基于光谱技术的稻飞虱虫害聚类研究

为了实现可靠的稻飞虱虫害诊断,使用PR715光谱辐射度计采集了健康水稻叶片、有幼虫寄生的叶片、单个体成虫及参考白板的光谱值,获得各个阶段稻飞虱病虫害的光谱。由于原始光谱在谱线的两端噪声比较大,因此用滤波方法对光谱进行平滑处理,得到相对平滑的反射率光谱曲线。采用主分析、欧式距离和余弦3中方法进行聚类分析,其中主成分分析、余弦的分类正确率为100%,欧式距离的平均分类正确率均为：92.24%,相对较差。实验结果证明,光谱分析方法在一定程度上能快速、精确提取稻飞虱病害的信息,实现聚类分析,为对稻飞虱病虫害进行快速、精确和非破坏性诊断提供技术支持。

Research Onclustering of Rice Planthoppers Based on Spectroscopy

To get the reliable diagnosis of rice planthoppers pest,in this paper,we used PR715 to collect the spectral reflectance below:leaves of healthy rice,leaves Parasitized by larvaes of rice planthoppers,alone imago and reference white under standard observation environment.Median filter and mean filter,which M=5,were been used to smooth the spectrum,so that smoothing spectrum could be obtained.The 4 samples were classified by using Principal Component Analysis(PCA),Euclidean distance and angle.The accuracy rate by using PCA and angle was 100%,and used Eculidean distance was 92.24%.Experiment result shows that the method realized a good accuracy in the diagnosis of the rice planthoppers.