An improved star identification method based on neural network

In order to increase the star identification speed and recognition rate of star sensor, an improved rapid star identification method based on neural network (NN) is presented. The proposed method is composed of three levels, including the coarse classification of navigation stars, star pattern recognition based on NN and the final validation of recognition results. The angular distance of characteristic triangle is employed for coarse classification to active the subnets for star pattern recognition. Then the star pattern obtained by the grid method for the main star is sent to the corresponding subnets for the star pattern identification. At last, the identification results of the active subnets are validated to obtain the only recognition result. The experimental results show that, compared with traditional triangle identification method, the proposed method has higher accurate recognition rate, lower redundancy and better robustness.