Satellite Attitude Estimation Based on ISAR Image Sequence Interpretation

This paper presents a novel approach for the estimation of satellite attitude based on ISAR image interpretation. The solar panel of the satellite is segmented from the initial ISAR image using Pix2pixGAN network, and morphological method is used for edge extraction according to the parallelogram characteristics of the solar panel. According to the projection process of ISAR imaging, the optimization function for the 3D absolute attitude estimation of the satellite solar panel is established, and the attitude of the satellite is solved by iteration. The experimental results show the effectiveness of the proposed method.