• Laser & Optoelectronics Progress
  • Vol. 58, Issue 12, 1211002 (2021)
Caidong Wang1、*, Fengyang Liu1, Zhihang Li1, Zhihong Chen2, Yan Cheng1, and Huadong Zheng1
Author Affiliations
  • 1College of Mechanical and Electrical Engineering, Zhengzhou University of Light Industry, Zhengzhou, Henan 450002, China
  • 2Zhengzhou Kehui Technology Co., Ltd., Zhengzhou, Henan 450001, China
  • show less
    DOI: 10.3788/LOP202158.1211002 Cite this Article Set citation alerts
    Caidong Wang, Fengyang Liu, Zhihang Li, Zhihong Chen, Yan Cheng, Huadong Zheng. Research on Image Mosaic Method Based on Binocular Vision Feature Point Matching[J]. Laser & Optoelectronics Progress, 2021, 58(12): 1211002 Copy Citation Text show less
    Image mosaic process based on feature point matching
    Fig. 1. Image mosaic process based on feature point matching
    Principle of area measurement of characteristic region of workpiece
    Fig. 2. Principle of area measurement of characteristic region of workpiece
    Platform of flexible vision detection system
    Fig. 3. Platform of flexible vision detection system
    Relationship between feature dimension and matching rate
    Fig. 4. Relationship between feature dimension and matching rate
    Results of image mosaic process. (a) Detection results of feature points; (b) coarse matching of feature points; (c) precision matching of feature points; (d) mosaic image
    Fig. 5. Results of image mosaic process. (a) Detection results of feature points; (b) coarse matching of feature points; (c) precision matching of feature points; (d) mosaic image
    Mosaic image after fusion
    Fig. 6. Mosaic image after fusion
    Mosaic imageⅠ
    Fig. 7. Mosaic imageⅠ
    Mosaic imageⅡ
    Fig. 8. Mosaic imageⅡ
    Complete workpiece image
    Fig. 9. Complete workpiece image
    Process of screening characteristic areas. (a) Binary image Ⅰ; (b) binary image Ⅱ; (c) binary image Ⅲ; (d) connected domain after screening
    Fig. 10. Process of screening characteristic areas. (a) Binary image Ⅰ; (b) binary image Ⅱ; (c) binary image Ⅲ; (d) connected domain after screening
    Labeling of geometric figures
    Fig. 11. Labeling of geometric figures
    Contrast curve of radius of workpiece circular hole
    Fig. 12. Contrast curve of radius of workpiece circular hole
    Relative error of radius of circular hole
    Fig. 13. Relative error of radius of circular hole
    ItemParameter
    Highest resolution /(pixel×pixel)1600×1200
    Pixel size /(μm×μm)4.4×4.4
    Sensor typeCCD
    Optical size /(″)1/1.8
    Lens focal length /mm16
    Aperture adjustment rangeF1.4-F32.0
    Photographic lens imaging size /(″)2/3
    Frame rate /(frame·s-1)20
    Exposure time /μs16-100000
    Output colorBlack and white
    Output modeGigEgigabit ethernet output
    Transmission distance/m100
    Synchronous modeExternal trigger or continuous acquisition
    Table 1. Parameters of MV-EM200M camera
    AlgorithmFeature points(Left/right)Matched logarithmCorrect matched logarithmCorrect matching rate /%Matched time /ms
    Improved SIFT300/20017814480.90187
    SIFT300/2001317456.49251
    ORB300/20015410668.83235
    Table 2. Test results for different algorithms
    Caidong Wang, Fengyang Liu, Zhihang Li, Zhihong Chen, Yan Cheng, Huadong Zheng. Research on Image Mosaic Method Based on Binocular Vision Feature Point Matching[J]. Laser & Optoelectronics Progress, 2021, 58(12): 1211002
    Download Citation