• Laser & Optoelectronics Progress
  • Vol. 58, Issue 20, 2015001 (2021)
Qi Meng1, Hua Miao1、*, Lin Li1, Bo Guo1, Tingting Liu1, and Shilong Mi2
Author Affiliations
  • 1School of Opto-Electronic Engineering, Changchun University of Science and Technology, Changchun, Jilin 130022, China
  • 2R&D Center, Dongguan Yutong Optical Technology Co.,Ltd, Dongguan, Guangdong 523841, China;
  • show less
    DOI: 10.3788/LOP202158.2015001 Cite this Article Set citation alerts
    Qi Meng, Hua Miao, Lin Li, Bo Guo, Tingting Liu, Shilong Mi. Data Enhancement of Lens Defect Based on Dual Channel Generative Adversarial Networks[J]. Laser & Optoelectronics Progress, 2021, 58(20): 2015001 Copy Citation Text show less
    Framework for generator
    Fig. 1. Framework for generator
    Framework for discriminator
    Fig. 2. Framework for discriminator
    Overall workflow
    Fig. 3. Overall workflow
    Dataset acquisition system and results. (a) Machine vision lens defect detection system; (b) pretreatment of segmented lenses; (c) defect labeling
    Fig. 4. Dataset acquisition system and results. (a) Machine vision lens defect detection system; (b) pretreatment of segmented lenses; (c) defect labeling
    Results of lens defect image generated by DualC-GAN
    Fig. 5. Results of lens defect image generated by DualC-GAN
    Comparison of generation of lens defect images
    Fig. 6. Comparison of generation of lens defect images
    No.DefectLabelCount
    1BubbleBubble332
    2ScratchScratch303
    3SpotSpot325
    4SmudgeSmudge318
    Table 1. Annotation of defect types
    Method1-NNMMDWD
    CGAN0.780.304.36
    WGAN-GP0.720.273.52
    Patch GAN0.610.183.12
    DualC-GAN0.520.152.81
    Table 2. Quality evaluation of each model
    Qi Meng, Hua Miao, Lin Li, Bo Guo, Tingting Liu, Shilong Mi. Data Enhancement of Lens Defect Based on Dual Channel Generative Adversarial Networks[J]. Laser & Optoelectronics Progress, 2021, 58(20): 2015001
    Download Citation