无人机飞行高度对植被覆盖度和植被指数估算结果的影响  

Effects of UAV flight height on estimated fractional vegetation cover and vegetation index

作　　者：何勇[1,2] 杜晓月 郑力源 朱姜蓬 岑海燕 许丽佳[3] He Yong;Du Xiaoyue;Zheng Liyuan;Zhu Jiangpeng;Cen Haiyan;Xu Lijia(College of Biosystems Engineering and Food Science,Zhejiang University,Hangzhou 310058,China;Key Laboratory of Spectroscopy Sensing,Ministry of Agriculture and Rural Affairs,Hangzhou 310058,China;College of Mechanical and Electrical Engineering,Sichuan Agricultural University,Sichuan 625014,China)

机构地区：[1]浙江大学生物系统工程与食品科学学院,杭州310058 [2]农业农村部光谱检测重点实验室,杭州310058 [3]四川农业大学机电学院,四川625014

出　　处：《农业工程学报》2022年第24期63-72,共10页Transactions of the Chinese Society of Agricultural Engineering

基　　金：浙江省重点研发计划项目(021C02023)。

摘　　要：将无人机与多种成像传感设备相结合可实现田间作物表型信息的全面获取。针对田间复杂环境下无人机搭载多种成像传感设备在不同飞行高度处提取的作物信息具有差异性的问题,该研究着重探究了无人机搭载2种成像传感设备获取图像时,不同飞行高度对估算植被覆盖度以及植被指数结果的影响。首先为防止外界环境变化对获取图像质量造成干扰,通过最近邻插值算法将无人机飞行高度为25 m处获取的2个多光谱和可见光图像数据集分别退化为10个不同地面分辨率的模糊图像数据集,模拟无人机在不同飞行高度下获取的作物图像。然后获取50 m高度处的无人机图像数据集,通过皮尔逊相关性分析验证模拟数据集的有效性。最后采用随机森林模型估算不同数据集中的植被覆盖度,分类精度大于91%。结果发现,当植被覆盖度小于1/2时,随着地面分辨率的降低该指标不断被低估,反之则被高估。飞行高度50 m的真实图像与模拟图像估算植被覆盖度结果的相关系数为0.9928,两者具有强相关性,模拟图像估算得到的植被覆盖度变化具备参考意义。植被指数估算结果中,首先对无人机图像数据集进行辐射校正、阈值分割等图像预处理,然后根据公式计算得到植被指数,最后通过假设性检验对10个图像数据集计算得出的植被指数进行分析。结果发现,可见光植被指数在飞行高度61m时均具备显著性差异(P<0.05),在飞行高度42m时没有差异(P>0.1),多光谱植被指数在10个高度下均没有显著性差异,因此为保证无人机获取数据的准确性与完整性,当无人机搭载该研究的两种相机获取作物信息时建议飞行高度为42 m。研究结果可为研究者利用无人机搭载多传感设备获取作物信息设定合适的飞行高度、减小作业成本提供参考。Accurate acquisition of crop information is very important for the real-time monitoring of crop growth status in crop breeding and agricultural precision management.The ever-increasing remote sensing(RS)and Unmanned Aerial vehicles(UAV)have been widely used to collect the big phenotypic data(especially image data)of various plants in a large area.Specific sensing devices are often used to obtain accurate and comprehensive information on crop growth.However,the diversity of sensing devices can bring a great challenge to the cooperation between them on UAV.It is very urgent to establish UAV flight parameters suitable for the complex field environment.In this study,an effective approach was developed to explore the effects of UAV flight height on the estimated fractional vegetation cover(FVC)and vegetation index(VI).The UAV can ultimately be utilized to collect the phenotypic data within the effective range under the multi-sensor combinations based on suitable flight height.The multispectral and high-resolution RGB cameras were simultaneously mounted on the UAV to acquire the images at a speed of 2.5 m/s.Among them,the heading and the side overlaps were set as 75%and 60%,respectively,particularly for the possibility of successful image stitching.The initial flight heights were set as 25 and 50 m,in order to exclude the wind field generated by the high-speed rotation of the UAV’s paddles from disturbing the crop.The high ground resolution(GR)images were collected by the UAV at the low flight height.These images were then degraded by image processing into a series of images with different GR.The simulation was finally carried out to predict the crop images obtained by the UAV at different flight heights.As such,the impact of environmental changes on image quality was reduced,such as the light intensity.The FVC of every sample plot was estimated using Random Forest(RF).The results show that the vegetation classification accuracy was greater than 91%.The regular FVC changes were found in the different GR,due to the im

关 键 词：无人机 植被指数 植被覆盖度 飞行策略 

分 类 号：S24[农业科学—农业电气化与自动化]