s intrinsic parameters and extrinsic parameters errors coupling make parameters calibration accuracy is different when the pose between camera and template is different. In order to improve the accuracy of the camera calibration parameters, it need to reduce the coupling errors influence of calibration parameter accuracy. This paper proposes decouling algorithm of camera’s intrinsic parameters and extrinsic parameters. This algorithm is based on the linear coupling between camera’s intrinsic parameters and extrinsic parameters calibration errors, and using multiple calibration result by translation template, to correct optical center point, camera focal distance and translation vector calibration errors. Experiment results show that the decouling algorithm reduces the coupling errors influence of calibration parameter accuracy, and improves the calibration parameter accuracy.These proposed algorithm has been validated in the aerospace satellite pre-research hardware platform. This platform includes two cameras, the measured target, embedded image processing hardware and ground verification equipment. The pose measurement algorithm is written in the embedded hardware. Ground verification equipment uses optical dark room to simulate space environment, and uses six degree of freedom motion system to detect the accuracy of the vision measurement system. Finally the experiment data show that the accuracy of vision measurement system meets the design requirements, thus verified the correctness of the algorithm proposed in this paper.