潜育期小麦条锈菌的高光谱定性识别
Hyperspectral qualitative identification on latent period of wheat stripe rust

小麦条锈病的发生流行严重影响小麦的生产安全，其早期的监测预警对病害发生的防控具有重要作用。本研究利用HandHeld 2地物光谱仪获取不同浓度小麦条锈菌胁迫下的潜育期小麦冠层高光谱数据，基于定性偏最小二乘（discriminant partial least squares，DPLS）、人工神经网络（artificial neural networks，ANN）和支持向量机（support vector machine，SVM）3种方法建立识别潜育期小麦条锈菌的模型，并分析接菌间隔日、建模比及光谱特征的差异对模型效果的影响。结果显示，在全波段（325~1 075 nm）建模，SVM识别效果优于ANN，而ANN优于DPLS，其中以一阶导数为光谱特征所建模型识别效果最优，在不同建模比下其识别准确率均可达到100.00%。
Wheat stripe rust seriously affects the safety of wheat production. The early monitoring and warning of this disease is of great significance to control the disease epidemic and ensure the quality and safety of wheat production. It is available to use the canopy hyperspectral reflectance which obtained from spectrograph HandHeld 2 to differentiate the different Pst concentrations in latent period. The effects of different inoculation days, different modeling ratios and different spectral features were assessed with the DPLS, ANN and SVM methods. The results showed that in the spectral region of 325-1 075 nm, the models accuracy based on ANN was better than the DPLS, and the SVM was the best. The model with the spectral feature of 1st derivative of reflectance had better accuracy than others, and the accuracy rate could be up to 100.00%.