基于无人机高光谱遥感的森林可燃物分类方法研究

Study on Forest Fuel Classification Method Based on UAV Hyperspectral Remote Sensing

以地处南亚热带季风气候区的佛山市高明区山地为研究区,探索利用无人机获取高光谱影像对森林可燃物类型进行划分的具体方法,将可燃物类型划分成针叶林、阔叶林、灌草地、竹林和非林地5个类型,采用无人机携带高光谱相机获取森林可燃物的高光谱影像设计针对森林可燃物的高光谱影像处理流程并进行遥感影像处理和分类。结果表明通过有监督的机器学习分类方法可以实现对试验区主要可燃物类型进行划分,经现场检验,分类准确度达到81.94%。研究表明基于无人机高光谱遥感的森林可燃物分类方法具备可行性。

Absract Using the mountainous area of Gaoming district,Foshan city,which is located in the southern subtropical monsoon climate zone as the research area,we explore the specific method of classifying the types of forest combustibles using hyperspectral images acquired by UAVs,classifying the fuel types into five types:coniferous forest,broad-leaved forest,scrub grass,bamboo forest and non-forested land.The results show that the supervised classification method can be applied to the forest combustibles.The results show that the classification of the main combustible material types in the test area can be achieved through the supervised machine learning classification method,and the accuracy of the classification reached 81.94%after the field inspection.The study shows that the method of forest combustibles classification based on UAV hyperspectral remote sensing is feasible.