• Laser & Optoelectronics Progress
  • Vol. 57, Issue 8, 081016 (2020)
Lingli Xu, Xiaopo Zhu, Yixing Hou, Min Li, and Xuewu Zhang*
Author Affiliations
  • College of Internet of Things Engineering, Hohai University, Changzhou, Jiangsu 213022, China
  • show less
    DOI: 10.3788/LOP57.081016 Cite this Article Set citation alerts
    Lingli Xu, Xiaopo Zhu, Yixing Hou, Min Li, Xuewu Zhang. Culvert Crack Defect Segmentation Algorithm Based on Enhanced Hue Features[J]. Laser & Optoelectronics Progress, 2020, 57(8): 081016 Copy Citation Text show less
    Flow chart of hue feature enhancement and extraction
    Fig. 1. Flow chart of hue feature enhancement and extraction
    Comparison of guided filtering effect. (a) Unfiltered blue channel; (b) partial enlargement of (a); (c) traditional guided filtering effect; (d) partial enlargement of (c); (e) proposed guided filtering effect; (f) partial enlargement of (e)
    Fig. 2. Comparison of guided filtering effect. (a) Unfiltered blue channel; (b) partial enlargement of (a); (c) traditional guided filtering effect; (d) partial enlargement of (c); (e) proposed guided filtering effect; (f) partial enlargement of (e)
    Rough segmentation results. (a) Enhanced hue feature map; (b) rough segmentation result
    Fig. 3. Rough segmentation results. (a) Enhanced hue feature map; (b) rough segmentation result
    Comparison of threshold segmentation effect. (a) Original images; (b) in Ref. [15]; (c) maximum entropy segmentation; (d) local AT method; (e) our rough segmentation
    Fig. 4. Comparison of threshold segmentation effect. (a) Original images; (b) in Ref. [15]; (c) maximum entropy segmentation; (d) local AT method; (e) our rough segmentation
    Comparison of segmentation effect. (a) Original images; (b)watershed segmentation algorithm; (c) in Ref. [16]; (d) our rough segmentation
    Fig. 5. Comparison of segmentation effect. (a) Original images; (b)watershed segmentation algorithm; (c) in Ref. [16]; (d) our rough segmentation
    Segmentation results of proposed method. (a) Segmentation results based on figure 4(e); (b) segmentation results based on figure 5(d)
    Fig. 6. Segmentation results of proposed method. (a) Segmentation results based on figure 4(e); (b) segmentation results based on figure 5(d)
    TypeItemValue
    SoftwareOperating systemWin 10
    HardwareCPUIntel(R) Core(TM) i7-8700 CPU @3.20 GHz 3.19 GHz
    RAM16 G
    Guided filtering radius20
    Experimental parametersMean filtering window size1/4 of the image size
    Median filtering kernel size5
    Table 1. Experimental parameters
    ImageMethodSe /%Sp /%Ac /%
    Ref. [16]70.72695.60091.467
    1AT10.48699.91185.052
    Proposed method82.05096.62394.217
    Ref. [16]76.90597.92694.403
    2AT44.18199.73090.421
    Proposed method73.22199.84895.418
    Ref. [16]85.24551.72653.597
    3AT54.32299.42096.902
    Proposed method94.20297.70897.514
    Table 2. Image segmentation performance qualitative index
    Lingli Xu, Xiaopo Zhu, Yixing Hou, Min Li, Xuewu Zhang. Culvert Crack Defect Segmentation Algorithm Based on Enhanced Hue Features[J]. Laser & Optoelectronics Progress, 2020, 57(8): 081016
    Download Citation