基于机器视觉的天基动目标自主监视技术

未来遥感卫星的分辨率将越来越高,覆盖范围越来越广,观测数据越来越多,现有的人工监视已经远远不能满足需求，因此监控的“智能化”变得越来越迫切。天基动目标自主视频监视技术是计算机视觉领域的研究热点，涉及到图像处理、模式识别以及人工智能领域，研究内容丰富，应用领域广泛。从三个模块对动目标自主识别关键技术进行了归纳总结，介绍了关键技术算法，并对典型的算法进行了优缺点分析，给出了相对鲁棒性强、实时性高、检测效果好的技术路线。

Autonomous Monitoring Technology of Space-based Moving Targets Based on Machine Vision

With the increasing of the imaging resolution of remote sensing satellites in the future, their coverage range will become wider and wider and their observation data will increase greatly. The existing manual monitoring is far from meeting the demand. Therefore, `intelligent'' monitoring becomes more and more urgent. At present, the space-based independent video surveillance technology is a hot spot in the field of computer vision, which involves in image processing, pattern recognition and artificial intelligence. This research is rich in content and has a wide application range. The key independent recognition technologies of abnormal moving targets are summarized from three aspects. The key algorithms are presented. The advantages and disadvantages of some typical algorithms are analyzed. The technical route with excellent robustness, high real-time performance and good detection effectiveness is given.