Laser & Optoelectronics Progress, Volume. 60, Issue 4, 0430002(2023)

Comparison of Spectral Pattern-Recognition Methods for Slimming Drugs in Express Delivery Channels

Aolin Zhang1, Jifen Wang1、*, Song Liu1, Xuejun Shi2, Xiaojie Xu2, Di Zhou2, and Zhen Zhang1
Author Affiliations
  • 1College of Investigation, People's Public Security University of China, Beijing 100038, China
  • 2Forensic Expertise Center of Beijing Customs Anti-Smuggling Bureau, Beijing 100000, China
  • show less
    Figures & Tables(12)
    Cumulative variance contribution rate of principal component
    Infrared spectrograms of 145 samples
    Spatial distribution of Fisher discriminant samples
    Classification accuracy of samples under KNN model
    Classification accuracy of SVM model
    A, B infrared spectrum. (a) A sample; (b) B sample
    • Table 1. Experimental equipment and parameters

      View table

      Table 1. Experimental equipment and parameters

      FacilityParameter

      Nicolet iS20

      Fourier transform infrared spectrometer

      Spectral resolution:2 cm-1
      Spectral repeatability:better than 0.02 nm
      Number of scans:32
      Measuring range:4000-400 cm-1
      Dynamic adjustment up to 130000 times/s
      Signal-to-noise ratio:50∶1
    • Table 2. Summary of discriminant functions of 5 groups of drug samples

      View table

      Table 2. Summary of discriminant functions of 5 groups of drug samples

      FunctionFunction testCorrelationTotalWilks’ LambdaSigEigenvalue
      FDA11 to 40.9982263.45100.000286.651
      FDA22 to 40.9961482.12900.000124.386
      FDA33 to 40.99815.3970.0030.00050.449
      FDA440.928271.5960.140.9136.157
    • Table 3. Accuracy of drug classification in 5 groups

      View table

      Table 3. Accuracy of drug classification in 5 groups

      TypeZolpidemEstazolamFlunirazepmSibutramineSildenafilAccuracy /%
      Zolpidem30100
      Estazolam30100
      Flunirazepm30100
      Sibutramine25100
      Sildenafil30100
    • Table 4. Classification accuracy of samples under KNN model units: %

      View table

      Table 4. Classification accuracy of samples under KNN model units: %

      KZolpidemEstazolamFlunirazepmSibutramineSildenafil
      159.459.459.459.459.4
      288.588.588.588.588.5
      393.593.593.593.593.5
      498.298.298.298.298.2
      5100100100100100
      69999999999
      7100100100100100
      8100100100100100
      9100100100100100
      10100100100100100
      11100100100100100
      12100100100100100
      13100100100100100
    • Table 5. Number of 8 brands of Zolpidem

      View table

      Table 5. Number of 8 brands of Zolpidem

      DrugNumber
      Miaozi Slimming capsulesC1
      Xianmei capsulesC2
      Xianfeixianwei tabletC3
      Quzixiu Slimming capsulesC4
      Zhuikelier capsulesC5
      Xianti pill ⅢC6
      Meizhixuan PillC7
      Xiuzhixiufu capsulesC8
    • Table 6. Classification accuracy of 8 brands of Zolpidem

      View table

      Table 6. Classification accuracy of 8 brands of Zolpidem

      BrandC1C2C3C4C5C6C7C8Accuracy
      C13100%
      C21150%
      C31150%
      C410100%
      C51100%
      C67100%
      C72100%
      C81266.70%
    Tools

    Get Citation

    Copy Citation Text

    Aolin Zhang, Jifen Wang, Song Liu, Xuejun Shi, Xiaojie Xu, Di Zhou, Zhen Zhang. Comparison of Spectral Pattern-Recognition Methods for Slimming Drugs in Express Delivery Channels[J]. Laser & Optoelectronics Progress, 2023, 60(4): 0430002

    Download Citation

    EndNote(RIS)BibTexPlain Text
    Save article for my favorites
    Paper Information

    Category: Spectroscopy

    Received: Nov. 22, 2021

    Accepted: Dec. 27, 2021

    Published Online: Feb. 13, 2023

    The Author Email: Wang Jifen (wangjifen58@126.com)

    DOI:10.3788/LOP213018

    Topics