1. 基于去除土壤效应的烤烟叶面积指数及烟碱含量的无人机高光谱监测研究.
- Author
-
蒋薇, 严定春, 李 栋, 孙伟超, 薛博文, 程 涛, 李军营, and 汤 亮
- Subjects
- *
LEAF area index , *OPTIMIZATION algorithms , *DRONE aircraft , *NICOTINE , *FIELD research , *TOBACCO - Abstract
[Objectives] Soil background interferences with unmanned aerial vehicle (UAV) remote sensing monitoring. This paper aimed to investigate soil removal background for UAV hyperspectral monitoring of tobacco growth and quality. [Methods] In this study, a flue-cured tobacco field experiment involving two varieties and four nitrogen application levels was conducted. A hyperspectral camera mounted on the UAV platform was used to obtain the canopy reflection spectrum data of flue-cured tobacco during key growth periods. The average spectrum of each plot after soil effect removal was removed by spectral separation of soil and vegetation (3SV) algorithm or vegetation index threshold method, respectively. The vegetation index optimization algorithm was used to construct the hyperspectral monitoring model of leaf area index and nicotine content of flue-cured tobacco during the whole growth period. [Results] After using the 3SV algorithm, the high correlation between band combinations and tobacco nicotine was distributed in combination with λ1:450-500 nm and λ2:580-660 nm, λ1:630-670 nm and λ2:680-700 nm. The high correlation between band combinations and leaf area index was found in combination with λ1:730-770 nm and λ2:750-800 nm, λ1:510-600 nm and λ2:680-700 nm. Compared to the vegetation index threshold algorithm, the 3SV algorithm significantly improved the monitoring model's validation accuracy for both tobacco nicotine content and leaf area index. The model validation determination coefficient R² of ‘Yunyan 87' nicotine increased from 0.64 to 0.88, and the RMSE decreased from 0.71% to 0.29%, which showed the most significant effect. Using 3SV algorithm and band optimization algorithm, the index with the best relationship of leaf area index in ‘Yunyan 87' and ‘K326' was NDLI515, 691 and NDLI764,799, respectively, the index with the best relationship of nicotine content of ‘Yunyan 87' and ‘K326' was NDNI450,658 and NDNI456,654, respectively. A general leaf area index and nicotine equation were established for monitoring the whole field growth period of the UAV platform. [Conclusions] By comparing different soil background removal algorithms, it was found that the 3SV algorithm improved the accuracy of tobacco leaf area index and nicotine monitoring model, and provided technical support for UAV hyperspectral large-area monitoring of tobacco growth and quality. [ABSTRACT FROM AUTHOR]
- Published
- 2024
- Full Text
- View/download PDF