A Fast Vision-Based Localization Algorithm for Spacecraft in Deep Space

Star light navigation can provide the current attitude and position of the spacecraft in deep space. However, the accuracy of stellar-inertial attitude determination is degraded because of star image smearing under high dynamic condition. To solve this problem, two key work, including accuracy star extraction and fast star identification, should be done. In this paper, we bring interpolation algorithm into contiguous area pixel searching for star extraction, and get sub-pixel coordinate information of the star points. In addition, a divisional method is proposed to improve star identification algorithm speed based on Hausdorff distance. The simulation results show that work not only has accuracy identification rate but also has better recognition speed. It was used successfully in the actual projects.