基于FTIR-CWT及ANN分类法的中药材菟丝子与同科属金灯藤子的识别方法研究  

Research of Recognition Method between Semen Cuscutae and Its Sibling Plant Japanese Dodder Seed Based on FTIR-CWT and ANN Classification Method

作　　者：程存归[1] 田玉梅[1] 张长江[2] 

机构地区：[1]浙江师范大学化学系浙江省固体表面反应化学重点实验室,金华321004 [2]浙江师范大学电子信息工程系,金华321004

出　　处：《化学学报》2008年第7期793-798,共6页Acta Chimica Sinica

摘　　要：采用水平衰减全反射傅里叶变换红外光谱法(HATR-FTIR)测定了种子植物中药材菟丝子及同科属种子植物金灯藤子的FTIR,运用基于连续小波变换多分辨率分析法对吸收较为相似的菟丝子及金灯藤子的FTIR进行特征提取.通过分析比较后选择第7,10,13分解层的特征向量,进行人工神经网络(ANN)训练,再用训练出来的网络对不同产地的植物种子菟丝子和金灯藤子所得FTIR小波提取的特征向量进行分类.通过对32个不同样本的验证,说明能够采用基于FTIR-连续小波特征提取及人工神经网络分类法对同科属中药材菟丝子与金灯藤子进行识别.Fourier transform infrared （FTIR） and horizontal attenuated total reflectance （HATR） techniques were used to obtain the FTIR of semen cuscutae （the seed from Cuscuta chinensis Lam.） and its sibling Japanese dodder seed （the seed from Cuscutajaponica Choisy）. The similar features between the semen cuscutae and Japanese dodder seed were extracted by continuous wavelet transform （CWT）. After comparison analysis, the decomposition levels 7, 10 and 13 were used to extract the feature vectors, which were used to train the artificial neural network （ANN）. The trained neural network was used to classify semen cuscutaes and Japanese dodder seeds, which were collected from different places all over the country. According to 32 testing samples, the sibling plants, semen cuscutae and Japanese dodder seed, could be effectively identified by FTIR with continuous wavelet feature extraction and artificial neural network classification.

关 键 词：水平衰减全反射傅里叶变换红外光谱法 连续小波特征提取 人工神经网络 同科属中药材 识别分析 

分 类 号：R284[医药卫生—中药学]