摘要
针对以往土地利用/土地覆盖分类数据应用,当前遥感智能解译样本库建设过程中与分类系统有关的问题,对遥感智能解译样本分类体系应有统一认识,可通过构建适于多尺度遥感影像的土地资源分类体系,实现统一分类体系下不同来源遥感分类数据的集成和有效对比。本文依据国土空间规划和治理新理念,地物类型在遥感影像上的可分性,采用自上而下方法构建土地资源多尺度遥感智能解译分类体系。该分类体系兼容土地利用与土地覆盖,包含9个一级类型,41个二级类型和43个三级类型,总计74个不同细分类型,比当前深度学习样本库的地物类型更加全面,并且采用开放框架和分层设计,支持类型扩展。该分类体系有助于对土地资源进行统一调查监测。
In view of the main scientific problems related to the classification system in the application of land use/land cover classification data in history and the construction of sample databases for remote sensing intelligent interpretation at present,it is considered that there should be a unified understanding of the classification system for remote sensing intelligent interpretation,because a unified land resource classification system that is suitable for multi-scale spatial resolution remote sensing image interpretation is very conductive to the integration and effective comparison of interpreted data from different producers.Based on the new concept of territorial spatial planning and governance,and the separability of feature types on remote sensing images,a top-down approach was adopted to construct a multi-scale remote sensing intelligent interpretation classification system for land resources.The classification system is compatible with land use and land cover,and contains 9 primary types,41 secondary types and 43 tertiary types,totaling 74 different subtypes,which are more comprehensive than the current sample databases for deep learning,and adopts an open framework and hierarchical design to support type expansion.The proposed classification system can promote the national unified survey and monitoring of land resources.
作者
徐进勇
汪潇
张增祥
孟瑜
张兵
XU Jinyong;WANG Xiao;ZHANG Zengxiang;MENG Yu;ZHANG Bing(Aerospace Information Research Institute,Chinese Academy of Sciences,Beijing 100101,China)
出处
《地理信息世界》
2022年第5期112-117,124,共7页
Geomatics World
基金
国家重点研发计划课题(2021YFB3900503)。
关键词
土地资源
遥感
智能解译
分类体系
land resources
remote sensing
intelligent interpretation
classification system
