中文摘要：

利用傅里叶变换红外光谱（FTIR）技术与统计分析相结合对柑橘褐斑病、黄龙病、溃疡病、煤烟病、芽枝孢霉斑病和正常的叶片进行研究.发现正常叶和病害叶的红外光谱很相似,导数光谱能够提高分辨率和放大差异,对光谱分别进行二阶导数分析,发现病害叶和正常叶的二阶导数光谱在1 200～700 cm^-1范围内有明显的差异,选取该区的光谱数据进行相关性分析.结果显示,柑橘正常叶之间、同种病害叶之间的相关系数都在0.918以上,不同病害叶之间及不同病害叶与正常叶之间相关系数差异较大.选取1 200～700 cm^-1范围内的原始光谱、一阶和二阶导数光谱分别进行主成分分析和聚类分析.所有样本用二阶导数光谱进行主成分分析的正确率为92.5％,比原始光谱和一阶导数光谱正确率高.原始光谱、一阶和二阶导数聚类分析正确率分别为52.5％、80.0％、90.0％.结果表明,傅里叶变换红外光谱技术能够快速、准确地检测区分这5种柑橘病害.

英文摘要：

Citrus brown spot,citrus yellow shoot,canker,fuliginous,cercospora sp.and healthy leaves were analyzed by Fourier transform infrared spectroscopy （FHR） combined with statistical analysis.The results showed that the spectra of samples were similar.Derivative spectra of FTIR can obviously enhance the spectral resolution and amplify small differences.The second derivative spectra were analyzed with obvious differences in the range of 1 200～700 cm^-1.The correlative analysis showed that the correlation coefficients were more than 0.918 between healthy leaves,and between the same diseased leaves.The preprocessed original dataset,first derivative dataset and second derivative dataset in the range of 1 200～700 cm-1 were used to make the principal component analysis（PCA） and hierarchical cluster analysis（HCA）.The performance of the overall accuracy of PCA was 92.5%,better than original dataset and first derivative dataset.The cluster analysis accuracy of original spectra,first derivative and second derivative was 52.5%,80.0%,90.0%,respectively.It is indicated that FTIR spectroscopy can detect citrus diseases fast and accurately.