【摘要】 目标跟踪是空间技术领域中非常重要的研究分支，针对一些运动轨迹复杂的天基空间动目标，提出基于可见光图像的动目标跟踪算法。首先，提出一种基于多窗口灰度统计迭代的自动曝光算法，有效提升跟踪目标的成像质量；其次，提出联合滤波算法对图像序列进行预处理，以保留更多目标细节；在跟踪过程中，通过自适应阈值分割及轨迹关联预测对空间动目标进行持续关联跟踪，以进一步提高目标跟踪的鲁棒性。实验结果表明，对于深空背景下运动轨迹复杂的空间动目标，所提算法的跟踪延迟时间不超过40.3 ms，跟踪偏移角度精度达到10.0′。在实际飞行工程实践中能同时应对100个以上干扰物，并与地面测试结果基本一致，表明其能够进行鲁棒的目标跟踪。更多还原

【Abstract】 Target tracking is a very important branch in the space technology research, to address the challenges of tracking space-based moving targets with complex trajectories, a visible light image-based moving target tracking algorithm is proposed. First, an algorithm for automatic exposure based on multi-window grayscale statistical iteration is proposed, which can effectively improve the imaging quality of tracked targets. Second, a joint filtering algorithm is proposed to preprocess image sequences, preserving more target details. During the tracking process, adaptive threshold segmentation and trajectory association prediction are employed to achieve continuous and robust tracking of moving targets in space. Experimental results show that for space moving targets with complex motion trajectories in deep space background, the tracking delay time is no more than 40.3 ms, and the tracking offset angle accuracy is 10.0′. In actual flight engineering practice, the proposed algorithm can simultaneously handle more than 100 interfering objects, and the results are basically consistent with ground testing, indicating that it can perform robust target tracking.更多还原