Early detection of pine wilt disease in Pinus tabuliformis in North China using a field portable spectrometer and UAV-based hyperspectral imagery

结合了无人机和地面的高光谱数据，使用植被指数(VI)、红边参(REP)、水分指数(MI)及三者组合建立了随机森林分类模型。结果表明：对于地面数据，结合所有参数的模型（OA：80.17%, Kappa：0.73）比 VI（OA：75.21%，Kappa：0.66）、REP（OA：79.34%，Kappa：0.67）和 MI（OA：74.38%, Kappa: 0.65) 预测单木感病阶段的精度高。在区分松材线虫病（PWD）早期的树木和健康的树木中，REP 的准确度最高(OA: 80.33%, Kappa: 0.58)。基于无人机高光谱数据产生了类似的结果。总的来说，我们的结果证实了使用高光谱数据来识别感染PWD的松树的有效性。

This study combined unmanned aerial vehicle (UAV)-based and ground-based hyperspectral data, and established random forest classification models using vegetation index (VI), red edge parameter (REP), moisture index (MI), and their combinations. The results demonstrated that for ground-based data, the model integrating all parameters (OA: 80.17%, Kappa: 0.73) achieved higher accuracy in predicting the disease stage of individual trees than the models using only VI (OA: 75.21%, Kappa: 0.66), REP (OA: 79.34%, Kappa: 0.67), and MI (OA: 74.38%, Kappa: 0.65). When differentiating between early-stage pine wilt disease (PWD)-infected trees and healthy trees, the REP-based model showed the highest accuracy (OA: 80.33%, Kappa: 0.58). Consistent results were obtained using UAV-based hyperspectral data. Overall, our findings confirm the effectiveness of utilizing hyperspectral data for identifying PWD-infected pine trees.

搜集汇总

背景与挑战

背景概述

该数据集研究了利用便携式光谱仪和无人机高光谱图像技术早期检测华北油松松材线虫病的有效性。研究通过植被指数、红边参数和水分指数建立随机森林分类模型，结果表明高光谱数据在识别感染松材线虫病的松树方面具有较高准确性。