• Laser & Optoelectronics Progress
  • Vol. 61, Issue 16, 1611009 (2024)
Yu Meng, Manchao Bao, Tao Yue**, and Xuemei Hu*
Author Affiliations
  • School of Electronic Science and Engineering, Nanjing University, Nanjing 210033, Jiangsu, China
  • show less
    DOI: 10.3788/LOP241380 Cite this Article Set citation alerts
    Yu Meng, Manchao Bao, Tao Yue, Xuemei Hu. Meta Light Field Event Imaging System (Invited)[J]. Laser & Optoelectronics Progress, 2024, 61(16): 1611009 Copy Citation Text show less
    General view of meta light-field event imaging system
    Fig. 1. General view of meta light-field event imaging system
    Aggregation diagram of geometrical multi-perspective weighted event
    Fig. 2. Aggregation diagram of geometrical multi-perspective weighted event
    Event queue manager algorithm
    Fig. 3. Event queue manager algorithm
    Event queue manager algorithm flow
    Fig. 4. Event queue manager algorithm flow
    Schematic diagram of simulation scenes. (a) Pedestrian scene; (b) light field imaging result of pedestrian scene; (c) ball drop scene; (d) light field imaging result of ball drop scene
    Fig. 5. Schematic diagram of simulation scenes. (a) Pedestrian scene; (b) light field imaging result of pedestrian scene; (c) ball drop scene; (d) light field imaging result of ball drop scene
    Simulation scenes and reconstruction results. (a) Ball slip and (b) pedestrian; meta light field event reconstruction results of (c) ball slip and (d) pedestrian
    Fig. 6. Simulation scenes and reconstruction results. (a) Ball slip and (b) pedestrian; meta light field event reconstruction results of (c) ball slip and (d) pedestrian
    Hardware system view of meta light field event camera. (a) Side view of hardware system and corresponding components;(b) micro-lens placed after main lens.
    Fig. 7. Hardware system view of meta light field event camera. (a) Side view of hardware system and corresponding components;(b) micro-lens placed after main lens.
    Deniosing results of different methods. (a) Original event data; (b) Ynoise filter; (c) Knoise filter; (d) proposed method
    Fig. 8. Deniosing results of different methods. (a) Original event data; (b) Ynoise filter; (c) Knoise filter; (d) proposed method
    Results of proposed method of removing manually added noise events.Results after (a) adding 300 noise events and (b) results after processing with proposed method; results after (c) adding 2000 noise events and (d) results after processing with proposed method
    Fig. 9. Results of proposed method of removing manually added noise events.Results after (a) adding 300 noise events and (b) results after processing with proposed method; results after (c) adding 2000 noise events and (d) results after processing with proposed method
    Reconstruction result of proposed method. (a) Imaging scheme; (b) reconstruction results with different numbers
    Fig. 10. Reconstruction result of proposed method. (a) Imaging scheme; (b) reconstruction results with different numbers
    Results of multi-view perception of scene by meta light-field event imaging system. (a) Intensity frame; (b) accumulation frame
    Fig. 11. Results of multi-view perception of scene by meta light-field event imaging system. (a) Intensity frame; (b) accumulation frame
    Image of two balls thrown at different distances. (a) Experimental setup of motion tracking; (b) results of motion trajectory
    Fig. 12. Image of two balls thrown at different distances. (a) Experimental setup of motion tracking; (b) results of motion trajectory
    ParameterSetting
    Local buffer activation threshold5
    Range of weight spreading3×3×3
    Length of buffer queue1500
    Weighting threshold of voting4.5
    Distribution of weight spreading0‒1 Gaussian
    Table 1. Processing parameters for simulation data
    SceneTime window /msGround truth /cmDepth STD /cmAverage error /cm
    Ball slip10541.5763.7962
    Pedestrian10541.5771.1260
    Table 2. Error analysis of simulation test
    ParameterSetting
    Local buffer activation threshold5
    Range of weight spreading3×3×3
    Length of buffer queue200
    Weighting threshold of voting4.5
    Distribution of weight spreading0‒1 Gaussian
    Table 3. Processing parameters for real data
    NiEiEoNoErrNRR
    300650895070.14590.023
    5006708900190.13410.038
    80070081034440.14750.055
    10007208996330.13810.033
    120074081005600.13560.050
    150077081020480.13230.032
    180080081007720.12570.040
    20008208992460.12080.023
    Table 4. Denoising results of proposed method
    Yu Meng, Manchao Bao, Tao Yue, Xuemei Hu. Meta Light Field Event Imaging System (Invited)[J]. Laser & Optoelectronics Progress, 2024, 61(16): 1611009
    Download Citation