基于地基激光雷达的单木森林参数提取分析
来源:用户上传
作者:傅根深 陈健 黄庆丰 唐雪海
摘要 [目的]利用激光雷达手段大范围估算森林参数。[方法]以安徽省黄山市休宁县岭南林场森林资源为研究对象,通过地面调查获取3种森林类型单木胸径和树高数据,并同步使用地基激光雷达获取点云数据,利用Hough变换检测算法提取每木胸径和树高,在Matlab2020a软件下构建基于地面调查数据和激光雷达提取值的估算模型。[结果]针叶林、阔叶林和毛竹林的胸径估测精度(R2c)分别为0.921、0.921和0.996,RMSEc分别为2.080、2.460和0.291,模型表现显著;树高估测精度分别为0.908、0.846和0.500,RMSEc分别为1.120、1.620和1.210。[结论]该研究可为运用激光雷达手段大范围估算森林参数提供依据,也可为森林类型点云识别与分割提供参考。
关键词 岭南林场;地基激光雷达;森林参数;回归分析;精度评价
中图分类号 S 758文献标识码 A文章编号 0517-6611(2022)04-0128-04
doi:10.3969/j.issn.0517-6611.2022.04.033
开放科学(资源服务)标识码(OSID):
Extraction and Analysis of Forest Parameters for Individual Tree Based on Terrestrial Laser Scanning
FU Gen-shen1, CHEN Jian2, HUANG Qing-feng1 et al
(1. School of Forestry and Landscape Architecture, Anhui Agricultural University, Hefei, Anhui 230061; 2. KQ GEO Technologies Co., Ltd, Beijing 100176)
Abstract [Objective]To estimate forest parameters in a wide range by lidar. [Method]The study studied by forest resources in Lingnan forestry site, Xiuning County, Anhui Province, single wood diameter at breast height and tree height data of three forest types were obtained by ground survey, and their point cloud data were obtained simultaneously by terrestrial laser scanning (TLS). Each wood diameter at breast height and tree height were extracted by Hough transform detection algorithm, and an estimation model based on ground survey data and TLS extracted values was constructed under Matlab2020a software. [Result]The results showed that the estimation accuracy (R2c) of breast diameter was 0.921, 0.921 and 0.996 for coniferous, broadleaf and moso bamboo forests, respectively, with RMSEc of 2.080, 2.460 and 0.291, and the model performance was significant; the estimation accuracy of tree height was 0.908, 0.846 and 0.500, respectively, with RMSEc of 1.120, 1.620 and 1.210, respectively. [Conclusion]This study can provide a basis for estimating forest parameters in a large scale by TLS, and also provide a reference for forest type point cloud identification and segmentation.
Key words Lingnan forest farm;Terrestrial laser scanning;Forest parameters;Regression analysis;Precision evaluation
基金项目 安徽省自然科学基金项目(1808085QC74)。
作者简介 傅根深(1997―),男,安徽合肥人,硕士研究生,研究方向:森林经理。*通信作者,讲师,博士,从事森林资源经营管理和林业3S技术研究。
收稿日期 2021-06-15
淠拘鼐逗褪鞲呤巧林调查中常用的参数,传统方法是通过胸径卷尺和测高器进行测量获取,但是这种方法存在着工作量大、人为误差等不足。除此之外,对于地形复杂、坡度大、调查人员难以进入的区域,这些林木因子获取难度较大。遥感技术一定程度上能够弥补这些不足,光学遥感技术在二维平面能够获得较好的水平森林参数,但是在获取树高、胸径、密度等三维参数时显得不足。作为一种新兴的主动遥感技术,激光雷达发射的激光脉冲则可轻易地从树木顶端到达地表,甚至穿透地表到达土壤内部,这种获取森林垂直结构参数的技术目前大部分光学遥感传感器难以做到[1]。地基激光雷达(terrestrial laser scanning,TLS)是一种基于激光的新型遥感仪器,通过光束偏转机制获取地表物体的三维点云数据,可以同时量测较高精度的地物范围和地物之间的角度,TLS获取的点云数据具有高密度、高精度等特点,可轻易捕捉林分尤其是下层林分的结构信息,这对于精确掌握森林生态系统的变化并研究这些变化对整个陆地生态系统的影响起着举足轻重的作用[2]。但是,由于大部分森林冠层之间相互遮挡的影响,机载激光雷达(airborne LiDAR scanning,ALS)则难以准确分离乔木层下的灌木和草本,ALS的优势体现在大区域森林垂直结构的快速获取,TLS则要耗费庞大的人力物力才能做到[3]。
转载注明来源:https://www.xzbu.com/1/view-15425652.htm