摘要
以福建平潭岛木麻黄人工林为研究对象,对比研究基于无人机可见光遥感数据的株数提取方法,探讨多种方法在不同林龄木麻黄人工林株数提取的适用性。通过轻型旋翼无人机获取研究区可见光相片,经预处理生成无人机可见光正射影像(DOM)和冠层高度模型(CHM);分别选取成熟林、幼龄林6块标准地区域,使用基于冠层高度模型的局部最大值方法(LMC)、基于正射影像的局部最大值方法(LMD)、基于正射影像的多尺度分割方法(MST)提取标准地株数;最后通过3种方法提取的株数和实测数据进行对比分析。结果表明:3种方法中LMC总体提取精度最高,Fscore为0.97,而LMD和MST总体提取精度明显降低,Fscore分别为0.90、0.78;LMC方法对幼龄林和成熟林的株数提取精度相近,幼龄林和成熟林的Fscore皆为0.97。LMD方法在成熟林的株树提取精度略高于幼龄林,尤其2种林龄类型的P值相差较大,幼龄林RE:0.97、P:0.81、Fscore:0.88;成熟林RE:0.94、P:0.89、Fscore:0.91。MST方法在成熟林的株树提取精度明显高于幼龄林,幼龄林RE:0.88、P:0.67、Fscore:0.76;成熟林RE:0.88、P:0.74、Fscore:0.81。因此,3种方法中,LMC总体株数提取精度最高,且适用于不同林龄的木麻黄人工林株数提取,可以满足实时、快速提取木麻黄人工林株数的需求。
Study on Casuarina equisetifolia plantation in Pingtan island of Fujian Province.The methods of plantation extraction based on unmanned aerial vehicle(UAV)visible light remote sensing data were compared and investigated for their applicability in plantation extraction of different ages of plantation forests.Firstly,we acquired the visible photos of the study area by a light-rotor UAV,and generated the Digital Orthophoto Map(DOM)and Canopy Height Model(CHM)through pre-processing;secondly,6 standard plot areas of mature and young forests were selected respectively,and the number of standard plot plants was extracted using the local maximum method(LMC)based on canopy height model,the local maximum method(LMD)based on orthophoto,and the multi-scale segmentation method(MST)based on orthophoto;finally,we compared the number of plants extracted by the 3 methods with the measured data.The results showed that LMC has the highest overall extraction accuracy among the 3 methods,with an Fscore of 0.97,while LMD and MST are significantly lower,with an Fscore of 0.90 and 0.78,respectively;the extraction accuracy of LMC method was similar for the stems of young and mature forests,with Fscore of 0.97 for both young and mature forests;the accuracy of LMD method was slightly higher in mature forests than in young forests,especially the P-values of the 2 different forest ages,as follows:young forests(RE:0.97,P:0.81,Fscore:0.88)and mature forests(RE:0.94,P:0.89,Fscore:0.91).The accuracy of MST method was significantly higher in mature forests than in young forests,specifically,young forests(RE:0.88,P:0.67,Fscore:0.76)and mature forests(RE:0.88,P:0.74,Fscore:0.81).It was concluded that LMC had the highest overall strain extraction accuracy among the 3 methods,and it was suitable for strain extraction of different ages of mosaic plantation forests,which could meet the demand of real-time and rapid extraction of C.equisetifolia plantation strain.
作者
赵各进
孙梦莲
宋贤芬
郝振帮
李明慧
武红梅
刘健
余坤勇
Zhao Gejin;Sun Menglian;Song Xianfen;Hao Zhenbang;Li Minghui;Wu Hongmei;Liu Jian;Yu Kunyong(College of Forestry,Fujian Agriculture and Forestry University,Fuzhou Fujian 350002,China;Fujian Provincial Key Laboratory of 3S Technology and Optimal Utilization of Resources,Fuzhou Fujian 350002,China)
出处
《西南林业大学学报（自然科学）》
CAS
北大核心
2023年第3期127-135,共9页
Journal of Southwest Forestry University：Natural Sciences
基金
福建省林业科技项目(2022FKJ03)资助
3S技术与资源优化利用福建省高校重点实验室建设项目(PTJH17014)资助
福州市科技项目(FZLYKJ202201)资助。
关键词
无人机
可见光遥感
局部最大值
多尺度分割
株数
unmanned aerial vehicle
visible light remote sensing
local maxima
multi-scale segmentation
plant number
