• Infrared and Laser Engineering
  • Vol. 49, Issue 7, 20200012 (2020)
Yang Zhao, Jia''an Fu, Haotian Yu, Jing Han, and Dongliang Zheng
Author Affiliations
  • 南京理工大学 电子工程与光电技术学院,江苏 南京 210094
  • show less
    DOI: 10.3788/IRLA20200012 Cite this Article
    Yang Zhao, Jia''an Fu, Haotian Yu, Jing Han, Dongliang Zheng. Defocus projection three-dimensional measurement based on deep learning accurate phase acquisition[J]. Infrared and Laser Engineering, 2020, 49(7): 20200012 Copy Citation Text show less
    Schematic diagram of conventional high-speed 3D measurement
    Fig. 1. Schematic diagram of conventional high-speed 3D measurement
    Schematic of accurate phase acquisition. (a) Phase with noise;(b) proposed PDNet;(c) result by the proposed method
    Fig. 2. Schematic of accurate phase acquisition. (a) Phase with noise;(b) proposed PDNet;(c) result by the proposed method
    Structure of the PDNet
    Fig. 3. Structure of the PDNet
    Simulation results. (a) Phase map with noise;(b) result by the proposed method;(c) comparison of the phase in the 100th column from Fig. 4(a),4(b) and the ground truth;(d) plotting error curves of the 100th column of Fig. 4(b) and 4(c)
    Fig. 4. Simulation results. (a) Phase map with noise;(b) result by the proposed method;(c) comparison of the phase in the 100th column from Fig. 4(a),4(b) and the ground truth;(d) plotting error curves of the 100th column of Fig. 4(b) and 4(c)
    Phase error of simulated wrapped phase at T = 48,54,60,66,72,78
    Fig. 5. Phase error of simulated wrapped phase at T = 48,54,60,66,72,78
    Experiment results. (a) Phase map with noise;(b) result by the proposed method;(c) comparison of the phase in the 100th column from Fig. 6(a), 6(b) and the ground truth;(d) plotting error curves of the 100th column of Fig. 6(b) and 6(c)
    Fig. 6. Experiment results. (a) Phase map with noise;(b) result by the proposed method;(c) comparison of the phase in the 100th column from Fig. 6(a), 6(b) and the ground truth;(d) plotting error curves of the 100th column of Fig. 6(b) and 6(c)
    Phase error of actual measurement of wrapped phase at T= 48, 54, 60, 66, 72, 78
    Fig. 7. Phase error of actual measurement of wrapped phase at T= 48, 54, 60, 66, 72, 78
    Wrapping phase map before and after denoising and its error comparison. (a) Phase map with noise;(b) result by the proposed method;(c) phase error before denoising;(d) phase error after denoising
    Fig. 8. Wrapping phase map before and after denoising and its error comparison. (a) Phase map with noise;(b) result by the proposed method;(c) phase error before denoising;(d) phase error after denoising
    Comparison of reconstruction results. (a) Reconstruction results before processing using the proposed method;(b) ideal reconstruction results;(c) reconstruction result by the proposed method;(d) reconstruction error curve
    Fig. 9. Comparison of reconstruction results. (a) Reconstruction results before processing using the proposed method;(b) ideal reconstruction results;(c) reconstruction result by the proposed method;(d) reconstruction error curve
    TypeOut-FOut-Res
    ENCODER1Downsampler block16512×256
    2Downsampler block64256×128
    3−75 × Non-bt-1D64256×128
    8Downsampler block128128×64
    9Non-bt-1D(dilated 2)128128×64
    10Non-bt-1D(dilated 4)128128×64
    11Non-bt-1D(dilated 8)128128×64
    12Non-bt-1D(dilated 16)128128×64
    13Non-bt-1D(dilated 2)128128×64
    14Non-bt-1D(dilated 4)128128×64
    15Non-bt-1D(dilated 8)128128×64
    16Non-bt-1D(dilated 16)128128×64
    DECODER17Deconvolution (unsampling)64256×128
    18−192 × Non-bt-1D64256×128
    20Deconvolution (unsampling)16512×256
    21−222 × Non-bt-1D16512×256
    23Deconvolution (unsampling)C1 024×512
    Table 1. Main network modules and parameters
    Yang Zhao, Jia''an Fu, Haotian Yu, Jing Han, Dongliang Zheng. Defocus projection three-dimensional measurement based on deep learning accurate phase acquisition[J]. Infrared and Laser Engineering, 2020, 49(7): 20200012
    Download Citation