• Laser & Optoelectronics Progress
  • Vol. 59, Issue 18, 1830002 (2022)
Wei Hou1, Jifen Wang1、*, and Yiran Liu2
Author Affiliations
  • 1School of Investigation, People’s Public Security University of China, Beijing 100038, China
  • 2School of Police Administration, People’s Public Security University of China, Beijing 100038, China
  • show less
    DOI: 10.3788/LOP202259.1830002 Cite this Article Set citation alerts
    Wei Hou, Jifen Wang, Yiran Liu. Spectral Pattern Recognition and Traceability Analysis of Human Fingernail Based on Machine Learning[J]. Laser & Optoelectronics Progress, 2022, 59(18): 1830002 Copy Citation Text show less
    Infrared spectra of fingernail samples. (a) Infrared spectra of different sampling sites for the same fingernail sample; (b) infrared spectra of ten fingernails from the same person; (c) infrared spectra of fingernail samples from different people
    Fig. 1. Infrared spectra of fingernail samples. (a) Infrared spectra of different sampling sites for the same fingernail sample; (b) infrared spectra of ten fingernails from the same person; (c) infrared spectra of fingernail samples from different people
    Variance contribution rate depending on number of principal components and factor components. (a) Principal components; (b) factor components
    Fig. 2. Variance contribution rate depending on number of principal components and factor components. (a) Principal components; (b) factor components
    Classification accuracy of MLP and RBF models based on PCA and FA dimensionality reduction. (a) PCA-MLP; (b) FA-MLP; (c) PCA-RBF; (d) FA-RBF
    Fig. 3. Classification accuracy of MLP and RBF models based on PCA and FA dimensionality reduction. (a) PCA-MLP; (b) FA-MLP; (c) PCA-RBF; (d) FA-RBF
    Spatial classification details of fingernail samples based on SVM model
    Fig. 4. Spatial classification details of fingernail samples based on SVM model
    Classification accuracy of SVM model based on different principal components
    Fig. 5. Classification accuracy of SVM model based on different principal components
    Spatial classification details of fingernail samples from five provinces of north China
    Fig. 6. Spatial classification details of fingernail samples from five provinces of north China
    Wavenumber /cm-1Mode of vibration
    3292O-H stretching,carboxyl acid and derivatives
    3068Amide A and B and NH stretching
    2925C-H symmetric stretching(CH2 and CH3 anti symmetric and symmetric stretching modes)
    2858
    1618C=C stretching
    1532Amide II,C-N stretch and N-H in plane bend
    1461C-H deformation in CH2
    1241Amide III band,C-N stretching vibrations
    1062C-C trans conformation
    756Cis-R,CH≡CHR
    Table 1. Spectral peaks and their modes of vibration
    RegionClassification accuracy /%
    CHAIDExhaustive CHAIDCRTQUEST
    Training setTest setTraining setTest setTraining setTest setTraining setTest set
    R192.994.482.692.997.684.287.281.0
    R262.550.057.166.771.466.7100.0100.0
    R397.292.996.486.486.185.794.393.3
    R4100.083.380.0100.066.7100.085.733.3
    R584.2100.091.762.585.783.394.7100.0
    R688.9100.0100.080.084.657.173.3100.0
    R794.4100.0100.0100.095.0100.046.720.0
    Total91.092.089.286.289.182.884.873.7
    Table 2. Classification accuracy of decision tree model based on different algorithms
    TypeClassification accuracy /%
    RBF kernelPolynomial kernelSigmoid kernelLinear kernel
    Training set90.8100.047.388.6
    Test set92.2100.050.082.8
    Table 3. Classification accuracy of support vector machine model based on different kernel functions
    ProjectTotal number of samplesNumber of unknown samplesNumber of correctly classified samplesClassification accuracy /%
    MLPSVMMLPSVM
    119557545794.7100.0
    260151515100.0100.0
    Table 4. Classification results of unknown samples by MLP and SVM models
    Wei Hou, Jifen Wang, Yiran Liu. Spectral Pattern Recognition and Traceability Analysis of Human Fingernail Based on Machine Learning[J]. Laser & Optoelectronics Progress, 2022, 59(18): 1830002
    Download Citation