• Laser & Optoelectronics Progress
  • Vol. 55, Issue 5, 051013 (2018)
Min Wang, Zeyang Li, Chun Wang, and Xinyuan Shi
Author Affiliations
  • School of Information and Control Engineering, Xi'an University of Architecture and Technology, Xi'an, Shaanxi 710055, China
  • show less
    DOI: 10.3788/LOP55.051013 Cite this Article Set citation alerts
    Min Wang, Zeyang Li, Chun Wang, Xinyuan Shi. Key Frame Extraction Algorithm of Sign Language Based on Compressed Sensing and SURF Features[J]. Laser & Optoelectronics Progress, 2018, 55(5): 051013 Copy Citation Text show less
    Matching image of two sign language frame
    Fig. 1. Matching image of two sign language frame
    Key frame extraction algorithm diagram of sign language based on compressed sensing and SURF feature
    Fig. 2. Key frame extraction algorithm diagram of sign language based on compressed sensing and SURF feature
    Partial test videos. (a) Laboratory environment; (b) different environment and light conditions
    Fig. 3. Partial test videos. (a) Laboratory environment; (b) different environment and light conditions
    Time sequence diagram of interframe similarity curve extreme point. (a) Extreme point; (b) key frame
    Fig. 4. Time sequence diagram of interframe similarity curve extreme point. (a) Extreme point; (b) key frame
    Output key frames of three algorithms. (a) Proposed algorithm; (b) non-specific dynamic sign language video key frame detection algorithm; (c) unsupervised clustering algorithm
    Fig. 5. Output key frames of three algorithms. (a) Proposed algorithm; (b) non-specific dynamic sign language video key frame detection algorithm; (c) unsupervised clustering algorithm
    Non-specific part of the key frames
    Fig. 6. Non-specific part of the key frames
    VideosampleNumber ofsubshotsAverage key framenumber /part position
    Test1276/[2, 12, 40, 86, 113, 149]
    Test2276/[7, 26, 51, 76, 100, 132]
    Test3247/[33, 49, 67, 101, 129, 140, 167]
    Test4247/[37, 55, 85, 115, 134, 153, 178]
    Table 1. Subshot segmentation and key frame detection
    ItemProposedalgorithmNon-specific dynamic sign language videokey frame detection algorithmUnsupervisedclustering algorithm
    Average number of key frames7.1259.37512.958
    Key frame leak detection rate /%0.751.753.25
    Average running time /s13.55111.23618.912
    Stability of key frame detection /%98.2593.7582.50
    Table 2. Comparison of experimental data
    Min Wang, Zeyang Li, Chun Wang, Xinyuan Shi. Key Frame Extraction Algorithm of Sign Language Based on Compressed Sensing and SURF Features[J]. Laser & Optoelectronics Progress, 2018, 55(5): 051013
    Download Citation