以机载激光雷达为核心的多源遥感巨树检测

Multi-source Remote Sensing Giant Tree Detection Method with Airborne LiDAR As Core

为探讨以机载激光雷达为核心的多源遥感巨树检测方法,准确、快速和高效地提取区域范围内巨树位置及信息,以有较大面积天然林分布、地形较复杂且高差较大的广西金秀瑶族自治县为研究区,以2018年获取的机载激光雷达点云及2015年获取的高分辨率航摄正射影像为数据源,建立基于多源遥感的巨树检测方法。对分类后的机载激光雷达点云进行高程归一化处理,以20 m×20 m分辨率生成点云最大高栅格图;对点云最大高栅格图进行异常值检测,筛选出>50 m的像元,并排除坡度>60°区域的像元,生成巨树初检矢量;将巨树初检矢量叠加机载激光雷达点云数据及高分辨率航摄正射影像,进行人工判读,确定其是否为巨树。结果显示,在金秀县共检测出14棵巨树,与实地调研结果一致。以机载激光雷达为核心的多源遥感巨树检测方法,具有坚实的物理机理,利用机载激光雷达对地表进行精准三维刻画,辅以高分辨率光学影像,可十分便捷、准确地识别巨树及提取巨树基本信息,在森林生态系统的保护和管理中具有良好的应用前景。

To explore giant tree detection method by multi-source remote sensing with airborne LiDAR as core so as to extract positions and information of giant trees accurately,quickly and comprehensively,Jinxiu county in Guangxi,which has large area of natural forests,complex terrain and large elevation difference,was taken as research area.Airborne LiDAR point cloud obtained in 2018 and high-resolution aerial orthophotography ob⁃tained in 2015 were used as data sources to build giant tree detection method based on multi-source remote sensing.Elevation normalization of airborne LiDAR point lound after classification was carried out,and maxi⁃mum height raster map of point cloud was generated with resolution of 20 m×20 m.Outlier detection was car⁃ried out on maximum height raster map of point cloud to select pixels>50 m and exclude pixels with slopes>60°,and initial detection vector of giant tree was generated.Finally,initial detection vector of giant tree was superim⁃posed on airborne LiDAR point cloud data and high-resolution aerial orthophotography,and artificial interpre⁃tation was carried out to determine whether it was a giant tree.Results showed that fourteen giant trees were de⁃tected in Jinxiu county,which was consistent with field survey.The multi-source remote sensing giant tree de⁃tection method with airborne LiDAR as core had solid physical mechanism.By using accurate three-dimension⁃al characterization of airborne LiDAR to ground surface and supplementing by high-resolution optical images,the detection method could identify giant tree and extract basic information easily and accurately,which had good application prospect in forestry ecosystem protection and management.