基于多源遥感数据的数字林场信息提取与系统构建  

Information extraction and system construction of digital forest system based on multisource remote sensing data

作　　者：谢东辉 李益 周坤 张智祥 金霖 阎广建 穆西晗 李文航 冯功耀 XIE Donghui;LI Yi;ZHOU Kun;ZHANG Zhixiang;JIN Lin;YAN Guangjian;MU Xihan;LI Wenhang;FENG Gongyao(State Key Laboratory of Remote Sensing Science,Jointly Sponsored by the Institute of Remote Sensing Applications of Chinese Academy of Sciences and Beijing Normal University,Beijing Normal University,Beijing 100875,China;Aerospace Information Research Institute,Chinese Academy of Sciences,Beijing 100094,China)

机构地区：[1]北京师范大学地理科学学部遥感科学国家重点实验室,北京100875 [2]中国科学院空天信息创新研究院,北京100094

出　　处：《遥感学报》2024年第5期1281-1294,共14页NATIONAL REMOTE SENSING BULLETIN

基　　金：国家自然科学基金(编号:42090013,42071304);国家重点研发计划(编号:2020YFA0608701,2022YFB3903304)。

摘　　要：数字林业是林业现代化的基础。本研究以河北省塞罕坝机械林场为研究区,利用遥感科学与技术,基于机载数据(激光雷达、高分辨率CCD影像和高光谱影像),研究单木分割和单木分类方法,实现了机载飞行区域270 km2范围内所有单木的信息(位置、树高、冠幅、树种等)提取。结合地面样方调查数据,单木分割匹配精度可以达到0.6以上,研究区内4种典型树种单木分类精度均能达到97%以上。基于地基激光雷达扫描数据,研究以单木为单位,通过枝干重建和叶片添加,实现单木三维模型重建和场景重建。该方法有利于从器官(枝干、叶片)—单木—区域多尺度分析森林结构特征。在此基础上,本研究探索了以单木为基础的林业数字化方法,并结合WebGIS技术开发了数字林业系统,初步实现了在研究区内单棵树为单位的数据存储、管理、查询、分析和可视化等功能,为后续的森林精准化经营管理和决策规划提供重要的基础数据和系统支持。Digital forestry is the foundation of forestry informationization and modernization,which can help management with scientific programming,scientific management,and scientific decision.With the development of 3S(GIS,RS,and GNSS)technology,improving the level of digital forestry becomes feasible,especially for the monitoring scale from forest-compartment level to single-tree level by using high spatial-and spectral-resolution remote sensing data and Light Detection And Ranging(LiDAR).This paper aims to develop a forestry digital system by combining the advantages of remote sensing and GIS technology,which can improve the precision management ability of forest farm.Saihanba Mechanical Forest Farm in Hebei Province,China is taken as the research area,and airborne data,including LiDAR,CCD image,and hyperspectral image,are used to study single tree segmentation and classification methods.First,an algorithm combined water shed and Connection Center Evolution methods is applied to segment each single tree based on the airborne LiDAR scanner(ALS).Then,the random forest algorithm is applied to classify the tree species of each single tree based on the combination of the 50 characteristics from ALS,CCD,and hyperspectral image.The 3D model reconstruction is also examined by using terrestrial laser scanning data,with the procedure of branch reconstruction and leaf addition.Forest scenarios are reconstructed based on each single tree model.Finally,WebGIS technology is used to develop a forestry digital system with the functions of data storage,management,query,analysis,and visualization.More than 9706 thousand single trees within the airborne flight area of 270 km2 are segmented and classified.Combined with the survey data of ground quadrats,the single tree segmentation accuracy can reach over 0.6 matching rate,and the single tree classification accuracy of four typical tree species in the study area can reach more than 97%.The information of single tree,including location,tree height,crown breadth,and the structures of branch

关 键 词：遥感 数字林业 激光雷达 高光谱 单木分割 分类 三维重建 

分 类 号：TP79[自动化与计算机技术—检测技术与自动化装置] P2[自动化与计算机技术—控制科学与工程]