State estimation of non-three-axis stabilized spacecraft based on double-station ISAR joint observation

State estimation of non-three-axis stabilized spacecraft is significant in space situational awareness tasks, such as on-orbit state analysis and collision warning. Traditional spacecraft state estimation methods based on Inverse Synthetic Aperture Radar (ISAR) images impose specific assumptions on the motion states of spacecraft and rely on long-term observation data to estimate state parameters. However, these methods exhibit strong dependency on the motion states and have certain limitations in practical applications. To address this issue, this paper proposes a state estimation method that does not depend on assumptions about the motion state. The method extends traditional ISAR observation to a double-station ISAR joint observation framework, leveraging dual-perspective measurements to decouple the relations between attitude parameters, motion parameters, and the projection morphology of spacecraft in ISAR images. A sequential estimation strategy for attitude and motion parameters is proposed: Firstly, by analyzing the double-station ISAR imaging geometry, the attitude parameters are estimated based on the invariance of the range projection vector. Then, an optimization function is constructed based on the cross-range projection morphology of the spacecraft to estimate the motion parameters. Moreover, the proposed deep multi-resolution keypoint extraction network substitutes traditional feature extraction methods. By autonomously extracting semantic features from ISAR images, it can effectively address the challenge of feature matching in double-station ISAR images. Experimental results validate the effectiveness and robustness of this method.