• Laser & Optoelectronics Progress
  • Vol. 56, Issue 22, 222801 (2019)
Xinlei Ren1、*, Yangping Wang1、2、4, Jingyu Yang1、3, and Decheng Gao4
Author Affiliations
  • 1School of Electronic and Information Engineering, Lanzhou Jiaotong University, Lanzhou, Gansu 730070, China
  • 2Experiment Teaching Center on Computer Science, Lanzhou Jiaotong University, Lanzhou, Gansu 730070, China
  • 3Gansu Provincial Engineering Research Center for Artificial Intelligence and Graphics & Image Processing, Lanzhou Jiaotong University, Lanzhou, Gansu 730070, China;
  • 4Gansu Provincial Key Laboratory of System Dynamics and Reliability of Rail Transport Equipment, Lanzhou Jiaotong University, Lanzhou, Gansu 730070, China
  • show less
    DOI: 10.3788/LOP56.222801 Cite this Article Set citation alerts
    Xinlei Ren, Yangping Wang, Jingyu Yang, Decheng Gao. Building Detection from Remote Sensing Images Based on Improved U-net[J]. Laser & Optoelectronics Progress, 2019, 56(22): 222801 Copy Citation Text show less
    Architecture of improved U-net network with enhanced low-dimensional features
    Fig. 1. Architecture of improved U-net network with enhanced low-dimensional features
    Architecture of fully convolutional neural networks
    Fig. 1. Architecture of fully convolutional neural networks
    Architecture of U-net network
    Fig. 2. Architecture of U-net network
    Flow chart of building detection using proposed method
    Fig. 4. Flow chart of building detection using proposed method
    Image of building detection area
    Fig. 5. Image of building detection area
    Results of building detection using different methods
    Fig. 6. Results of building detection using different methods
    Local results of building detection using different methods
    Fig. 7. Local results of building detection using different methods
    Test areaIoUPAKappa
    FCMFCNU-netImprovedU-netFCMFCNU-netImprovedU-netFCMFCNU-netImprovedU-net
    166.467.868.873.883.690.785.692.470.364.273.380.0
    269.384.887.591.683.089.192.994.472.279.987.388.8
    373.578.588.090.188.494.294.297.376.373.084.584.7
    470.679.977.280.195.281.785.987.275.084.078.180.8
    Mean69.977.780.483.987.588.989.692.873.575.380.883.6
    Table 1. Quantitative evaluation results of building detection for different methods%
    Xinlei Ren, Yangping Wang, Jingyu Yang, Decheng Gao. Building Detection from Remote Sensing Images Based on Improved U-net[J]. Laser & Optoelectronics Progress, 2019, 56(22): 222801
    Download Citation