光学学报, 2019, 39 (3): 0330001, 网络出版: 2019-05-10
基于拉曼光谱和模式识别算法的软玉产地鉴别 下载: 1230次
Nephrite Origin Identification Based on Raman Spectroscopy and Pattern Recognition Algorithms
光谱学 产地鉴别 拉曼光谱 马氏距离 随机森林 软玉 模式识别 spectroscopy origin identification Raman spectroscopy Mahalanobis distance random forest nephrite pattern recognition
摘要
利用自行研制的便携式拉曼光谱仪,研究了甘肃、青海和新疆3个产地软玉的拉曼光谱特征,并对其光谱差异进行解析;基于马氏距离判别方法和随机森林判别方法实现了3个产地软玉的无损鉴别。结果表明:利用马氏距离判别方法和随机森林判别方法可以对具有相同拉曼峰的不同产地的软玉进行鉴别,鉴别准确率分别为87.5%和95.83%。
Abstract
The self-developed portable Raman spectrometer is used to analyze the Raman spectral characteristics of nephrites from Gansu, Qinghai, and Xinjiang, and the spectral differences are analyzed. The nondestructive identification of nephrites from three different origins is realized based on the two pattern recognition algorithms of Mahalanobis distance discriminated method and random forest discriminated method. The results show that the Mahalanobis distance discriminated method and the random forest discriminated method can be used to distinguish the nephrites from different origins with the same Raman peak, and the identification accuracies of them are 87.5% and 95.83%, respectively.
徐荟迪, 林露璐, 李征, 黄梅珍, 周征宇. 基于拉曼光谱和模式识别算法的软玉产地鉴别[J]. 光学学报, 2019, 39(3): 0330001. Huidi Xu, Lulu Lin, Zheng Li, Meizhen Huang, Zhengyu Zhou. Nephrite Origin Identification Based on Raman Spectroscopy and Pattern Recognition Algorithms[J]. Acta Optica Sinica, 2019, 39(3): 0330001.