首页|AI技术赋能遥感卫星数据在轨处理

AI技术赋能遥感卫星数据在轨处理

扫码查看
原文链接
  • 万方数据
  • 维普
遥感卫星数据在轨智能处理已成为遥感领域的前沿技术,是实现"星数天算"的有效技术途径.本文综述了当前国内外主流的数据在轨智能处理系统发展现状,分析了行业面对的问题,提出了天地一体的智慧星座系统架构,构建了一种集固有处理、信息提取、知识生成功能于一体的卫星数据在轨智能处理系统,探索了利用人工智能AI模型及算法以及星地一体化处理系统实施在轨信息提取和知识生成的方法.为验证上述系统框架和方法,利用高光谱卫星数据进行了船舶检测和林火火点检测及预警的仿真试验.试验结果表明该系统框架的总体设计是可行的,提出的在轨信息提取和知识生成的方法是有效的.提出的系统架构为卫星在轨信息提取及知识生成并最终赋予卫星在轨思考能力,提供了一种可行的技术路线.卫星在轨思考的能力取决于在轨处理系统的架构设计、信息提取和知识生成的AI模型和算法,而算法的好坏将依赖卫星数据地面计算中心对其的不断优化和长期的迭代.无疑,卫星在轨处理系统将对卫星地面系统的架构和算力提出更加苛刻的要求,长远来看,这将助力"星数地算"的快速发展.
AI Technology Empowers In-orbit Processing of Remote Sensing Satellite Data
The in-orbit intelligent processing of remote sensing satellite data has become a cutting-edge technology in remote sensing science&engineering,and it is an effective technical approach to implement"data of remote sensing satellite processed at space".In this paper,the state of the art of mainstream in-orbit intelligent processing systems of remote sensing satellite in the domestic and international industries is summarized.The problems faced by the industry are analyzed.An intelligent satellite constellation system architecture integrated with satellite-terres-trial systems is proposed.An in-orbit intelligent processing system of remote sensing satellite is constructed that in-tegrates inherent routine processing module,information extraction module and knowledge generation module.The methods of implementation of in-orbit information extraction and knowledge generation are explored by utilizing AI models and algorithms and integrated satellite-terrestrial processing system.To validate the above system framework and methods,the simulation experiments on ship detection,forest fire point detection,and early warning using hy-perspectral satellite data are conducted.The results analysis show that the overall design of the system framework is feasible,and the proposed methods for in-orbit information extraction and knowledge generation are effective.The proposed system architecture can provide a feasible technical solution for satellite in-orbit information extraction and knowledge generation,ultimately endowing satellites with in-orbit thinking capabilities.The ability of satellites to think in-orbit depends on the in-orbit processing system architecture design,AI models and algorithms of infor-mation extraction and knowledge generation.The quality of the AI algorithms will depend on the continuous optimi-zation and long-term iteration executed by the ground data computing center.Undoubtedly,the satellite in-orbit processing system will put forward more stringent requirements for the architecture and computing power of the sat-ellite ground system.In the long run,this will help the rapid development of"data of remote sensing satellite pro-cessed at ground data computing center".

remote sensing satellitein-orbit processingintelligent satellite constellationAI models and algo-rithmsintegrated satellite-terrestrial processing systeminformation extractionknowledge generationdata of re-mote sensing satellite processed at spaced

王家耀、颜军、吴佳奇、邓剑文、董文岳、张强、龚永红、蒋晓华、张建辰、胡胜华

展开 >

河南大学 地理与环境学院,河南 郑州 450046

河南省时空大数据产业技术研究院(河南大学),河南 郑州 450046

珠海航宇微科技股份有限公司,广东 珠海 519080

青岛科技大学 数据科学学院,山东 青岛 266061

广东省国土资源技术中心,广东 广州 510062

展开 >

遥感卫星 在轨处理 智慧星座 AI模型和算法 星地一体化处理系统 信息提取 知识生成 星数天算 星数地算

2024

10.3969/j.issn.1673-6338.2024.06.001

测绘科学技术学报
信息工程大学科研部

测绘科学技术学报

影响因子：0.594
ISSN：1673-6338
年,卷(期)：2024.40(6)