首页|基于无人机高光谱影像的水稻叶绿素含量反演

基于无人机高光谱影像的水稻叶绿素含量反演

扫码查看
原文链接
  • NETL
  • NSTL
  • 万方数据
  • 维普
[目的]为实现快速无损地监测水稻叶绿素含量，采用大疆M600 Pro无人机搭载SENOP RIKOLA高光谱仪获取水稻分蘖期冠层高光谱影像。[方法]利用相关性分析筛选出光谱指数的特征波长，构建DSI、RSI、NDSI、MSR、OSAVI和RDVI 6种植被指数，并利用一阶光谱导数计算其红边面积和红边幅值，分析8种光谱参数参与水稻叶绿素含量之间的相关性分析。将这些光谱参数作为CatBoost回归模型的输入变量，分析8种光谱参数对水稻叶绿素含量的估算能力。[结果]基于红边参数的反演模型中红边幅值拟合效果最好，其R2为0。952 4，RSME为0。638 1;基于植被指数的反演模型中OSAVI指数拟合效果最好，其R2为0。941 6，RSME为0。588 5。2种模型均能有效预测水稻叶绿素含量信息，可以作为水稻叶绿素含量监测的依据。[结论]将无人机高光谱遥感影像与机器回归算法相结合，可以实现对水稻冠层叶绿素含量的精准预测，从而对水稻的生长和健康状况进行实时监测，进而实现对水稻的精准施肥和精准灌溉，对水稻的增产增收以及精准农业的发展具有重要意义。
Inversion of Chlorophyll Content in Rice Based on UAV Hyperspectral Imagery
[Objective]This study is to quickly and non-destructively monitor the chlorophyll content of rice,using the DJI M600 Pro UAV equipped with the SENOP RIKOLA hyperspectrometer to obtain hy-perspectral images of the rice canopy at the tillering stage.[Method]The characteristic wavelengths of the spectral indices were screened out by correlation analysis,and six vegetation indices,DSI,RSI,NDSI,MSR,OSAVI and RDVI were constructed.The first-order spectral derivatives were used to calculate the red-edge area and red-edge amplitude,and 8 kinds of spectra were analyzed.The parameters were in-volved in the correlation analysis between different rice chlorophyll contents.Taking these spectral param-eters as the input variables of the CatBoost regression model,the estimating ability of the eight spectral parameters on rice chlorophyll content was analyzed.[Result]The research results show that the red-edge amplitude fitting effect is the best in the inversion model based on the red-edge parameter,its R2 is 0.952 4,and the RSME is 0.638 1;the OSAVI index in the vegetation index-based inversion model has the best fitting effect,and its R2 is 0.941 6 and the RSME is 0.588 5.Both models can effectively predict the information of rice chlorophyll content and can be used as the basis for rice chlorophyll content monitor-ing.[Conclusion]The combination of UAV hyperspectral remote sensing data and machine regression algorithm can accurately predict the chlorophyll content of rice canopy,so as to monitor the growth and health of rice in real time,and then realize the precise fertilization and irrigation of rice,which provides a powerful reference for increasing rice yield and income and the development of precision agriculture in the future.

UAVricechlorophyll retrievalhhyperspectral remote sensingCatBoost regression

何嘉晨、何敬、王彬、苟静、林远杨、刘刚

展开 >

成都理工大学地球科学学院,成都 610059

成都理工大学地质灾害防治与地质环境保护国家重点实验室,成都 610059

无人机 水稻 叶绿素反演 高光谱遥感 CatBoost回归

国家自然科学基金成都市技术创新研发项目成都理工大学研究生质量工程项目地质灾害防治与地质环境保护国家重点实验室项目四川省科技计划项目四川省自然资源厅

418713032022-YF05-01090-SN2022YJG022SKLGP2018Z0102021YFG0365kj-2021-3

2024

10.16036/j.issn.1000-2650.202205194

四川农业大学学报
四川农业大学

四川农业大学学报

CSTPCD北大核心
影响因子：0.657
ISSN：1000-2650
年,卷(期)：2024.42(1)
  • 25