光谱学与光谱分析, 2017, 37 (1): 42, 网络出版: 2017-02-09
光谱匹配算法应用于鸡血藤与大血藤的太赫兹光谱区分
Spectral Matching Algorithm Applied to Identify the Terahertz Spectrum of Caulis Spatholobi and Caulis Sargentodoxae
太赫兹时域光谱 鸡血藤 大血藤 光谱角算法 准确率 Terahertz time-domain spectroscopy Caulis spatholobi Caulis sargentodoxae Spectral angle method Accuracy
摘要
鸡血藤和大血藤是药效、 产地不同而外形极相似的两种不同的藤茎类药材, 使用时极易混淆。 利用太赫兹时域光谱技术获得测试样品太赫兹频谱, 并结合光谱匹配算法分辨五种不同产地、 不同批次的鸡血藤和大血藤。 因中药化学成分的复杂性导致其不同成分的特征频谱发生相互的混叠, 使得鸡血藤和大血藤的太赫兹时域数谱均没有出现明显的特征峰。 重复实验测试同一样品同一位置10次, 通过计算得到不同频率下光谱数据的标准差。 数据在0.2~1.3 THz范围内有较好的稳定性。 对光谱数据进行二阶微分处理, 使光谱间的微小差异得以显现。 选取0.2~1.3 THz以及0.2~2范围内的光谱数据进行对比研究, 分别使用相关系数法、 分段编码法、 光谱角法和信息散度法四种光谱匹配算法对实验数据所属类别进行判定。 数据分析结果: 在0.2~1.3 THz范围内, 光谱角算法对鸡血藤与大血藤的太赫兹光谱数据有最好的分类效果, 对两类中药样品总计100组数据的分类正确率达到95%。 为易混中药的鉴定区分提供了一种新的技术与方法。
Abstract
Caulis spatholobi and caulis sargentodoxae are two different kinds of traditional Chinese medicine, which are similar to each other in appearance, but they are from differ sources with different functions.. This paper presents their spectrum using terahertz time-domain spectroscopy technology combines spectral matching algorithm to identify five different kinds of these samples. Due to the complex chemical composition of traditional Chinese medicine, different components of the characteristic absorption peak occurred easily mutually overlapping, therefore, it’s very difficult to find obvious characteristic absorption peak in terahertz range, caulis spatholobi and caulis like this. Every sample was measured 10 times in some position. These standard spectral data have been calculated in different frequency, data show good stability in range of 0.2~1.3 THz. The second order differential method is used to process the THz spectrum data from 0.2~1.3 THz. These tiny differences of absorption coefficient have been hemerged between caulis sargentodoxae and caulis spatholobi and compare with these data using the method of correlation coefficient block coding method, spectral angle method and information divergence method, respectively. The analysis results, the spectral angle method is the best algorithm, the discriminate accuracy of a total of 100 sets of data is 95%. It is a new way to distinguish difference of traditional Chinese medicine.
徐哲, 何明霞, 李鹏飞, 王鹏騛. 光谱匹配算法应用于鸡血藤与大血藤的太赫兹光谱区分[J]. 光谱学与光谱分析, 2017, 37(1): 42. XU Zhe, HE Ming-xia, LI Peng-fei, WANG Peng-fei. Spectral Matching Algorithm Applied to Identify the Terahertz Spectrum of Caulis Spatholobi and Caulis Sargentodoxae[J]. Spectroscopy and Spectral Analysis, 2017, 37(1): 42.