An Efficient Virtual Stereo-vision Measurement Method of a Space Non-cooperative Target

The pose (position and attitude) measurement of a space un-controlled target is very important for on-orbit service. However, there are no cooperative markers on these targets, so it is difficult to identify and measure of such targets. In this paper, an efficient VSVM (virtual stereo-vision measurement) method is proposed, which can recognize the natural objects with the circular and triangle features on the target, and largely reducing the computation load. Firstly, based on 3D reconstruction, the 3D point cloud feature of non-cooperative target is reconstructed. According to the 3D point cloud information, the relative pose of the satellite nozzle (circular center) and the triangular bracket (the center of the inscribed circle in the triangle) can be obtained accordingly. Secondly, a stereo-vision measurement method of circular features and triangular features is derived. Thirdly, the efficient VSVM method is used to solve the pose coupling equation and obtain the pose information, reducing the number of sensors and the calculation complexity; it also reduces the cost of space measurement. Finally, an experimental system is built and the validity of the algorithm is verified by the practical experiments. The experiment system is composed of two cameras, a FARO laser tracker, and satellite mockup. The experiment results demonstrate the effectiveness of the proposed method.