红外技术, 2013, 35 (5): 304, 网络出版: 2013-05-24
基于茶叶红外光谱的特征识别方法
The Characteristic Identification Method Based on Infrared Spectra of Tea
红外光谱 特征基 茶叶识别 特征识别 infrared spectrum characteristic basis tea identification characteristic identification
摘要
通过对识别理论的研究分析, 建立了特征基理论, 选择合适样品建立的光谱特征基, 能反映样品的综合特征信息, 从而对多组分物质进行识别研究; 以茶叶样品为分析对象, 进行了特征基理论的应用研究, 实现了对重度发酵茶和普洱熟茶的识别; 以普洱熟茶识别为例, 通过比较特征提取前后茶叶的识别效果, 分析了光谱分辨率问题, 结果显示: 特征提取后茶叶类内平均距离减小为原来的 2× 10-3倍, 类间相对平均距离增大为原来的 4~66倍, 类间最小最大距离比值增大为原来的 34~156倍, 所以类内聚集度增大, 类间分散度增大, 即特征基理论使光谱分辨率大幅度提高。
Abstract
On the analysis of the identification theory, the paper presents the characteristic basis theory. The multicomponent material can be distinguished by characteristic basis that is composed by proper samples and can reflect the comprehensive characteristic information. Taking tea as analytic target, the application research of characteristic basis theory was made. The work could distinguish severe fermented tea and zymic pu-er tea. Taking an example of the recognition of zymic pu-er tea, the resolution of the spectrum was ana-lyzed by comparing the identification effect of tea before and after feature extraction. The results showed: The intra-class average distance reduced to 2×10-3 times of original data and the relative inter-class average distance increased to the 4~66 times of original data, the inter-class min-max distance radio increased to the 34~156 times of original data. Thus the degree of aggregation of same samples and the degree of dispersion of different samples increased, that is to say, spectral resolution was greatly improved by characteristic basis theory.
张荣香, 张玮, 张艳伟, 李晓苇. 基于茶叶红外光谱的特征识别方法[J]. 红外技术, 2013, 35(5): 304. ZHANG Rong-xiang, ZHANG Wei, ZHANG Yan-wei, LI Xiao-wei. The Characteristic Identification Method Based on Infrared Spectra of Tea[J]. Infrared Technology, 2013, 35(5): 304.