遥感智能信息处理的发展及技术前景

Development and Technical Prospect of Remote Sensing Intelligent Information Processing

摘要:

遥感信息提取技术虽不断推陈出新，但在智能化、精准实用性方面始终存在巨大的瓶颈问题，有必要围绕遥感智能计算和信息提取这个发展主题进行总结和讨论。从“机理—尺度—数据—智能”4个层面，逐步就遥感信息提取与定量反演路径的发展融合、基于像素和面向对象不同处理单元模式、时空谱数据融合、遥感解译的智能化因素四方面进行剖析，从而提出未来“数据获取知识”和“知识引导数据”双向驱动、遥感大数据和地学知识图谱相融合的遥感智能计算架构，尝试推动遥感科学从经典向现代化的跃迁。

关键词:遥感智能计算;信息提取;智能解译;面向对象;深度学习;时空融合;遥感大数据;地学知识图谱

Abstract:

Although the remote sensing information extraction technology is constantly being innovated， there are still huge bottlenecks in terms of intelligence， precision， and practicality. Therefore， it is necessary to conduct a comprehensive summary and discussion on the development topic of remote sensing intelligent computing and information extraction. From the four levels of “mechanism-scale-data-intelligence”， this paper gradually discusses the development and fusion of remote sensing information extraction and quantitative inversion paths， the different processing unit modes based on pixels and object-oriented， the spatial-temporal spectral data fusion， the intelligent factors of remote sensing interpretation. It proposes a future-oriented remote sensing intelligent computing architecture which is driven by two-way “data acquisition knowledge” and “knowledge-guided data”， and integrates remote sensing big data and geoscience knowledge maps， trying to promote the transition of remote sensing science from classics to modernization.

Key words:remote sensing intelligent computing;information extraction;intelligent interpretation;oriented object;deep learning;spatial-temporal fusion;remote sensing big data;geoscience knowledge graph