基于多源卫星联合观测的舰船目标信息融合技术研究

With the increase of the number and the variety of satellites, multi-source satellite joint observation has become the development trend of space-based maritime surveillance. Faced with complex and variable multi-source satellite observation information, how to effectively deal with more target information and improve the capability of ship surveillance has become an urgent problem to be solved. This project focuses on the information fusion of multi-source satellite observation for ships and the specific research includes: (1) Aiming at the problem of information fusion of ships under the joint observation of multi-source low-orbit satellites, studying the multi-source heterogeneous sensors fusion in detection, location and recognition in complex environment, breaks through the fusion technology at the recognition level; (2) Aiming at the problem of information fusion of ships under the joint surveillance of multi-source high-orbit optical satellites, studying ship target tracking, track segment correlation and fusion in complex environment, and the related theories of multi-satellite track correlation and fusion under systematic error, breaks through the fusion technology at the tracking level; (3) Aiming at the problem of information fusion under the joint observation of high-orbit and low-orbit imaging satellites, studying the track mining, track prediction under long time interval combined with the big data, and multi-feature multi-level target association algorithm, breaks through high-level fusion techniques such as behavior analysis and dead reckoning.

随着卫星数量的增加以及种类的丰富，多源卫星联合观测已经成为天基海洋监视的发展趋势。面对复杂多变的多源卫星观测信息，如何有效地处理来获取更多的目标信息，提高舰船目标的监视能力，已经成为当前迫切需要解决的难题。本项目着眼于多源卫星舰船目标观测信息融合问题，具体研究内容包括：（1）针对多源低轨卫星联合观测下的舰船目标信息融合的问题，研究复杂环境下多源异质卫星传感器的舰船目标检测、定位与识别理论，突破识别层级上的融合技术；（2）针对多源高轨光学卫星联合监视下的舰船目标信息融合问题，研究复杂环境下舰船目标跟踪、航迹片段关联与融合，以及系统误差下多星的航迹关联与融合等相关理论，突破跟踪层级上的融合技术；（3）针对高轨与低轨成像卫星协同观测下的舰船目标信息融合问题，结合航迹大数据研究航道挖掘、长时间间隔下的航迹预测以及基于多特征多层次的目标关联算法，突破行为分析、航迹推演等高层级的融合技术。

多源卫星观测信息下的舰船目标感知是天基海洋监视的重要发展方向，能够获得舰船目标更多的信息，在海上交通管理、海洋权益保护等方面具有重要意义，研究多源卫星舰船目标观测信息融合技术具有重要的理论意义和实用价值。本项目针对多源卫星舰船目标观测信息融合问题，主要研究内容包括：（1）针对多源低轨卫星联合观测下的舰船目标信息融合的问题，研究了基于深度学习的高分辨率光学卫星图像舰船目标检测、基于深度学习的可见光与红外卫星图像舰船目标融合检测，基于卫星图像与AIS信息的舰船目标联合定位等相关技术，提升了低轨卫星对舰船目标的感知能力；（2）针对多源高轨光学卫星联合监视下的舰船目标信息融合问题，研究了复杂背景下基于高轨低分辨率卫星图像序列的舰船目标快速检测与跟踪、航迹片段关联与连接等相关技术，提升了高轨卫星对舰船目标的感知能力；（3）针对高轨与低轨成像卫星协同观测下的舰船目标信息融合问题，研究了基于低轨卫星图像序列的舰船目标运动参数估计、基于高轨与低轨卫星的舰船目标点迹精准关联等相关技术，提升了高低轨卫星对舰船目标的融合跟踪能力。本项目执行期间，发表与录用学术论文7篇，授权国家发明专利2项，达到预期目标。本项目研究，进一步完善了天基海洋监视中多源卫星联合应用的理论基础，为相关软硬系统开发提供了技术支撑，具有很好的应用前景，部分算法已开展工程应用。