• Laser & Optoelectronics Progress
  • Vol. 60, Issue 16, 1628002 (2023)
Rujun Chen1, Yunwei Pu1,2,*, Jiahou Zhou1, Jun Li1, and Xuefeng Wang3
Author Affiliations
  • 1Faculty of Land Resources Engineering, Kunming University of Science and Technology, Kunming 650093, Yunnan, China
  • 2Compute Center, Kunming University of Science and Technology, Kunming 650500, Yunnan, China
  • 3Puer 3d Mapping Engineering Co., Ltd., Puer665000, Yunnan, China
  • show less
    DOI: 10.3788/LOP222488 Cite this Article Set citation alerts
    Rujun Chen, Yunwei Pu, Jiahou Zhou, Jun Li, Xuefeng Wang. Small Water Body Extraction Based on GF-2 Image[J]. Laser & Optoelectronics Progress, 2023, 60(16): 1628002 Copy Citation Text show less
    Schematic of the study area
    Fig. 1. Schematic of the study area
    Analysis between brightness value and gray scale value at each band and SWI analysis diagram. (a) Analysis diagram between brightness value and gray scale value at green band (G); (b) analysis diagram between brightness value and gray scale value at red band (R); (c) analysis diagram between brightness value and gray scale value at blue band (B); (d) analysis diagram between brightness value and gray scale value at near-infrared band (NIR); (e) SWI analysis diagram
    Fig. 2. Analysis between brightness value and gray scale value at each band and SWI analysis diagram. (a) Analysis diagram between brightness value and gray scale value at green band (G); (b) analysis diagram between brightness value and gray scale value at red band (R); (c) analysis diagram between brightness value and gray scale value at blue band (B); (d) analysis diagram between brightness value and gray scale value at near-infrared band (NIR); (e) SWI analysis diagram
    Overlay image results after different algorithms distinguish water bodies from shadows and other ground objects. (a) Manual labeled shadow and water body; (b) LGR segmentation result; (c) SWI segmentation result
    Fig. 3. Overlay image results after different algorithms distinguish water bodies from shadows and other ground objects. (a) Manual labeled shadow and water body; (b) LGR segmentation result; (c) SWI segmentation result
    Technology roadmap
    Fig. 4. Technology roadmap
    Water extraction results of different methods. (a) Decision tree; (b) SVM; (c) RF; (d) CNN; (e) NDWI+NIR; (f) proposed method; (g) manual visual interpretation
    Fig. 5. Water extraction results of different methods. (a) Decision tree; (b) SVM; (c) RF; (d) CNN; (e) NDWI+NIR; (f) proposed method; (g) manual visual interpretation
    Sensor typeBand numberWavelength /μmSpatial resolution/m
    Panchromatic camera(2 sets)10.45-0.901(subsatellite point 0.81)
    Multispectral camera(2 sets)4B:0.45-0.524(subsatellite point 3.24)
    G:0.52-0.59
    R:0.63-0.69
    NIR:0.77-0.89
    Table 1. Technical indexes of payload of GF-2 satellite
    ParameterDecision treeSVMRFNDWI+NIRCNNProposed methodManual visual interpretation
    Area /m²6627652616865562373377141186662818266461076731631
    Extraction difference /m2104006562976494294409555103449855240
    Difference ratio /%1.558.367.346.081.541.20
    Table 2. Water extraction area and extraction difference
    MethodPA /%UA /%Hellden accuracy /%Short accuracy /%Kappa coefficient /%
    Decision tree90.2598.594.289.0288.85
    SVM88.7995.8392.1685.4987.15
    RF90.2198.4594.1588.9588.8
    NDWI+NIR92.787.5190.0381.8791.46
    CNN93.449594.289.0692.42
    Proposed method95.5697.4196.4793.294.86
    Table 3. Water extraction accuracy of various methods
    Rujun Chen, Yunwei Pu, Jiahou Zhou, Jun Li, Xuefeng Wang. Small Water Body Extraction Based on GF-2 Image[J]. Laser & Optoelectronics Progress, 2023, 60(16): 1628002
    Download Citation