基于无人机影像的边坡植物物种分类  

Classification of Slope Plant Species Based on Image of UAV

作　　者：翟浩 唐彬童 辜彬[1] ZHAI Hao;TANG Bin-tong;GU Bin(College of Life Sciences,Sichuan University,Key Laboratory of Bio-resource and Eco-environment of Ministry of Education,Chengdu 610065,Sichuan,China)

机构地区：[1]四川大学生命科学学院,生物资源与生态环境教育部重点实验室,四川成都610065

出　　处：《西北林学院学报》2020年第3期185-190,249,共7页Journal of Northwest Forestry University

摘　　要：无人机的出现,给生态调查带来关键性革新。而使用无人机进行生态调查,植物遥感分类是关键,基于平地的无人机植物物种分类创新运用于边坡,使用可见光正射影像联合nDSM(normalized digital surface model,归一化数字表面模型)对边坡植物物种进行分类。结果表明,边坡样地的分类的精度达85%,自然样地达84%,与没有加入nDSM的分类结果对比,边坡、自然样地分类精度分别增加了32%和16%。在边坡条件下可见光正射影像与nDSM结合,可大幅度提升边坡植物物种分类精细度。The appearance of unmanned aerial vehicle(UAV)has introduced a critical and revolutionary way to the ecological investigation.When applying UAV to the ecological investigation,classification of plant species from the remote sensing images plays a decisive role.This paper innovatively applied the flat-terrain-based plant species classification with UAV to slope terrain,in which the plant species on slope terrain were interpreted by visible orthophoto images combined with normalized digital surface model(nDSM).The results showed that the classification accuracy could be as high as 85%on the sample plots of slope terrain,while it was 84%on natural terrain.Compared with those without the combination of nDSM,the accuracy increased 32%and 16%on slope and natural terrains,respectively,indicating that the combination of visible orthophoto images and nDSM in the classification on slope could improve the precision of plant species interpretation.

关 键 词：边坡 无人机 归一化数字表面模型 植物物种分类 

分 类 号：S731.1[农业科学—林学]