• Acta Optica Sinica
  • Vol. 38, Issue 6, 0620001 (2018)
Fang Liu*, Lixia Lu, Guangwei Huang, Hongjuan Wang, and Xin Wang
Author Affiliations
  • Faculty of Information Technology, Beijing University of Technology, Beijing 100022, China
  • show less
    DOI: 10.3788/AOS201838.0620001 Cite this Article Set citation alerts
    Fang Liu, Lixia Lu, Guangwei Huang, Hongjuan Wang, Xin Wang. Landform Image Classification Based on Discrete Cosine Transformation and Deep Network[J]. Acta Optica Sinica, 2018, 38(6): 0620001 Copy Citation Text show less
    (a) Original image and (b) energy distribution after DCT
    Fig. 1. (a) Original image and (b) energy distribution after DCT
    Structure of CNN
    Fig. 2. Structure of CNN
    Three-dimensional spectrum diagram. (a) DCT coefficient spectrum of original image; (b) spectrum after the coefficient selection
    Fig. 3. Three-dimensional spectrum diagram. (a) DCT coefficient spectrum of original image; (b) spectrum after the coefficient selection
    Structure of DCT-CNN model
    Fig. 4. Structure of DCT-CNN model
    Flow chart of landform image classification algorithm based on DCT and deep network
    Fig. 5. Flow chart of landform image classification algorithm based on DCT and deep network
    Example images in database. (a) UC Merced LU database; (b) UAV landing landform database
    Fig. 6. Example images in database. (a) UC Merced LU database; (b) UAV landing landform database
    Classification performance of each method when the number of training samples is different. (a) UC Merced LU database; (b) UAV landing landform database
    Fig. 7. Classification performance of each method when the number of training samples is different. (a) UC Merced LU database; (b) UAV landing landform database
    Image classification confusion matrix for UAV landing landform database
    Fig. 8. Image classification confusion matrix for UAV landing landform database
    LayerTypePatch sizeStrideZero paddingOutput size
    xInput128×128
    h1Convolution5×512128×128×32
    h2ReLU
    h3Mean pooling3×3264×64
    h4Convolution3×32032×32×32
    h5ReLU
    h6Max pooling3×3216×16
    h7Convolution7×71214×14×64
    h8ReLU
    h9Max pooling3×327×7
    h10Convolution7×7101×1×64
    h11ReLU
    h12Convolution1×1101×1×10
    oSVMn(class)
    Table 1. Layer parameters of DCT-CNN network structure
    MethodAccuracy /%SDTraining time /h
    Method 184.250.780.8
    Method 295.760.281.0
    Method 392.830.523.3
    Table 2. Effect of different methods on classification of UC Merced LU database
    MethodAccuracy/%SDTraining time /h
    Method 183.730.851.0
    Method 294.380.341.3
    Method 392.100.613.9
    Table 3. Effect of different methods on classification of UVA landing landform database
    MethodAccuracy /%SD
    RF79.250.82
    LDA-RF82.920.69
    CS-CNN[5]92.860.59
    PSR[15]89.100.69
    MS-DCNN[16]91.340.63
    DCT-CNN95.760.28
    Table 4. Comparison of the classification accuracy of different methods for UC Merced LU database
    MethodAccuracy /%SD
    RF77.100.70
    LDA-RF80.230.74
    CS-CNN[5]91.780.62
    DCT-SAE[12]86.490.96
    MS-DCNN[16]90.160.71
    DCT-CNN94.380.34
    Table 5. Comparison of the classification accuracy of different methods for UAV landing landform database
    Fang Liu, Lixia Lu, Guangwei Huang, Hongjuan Wang, Xin Wang. Landform Image Classification Based on Discrete Cosine Transformation and Deep Network[J]. Acta Optica Sinica, 2018, 38(6): 0620001
    Download Citation