遥感卫星在轨图像智能处理设计与验证  

Design and Verification of On-board Intelligent Processing Technology for Remote Sensing Satellites

作　　者：肖化超[1] 王鹏[1] 张建华[1] 师勇 孙钰林[1] 郭琪[2] 杨芳[2] XIAO Huachao;WANG Peng;ZHANG Jianhua;SHI Yong;SUN Yulin;GUO Qi;YANG Fang(China Academy of Space Technology(Xi’an),Xi’an 710100,China;DFH Satellite Co.,Ltd.,Beijing 100094,China)

机构地区：[1]中国空间技术研究院西安分院,西安710100 [2]航天东方红卫星有限公司,北京100094

出　　处：《航天器工程》2022年第2期43-48,共6页Spacecraft Engineering

基　　金：高分辨率对地观测国家重大专项基金(GFZX04061502)。

摘　　要：传统遥感卫星星上成像、地面处理的方式制约了用户信息获取的效率,为提升遥感卫星在轨处理的智能化水平,文章针对多项在轨智能处理技术,设计并研制了星载智能处理器,突破了基于深度学习的目标检测、在轨云检测实时处理和多层级在轨重构等关键技术,并基于某遥感卫星对软硬件设计进行了在轨验证,测试结果表明:在轨智能处理可显著提高用户信息获取的有效性和时效性,倍增卫星对地观测效能及服务能力,可应用于后续遥感卫星设计。In order to improve the intelligent level of on-board data processing for remote sensing satellites,a research of several on-board intelligent processing technologies has been carried out.An on-board intelligent system is designed and developed,which breaks through key technologies such as on-board target detection based on deep learning,real-time cloud detection and multi-layer reconstruction.Software and hardware of the intelligent processing system has been verified on-orbit.The results show that the on-orbit intelligent processing can significantly improve the efficiency and service capability of satellite earth observation,and can be applied to subsequent remote sensing satellites.

关 键 词：遥感卫星 在轨智能处理技术 在轨重构 深度学习 

分 类 号：TN919.3[电子电信—通信与信息系统]