• Laser & Optoelectronics Progress
  • Vol. 60, Issue 2, 0228009 (2023)
Mengjia Niu1, Yongjun Zhang1,*, Zhi Li1, Gang Yang2..., Zhongwei Cui3 and Junwen Liu1|Show fewer author(s)
Author Affiliations
  • 1College of Computer Science and Technology, Guizhou University, Guiyang 550025, Guizhou, China
  • 2Guiyang Orbita Aerospace Science&Technology Co., Ltd., Guiyang 550027, Guizhou, China
  • 3Big Data Science and Intelligent Engineering Research Institute, Guizhou Education University, Guiyang 550018, Guizhou, China
  • show less
    DOI: 10.3788/LOP220525 Cite this Article Set citation alerts
    Mengjia Niu, Yongjun Zhang, Zhi Li, Gang Yang, Zhongwei Cui, Junwen Liu. Remote Sensing Image Segmentation Network Based on Adaptive Multiscale and Contour Gradient[J]. Laser & Optoelectronics Progress, 2023, 60(2): 0228009 Copy Citation Text show less
    Framework of remote sensing image semantic segmentation algorithm based on fused contour learning with deep convolutional neural network
    Fig. 1. Framework of remote sensing image semantic segmentation algorithm based on fused contour learning with deep convolutional neural network
    Multi-channel network framework based on attention mechanism
    Fig. 2. Multi-channel network framework based on attention mechanism
    Diagrams of internal structure of MA block. (a) Internal structure of a single MA block; (b) SK weight module
    Fig. 3. Diagrams of internal structure of MA block. (a) Internal structure of a single MA block; (b) SK weight module
    Diagram of internal structure of D-MA block
    Fig. 4. Diagram of internal structure of D-MA block
    Visualization of ablation experiments compared in Vaihingen test set
    Fig. 5. Visualization of ablation experiments compared in Vaihingen test set
    Comparison of prediction results with mainstream models on Vaihingen test set
    Fig. 6. Comparison of prediction results with mainstream models on Vaihingen test set
    Comparison of prediction results with mainstream models on Potsdam test set
    Fig. 7. Comparison of prediction results with mainstream models on Potsdam test set
    Comparison of prediction results with mainstream models on WHU building test set
    Fig. 8. Comparison of prediction results with mainstream models on WHU building test set
    LayerOutput sizeOperatorStrideSize
    Branch-1256×256,64MA block11
    Branch-2256×256,64MA block11
    Down layer-1128×128,64Conv-block21/2
    Branch-3128×128,64MA block11/2
    Branch-4128×128,64MA block11/2
    Down layer-264×64,64Conv-block21/4
    Branch-564×64,128MA block11/4
    Table 1. Configuration parameters for profile extraction module
    ModelmIoU /%OA /%F1 /%ParamsFLOPs /109
    SegNet76.5087.9686.9920.7327.56
    Improved SegNet80.6988.6389.8120.9128.29
    with D-MMA84.2889.9390.3621.8231.39
    with D- MMA +CME86.5692.9392.5122.6934.36
    Table 2. Model ablation experiments on Vaihingen dataset
    ModelIoUF1OAmIoU
    impervious surfacesbuildinglow vegetationtreecar
    U-Net79.4585.2364.9374.7738.5186.9485.4368.58
    SegNet81.6986.4173.4378.3642.6386.9987.9676.50
    ERFNet77.5179.2762.3571.2735.2983.5782.0965.13
    PSPNet87.6991.9481.5284.7955.7989.8690.1981.16
    DSMNet2590.8291.5
    Fres- MFDNN2692.091.085.0
    Proposed model91.8993.6186.7288.6971.9392.5192.9386.56
    Table 3. Comparison with other networks on Vaihingen dataset
    ModelIoUF1OAmIoU
    impervious surfacesbuildinglow vegetationtreecar
    U-Net76.4483.2265.9258.0371.8783.1277.4571.10
    SegNet83.6391.7274.7071.6778.0085.3187.4577.94
    ERFNet61.4674.7851.8345.8517.0780.5772.3550.20
    PSPNet83.5992.9976.2873.0977.1185.7889.7880.61
    BAM-Unet-sc2788.5989.13
    ResUNet-a392.0991.50
    Proposed model85.3993.6478.8276.4881.5791.6592.1883.18
    Table 4. Comparison with other networks on Potsdam dataset
    ModelF1OAmIoU
    U-Net89.3387.5983.01
    SegNet93.7392.5388.76
    ERFNe89.3387.5978.72
    PSPNet93.2892.1587.26
    DeNet2894.8090.12
    MA-FCN2995.1590.70
    Proposed model95.8194.0692.30
    Table 5. Comparison with other network indicators on WHU building dataset
    Mengjia Niu, Yongjun Zhang, Zhi Li, Gang Yang, Zhongwei Cui, Junwen Liu. Remote Sensing Image Segmentation Network Based on Adaptive Multiscale and Contour Gradient[J]. Laser & Optoelectronics Progress, 2023, 60(2): 0228009
    Download Citation