Journals Articles Conferences News About CLP
Submit a paper Email Alert
Search
Search
Articles
Journals
News
Advanced Search
Top Searches
2D MaterialsTransformation opticsQuantum PhotonicsSmall lasersSemiconductor UV PhotonicsSupersymmetric microring laser arrays

Laser & Optoelectronics Progress, Volume. 55, Issue 11, 111007(2018)

A Pedestrian Detection Method Based on Dark Channel Defogging and Deep Learning

Qing Tian1, Tongyang Yuan1、*, Dan Yang1, and Yun Wei2
Author Affiliations
  • 1 School of Electronic Information Engineering, North China University of Technology, Beijing 100144, China
  • 2 Beijing Urban Construction Design & Development Group Co., Ltd., Beijing 100037, China;
    • show less
    AbstractGet PDF(in Chinese)
    Figures&Tables (12)
    Equations (0)
    References (11)
    Cited By (4)
    Tools
    Paper Information
    Topics
    Figures & Tables(12)
    Structure of ZF network

    Fig. 1. Structure of ZF network

    Download full size
    Structure of RPN

    Fig. 2. Structure of RPN

    Download full size
    (a) Sample label image; (b) label file in XML

    Fig. 3. (a) Sample label image; (b) label file in XML

    Download full size
    Flow chart of dark channel defogging algorithm

    Fig. 4. Flow chart of dark channel defogging algorithm

    Download full size
    Flow chart of Faster R-CNN training algorithm

    Fig. 5. Flow chart of Faster R-CNN training algorithm

    Download full size
    (a) (b) Original images; (c)(d) image processed by dark channel defogging algorithm

    Fig. 6. (a) (b) Original images; (c)(d) image processed by dark channel defogging algorithm

    Download full size
    (a)(c) Test results of model 1; (b)(d) test results of model 2

    Fig. 7. (a)(c) Test results of model 1; (b)(d) test results of model 2

    Download full size
    (a)(c)(e) Test results of model 1; (b)(d)(f) test results of model 2

    Fig. 8. (a)(c)(e) Test results of model 1; (b)(d)(f) test results of model 2

    Download full size
    (a) Test results 1 of model 1; (b) test results 2 of model 1

    Fig. 9. (a) Test results 1 of model 1; (b) test results 2 of model 1

    Download full size
    (a) Test results 1 of model 2; (b) test results 2 of model 2

    Fig. 10. (a) Test results 1 of model 2; (b) test results 2 of model 2

    Download full size

    • Table 1. Detection rate and false alarm rate of model 1 and model 2%

      View table

      Table 1. Detection rate and false alarm rate of model 1 and model 2%

      Evaluation indexTest sample
      Poor qualitytest pictureFog testpictureAverage results withtest image enhancementAverage results withouttest image enhancement
      Detection rate of model 190889089
      Detection rate of model 2939092.591.5
      Flse alarm rate of model 1585.57
      False alarm rate of model 22534

    • Table 2. Detection time of model 1 and model 2

      View table

      Table 2. Detection time of model 1 and model 2

      ModelPixel size
      500×375500×333640×4801280×720
      Detection time of model 1 /s2.380912.455012.478092.490890
      Detection time of model 2 /s2.370172.454962.470912.480745
    Tools

    Get Citation

    Copy Citation Text

    Qing Tian, Tongyang Yuan, Dan Yang, Yun Wei. A Pedestrian Detection Method Based on Dark Channel Defogging and Deep Learning[J]. Laser & Optoelectronics Progress, 2018, 55(11): 111007

    Download Citation

    EndNote(RIS)BibTexPlain Text
    Set citation alerts for article
    Save article for my favorites
    Paper Information

    Category: Image Processing

    Received: Apr. 16, 2018

    Accepted: May. 29, 2018

    Published Online: Aug. 14, 2019

    The Author Email: Tongyang Yuan (2017311020137@mail.ncut.edu.cn)

    DOI:10.3788/LOP55.111007

    Topics

    laser devices and laser physics
    Lasers and Laser Optics
    Laser physics
    laser manufacturing
    Instrumentation, Measurement and Metrology