• Acta Photonica Sinica
  • Vol. 49, Issue 6, 0630004 (2020)
Zhi-wei WANG1, Kun TAN1、2、*, Xue WANG1、2, Jian-wei DING3, and Yu CHEN1
Author Affiliations
  • 1Key Laboratory of Land, Environment and Disaster Monitoring, Ministry of Natural Resources, China University of Mining and Technology, Xuzhou, Jiangsu 221116, China
  • 2Key Laboratory of Geographic Information Science, Ministry of Education, East China Normal University, Shanghai 200241, China
  • 3The Second Surveying and Mapping Institute of Hebei, Shijiazhuang, 050037, China
  • show less
    DOI: 10.3788/gzxb20204906.0630004 Cite this Article
    Zhi-wei WANG, Kun TAN, Xue WANG, Jian-wei DING, Yu CHEN. Unsupervised Nearest Regularized Subspace Based on Spectral Space Reconstruction for Hyperspectral Anomaly Detection[J]. Acta Photonica Sinica, 2020, 49(6): 0630004 Copy Citation Text show less
    Band trace curve and trace value matrix at different bands
    Fig. 1. Band trace curve and trace value matrix at different bands
    Comparison of band trace curve before and after denoising and noise map
    Fig. 2. Comparison of band trace curve before and after denoising and noise map
    The model of dual window
    Fig. 3. The model of dual window
    Framework of the proposed method
    Fig. 4. Framework of the proposed method
    AVIRIS dataset
    Fig. 5. AVIRIS dataset
    ROSIS dataset
    Fig. 6. ROSIS dataset
    Avon dataset
    Fig. 7. Avon dataset
    GF-5 Xuzhou dataset
    Fig. 8. GF-5 Xuzhou dataset
    Detection results of four data sets under different window sizes
    Fig. 9. Detection results of four data sets under different window sizes
    Detection results of various detection algorithms in AVIRIS data
    Fig. 10. Detection results of various detection algorithms in AVIRIS data
    Detection results of various detection algorithms in ROSIS data
    Fig. 11. Detection results of various detection algorithms in ROSIS data
    Detection results of various detection algorithms in Avon data
    Fig. 12. Detection results of various detection algorithms in Avon data
    Detection results of various detection algorithms in GF-5 data
    Fig. 13. Detection results of various detection algorithms in GF-5 data
    ROC curves of four datasets
    Fig. 14. ROC curves of four datasets
    σdAVIRISROSISAvonGF-5
    0.10.937 60.993 20.949 10.921 0
    10.937 60.993 20.949 10.922 7
    50.939 10.993 50.950 90.927 4
    100.942 20.995 00.959 10.938 7
    500.951 20.995 40.959 10.956 5
    1000.943 40.995 40.959 10.961 1
    Table 1. Detection results of four datasets at different σd parameters
    λAVIRISROSISAvonGF-5
    0.010.942 70.993 50.960 40.960 5
    0.10.945 60.994 80.961 50.957 1
    10.951 20.995 40.959 10.956 5
    100.950 10.993 60.945 10.959 0
    1000.944 20.992 20.945 10.957 3
    1 0000.942 60.991 90.943 90.956 9
    Table 2. Detection results of four datasets at different λ parameters
    GRXRSADCRDUNRSLSADCRIDWUNRS-SSR
    AVIRISAUC0.833 40.916 50.948 50.984 30.984 60.996 2
    Time/s1.1015.5723.2622.07980.4844.03
    ROSISAUC0.991 60.980 80.911 70.899 20.978 60.999 5
    Time/s1.1525.5250.1947.571 480.94152.65
    AvonAUC0.925 20.953 30.959 50.977 80.94670.995 2
    Time/s8.37442.21161.61171.582 194.29314.34
    GF-5AUC0.971 00.985 60.966 70.969 60.98 020.991 0
    Time/s1.1922.1823.7322.10993.5198.31
    Table 3. Comparison of AUC and execution times of the different methods for four datasets
    Zhi-wei WANG, Kun TAN, Xue WANG, Jian-wei DING, Yu CHEN. Unsupervised Nearest Regularized Subspace Based on Spectral Space Reconstruction for Hyperspectral Anomaly Detection[J]. Acta Photonica Sinica, 2020, 49(6): 0630004
    Download Citation