红外光谱结合光谱检索对石斛品种的鉴别研究

摘要

The infrared spectral of stems of 165 trees of 23 dendrobium varieties were obtained by means of Fourier transform in-frared spectroscopy technique. The spectra show that the spectra of all the samples were similar,and the main components of stem of Dendrobium is cellulose. By the spectral professional software Omnic8. 0 ,three spectral databases were constructed. Lib01 includes of the average spectral of the first four trees of every variety,while Lib02 and Lib03 are constructed from the first-derivative spectra and the second-derivative spectra of average spectra,separately. The correlation search,the square differ-ence retrieval and the square differential difference retrieval of the spectra are performed with the spectral database Lib01 in the specified range of 1 800~500 cm-1 ,and the yield correct rate of 92. 7%,74. 5%and 92. 7%,respectively. The square differen-tial difference retrieval of the first-derivative spectra and the second-derivative spectra is carried out with Lib02 and Lib03 in the same specified range 1 800~500 cm-1 ,and shows correct rate of 93. 9% for the former and 90. 3% for the later. The results show that the first-derivative spectral retrieval of square differential difference algorithm is more suitabe for discerning dendrobi-um varieties,and FTIR combining with the spectral retrieval method can identify different varieties of Dendrobium,and the cor-relation retrieval,the square differential retrieval,the first-derivative spectra and second-derivative spectra retrieval in the speci-fied spectral range are effective and simple way of distinguishing different varieties of dendrobium.%用傅里叶变换红外光谱技术结合光谱检索的方法对石斛茎进行品种鉴别研究。测试了23个品种，165株石斛茎样品的红外光谱，光谱显示，各样品的红外光谱非常相似，石斛茎中的主要成分是纤维素等多糖物质。利用光谱软件Omnic8．0建立了三个光谱数据库Lib01、Lib02和Lib03，Lib01由每个品种前4株样品的平均光谱组成，Lib02由平均光谱的一阶导数光谱组成，Lib03由平均光谱的二阶导数光谱组成。各样品光谱分别与Lib01进行指定在1800～500 cm-1光谱范围的相关性、平方差和平方微分差检索，相关性检索的正确率达到92．7％，平方差检索的正确率达到74．5％，平方微分差检索的正确率达到92．7％。各样品光谱的一阶导数光谱和二阶导数光谱分别与Lib02和Lib03进行指定在1800～500 cm-1光谱范围的平方微分差检索，一阶导数光谱的检索正确率达到93．9％，二阶导数光谱的检索正确率达到90．3％。结果表明，平方微分差算法的一阶导数红外光谱检索更适合于石斛的品种鉴别。红外光谱结合光谱检索的方法能较好的鉴别不同品种的石斛，有望成为简便、易行的植物品种鉴别方法。