Laser & Optoelectronics Progress, Volume. 60, Issue 2, 0210014(2023)

Video Skin-Color Enhancement Method Based on Video-Guided Model Updates

Shaobo Ding*, Yali Zhang, and Kun Zhang
Author Affiliations
  • School of Microelectronics, Tianjin University, Tianjin 300072, China
  • show less
    Figures & Tables(17)
    System structure of the proposed method
    Common shot boundary types. (a) Hard transition; (b) soft transition
    Confirmation process of HT
    Confirmation process of ST
    Variation of shot boundary detection precision with fusion weight
    Skin color sampling process. (a) 68 feature points; (b) feature point number and area division; (c) calculating the midpoint coordinates to determine the reference point; (d) 210 skin-color reference points
    Distribution of human skin color in the YCbCr color space
    Skin mask merging
    Skin-color classification results for different connected domains in the current frame
    Distribution of preferred skin color of three races. (a) Asian; (b) Caucasian; (c) African
    Distribution of preferred skin color of Asian. (a) Distribution of 30 points in the YCbCr color space; (b) 1D plane fitting; (c) 2D quadric surface fitting; (d) curve determined by intersection of the 1D plane and the 2D quadric surface
    Comparison of visual continuity. (a) Original frames; (b) Ellipse mapping method; (c) proposed method
    Skin color enhancement. (a) Asian; (b) Caucasian; (c) African
    • Table 1. Distribution of skin color and average chrominance of three races at each luminance level

      View table

      Table 1. Distribution of skin color and average chrominance of three races at each luminance level

      Luminance levelAsianCaucasianAfrican
      Frequency /%DCb¯DCr¯Frequency /%DCb¯DCr¯Frequency /%DCb¯DCr¯
      13.41171409.112314016.6125132
      212.611014721.510914538.5120143
      334.910714533.410115536.7111151
      438.410514928.11061535.6118140
      510.71171407.91161402.6123131
    • Table 2. Update times of different skin models

      View table

      Table 2. Update times of different skin models

      ModelMoviesTV DramasSports programsReality showsTotal
      Guided by video3322849051331654
      Frame by frame9547941115462916443584
    • Table 3. Execution time of each module

      View table

      Table 3. Execution time of each module

      MethodExecution time /msTotal time /ms

      Speed /

      (frams·s-1

      Shot boundary detectionSkin segmentationColor enhancement
      Skin color centerNone14.0312.0626.0938.32
      Ellipse mappingNone13.4511.9925.4439.31
      Proposed method1.539.1212.1422.7943.87
    • Table 4. Mean opinion score for subjective assessment of test video

      View table

      Table 4. Mean opinion score for subjective assessment of test video

      IndexMOS
      Original videoSkin color centerEllipse mappingProposed method
      15.15.95.26.4
      25.25.15.65.8
      35.65.65.45.7
      44.84.44.24.9
      54.94.24.35.4
      64.54.04.34.7
      74.64.94.75.2
      86.96.86.47.2
      95.86.36.16.6
      106.87.07.47.8
    Tools

    Get Citation

    Copy Citation Text

    Shaobo Ding, Yali Zhang, Kun Zhang. Video Skin-Color Enhancement Method Based on Video-Guided Model Updates[J]. Laser & Optoelectronics Progress, 2023, 60(2): 0210014

    Download Citation

    EndNote(RIS)BibTexPlain Text
    Save article for my favorites
    Paper Information

    Category: Image Processing

    Received: Feb. 8, 2022

    Accepted: Mar. 14, 2022

    Published Online: Jan. 6, 2023

    The Author Email: Shaobo Ding (akat@tju.edu.cn)

    DOI:10.3788/LOP220685

    Topics