Under high dynamic and long exposure conditions, the number of recognized stars on motion-blurred star images decreases, thereby degrading the attitude accuracy of star sensors. To improve the attitude accuracy, a restoration method based on multi-frame superposition, which focuses on the noise removal and quality of restored star images, is proposed for a star sensor. During each short exposure time, the corrected coordinate variation of the same star spot between adjacent star images is determined using a motion recursive model. Subsequently, the corrected star spot region is obtained, and the noise is removed. A restoration algorithm based on multi-frame superposition is proposed, taking the time consumption and quality of restored star image considered simultaneously. Simulation results indicate that the proposed restoration method based on multi-frame superposition is effective in removing noise and improving the quality of restored star images. The star recognition rate in simulation experiments verifies the advantages of the proposed method.

在高动态和长时间曝光条件下，运动模糊的恒星图像上可识别的恒星数量会减少，从而降低了恒星传感器的姿态精度。为了提高姿态精度，提出了一种基于多帧叠加的恒星传感器图像复原方法，该方法着眼于噪声去除和恒星图像质量的提高。在每个短时间曝光期间，使用运动递归模型确定相邻恒星图像之间相同恒星点的校正坐标变化。随后，获得校正的星点区域，并且去除噪声。提出了一种基于多帧叠加的复原算法，同时考虑了复原星图的时间消耗和质量。仿真结果表明，所提出的基于多帧叠加的图像复原方法可以有效地去除噪声，提高图像复原质量。仿真实验中的恒星识别率证明了该方法的优越性。