Star identification algorithm for un-calibrated star sensor cameras

When optic parameters of star sensor are not accurate, identification rate of most star identification algorithms falls sharply. A new identification algorithm is presented for solving the problem. Rate distance is introduced as the radial pattern and angle as cyclic pattern. The patterns do not require any optic parameter such as the focal length, and the optical axis offsets. They are decided only by stars' location on the image. Two patterns are used for initial match firstly, and FOV (Field of View) restriction is used for accurate match. To speed up the search process, fast grouping method for candidate stars is proposed. Simulation is done to verify the performance and the result shows that the average runtime is short and identification rate reaches 97% without using optical parameters.