• Acta Photonica Sinica
  • Vol. 50, Issue 7, 113 (2021)
Xiangxiang JIA, Baofeng GUO, Fanchang DING, and Wenjie XU
Author Affiliations
  • School of Automation, Hangzhou Dianzi University, Hangzhou310018, China
  • show less
    DOI: 10.3788/gzxb20215007.0710005 Cite this Article
    Xiangxiang JIA, Baofeng GUO, Fanchang DING, Wenjie XU. Hyperspectral Unmixing Based on Constrained Nonnegative Matrix Factorization[J]. Acta Photonica Sinica, 2021, 50(7): 113 Copy Citation Text show less
    Spectral signatures used to generate the simulated data and abundance map of endmember 1-4
    Fig. 1. Spectral signatures used to generate the simulated data and abundance map of endmember 1-4
    Performance of AEC-NMF with respect to parameters λ,μ and τ in terms of SAD and RMSE
    Fig. 2. Performance of AEC-NMF with respect to parameters λ,μ and τ in terms of SAD and RMSE
    Performance of AEC-NMF with respect to parameters γ and β in terms of SAD and RMSE
    Fig. 3. Performance of AEC-NMF with respect to parameters γ and β in terms of SAD and RMSE
    Jasper Ridge hyperspectral data
    Fig. 4. Jasper Ridge hyperspectral data
    Comparison of the library spectra (blue) with the endmember signatures extracted by AEC-NMF (red) on the jasper Ridge data set and the estimated fractional abundance map for each endmember
    Fig. 5. Comparison of the library spectra (blue) with the endmember signatures extracted by AEC-NMF (red) on the jasper Ridge data set and the estimated fractional abundance map for each endmember
    Cuprite hyperspectral data and twelve kinds of endmember spectral signatures
    Fig. 6. Cuprite hyperspectral data and twelve kinds of endmember spectral signatures
    Comparison of the library spectra (blue) with the endmember signatures extracted by AEC-NMF (red) on the Cuprite data set
    Fig. 7. Comparison of the library spectra (blue) with the endmember signatures extracted by AEC-NMF (red) on the Cuprite data set
    SNR/dBAEC-NMFTV-RSNMFL1/2NMFVCA-FCLS
    150.034 70.038 30.044 00.056 4
    250.009 80.012 50.012 80.015 3
    350.004 40.004 60.004 50.004 9
    Table 1. SAD values of different methods with the simulated data
    SNR/dBAEC-NMFTV-RSNMFL1/2NMFVCA-FCLS
    150.037 50.041 80.049 20.047 3
    250.009 10.010 40.015 80.016 0
    350.005 70.005 80.005 90.007 4
    Table 2. RMSE values of different methods with the simulated data
    MethodAEC-NMFTV-RSNMFL1/2NMFVCA-FCLS
    Tree0.041 30.044 00.153 00.155 9
    Soil0.163 30.186 90.092 60.245 3
    Water0.020 80.019 30.121 00.133 6
    Road0.082 80.083 90.060 40.106 9
    Mean0.077 00.083 50.106 80.160 4
    Table 3. SAD values of different methods with the real jasperRidge data set
    MethodAEC-NMFTV-RSNMFL1/2NMFVCA-FCLS
    Tree0.082 00.086 90.160 70.159 9
    Soil0.189 90.190 10.206 80.208 5
    Water0.094 70.098 00.128 10.130 0
    Road0.128 40.124 10.122 30.123 3
    Mean0.123 70.124 80.154 50.155 2
    Table 4. RMSE values of different methods with the real jasperRidge data set
    MethodAEC-NMFTV-RSNMFL1/2NMFVCA-FCLS
    Alunite0.103 10.106 40.092 10.085 9
    Andradite0.087 60.087 80.065 20.058 2
    Buddingtonite0.091 50.096 40.064 80.072 4
    Dumortierite0.098 70.111 20.097 20.097 8
    Kaolinite10.131 20.131 60.126 80.122 2
    Kaolinite20.043 60.044 90.044 00.045 8
    Muscovite0.127 60.127 91.166 71.152 2
    Montmorillonite0.069 60.069 80.072 00.071 7
    Nontronite0.090 70.090 40.117 30.107 0
    Pyrope0.088 00.088 10.189 70.178 3
    Sphene1.013 01.013 00.082 60.087 6
    Chalcedony0.074 00.076 10.191 90.167 5
    Mean0.168 20.170 30.192 50.187 2
    Table 5. SAD values of different methods with the real Cuprite data set
    Xiangxiang JIA, Baofeng GUO, Fanchang DING, Wenjie XU. Hyperspectral Unmixing Based on Constrained Nonnegative Matrix Factorization[J]. Acta Photonica Sinica, 2021, 50(7): 113
    Download Citation