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

Journal of Infrared and Millimeter Waves, Volume. 40, Issue 1, 122(2021)

Point target detection based on deep spatial-temporal convolution neural network

Miao LI1、*, Zai-Ping LIN1, Jian-Peng FAN1, Wei-Dong SHENG1, Jun LI1, Wei AN1, and Xin-Lei LI2
Author Affiliations
  • 1College of electronic science and technology，National University of Defense Technology，Changsha 410073，China
  • 2The Xian Chinese Space Tracking Control Center，Xian，Shanxi 710000，China
    • show less
    AbstractGet PDF(in Chinese)
    Figures&Tables (19)
    Equations (12)
    References (30)
    Tools
    Paper Information
    Topics
    Figures & Tables(19)
    The proposed network architecture.

    Fig. 1. The proposed network architecture.

    Download full size
    View in Article
    The sketches of 3D covolution and factorized 3D convolution: (a) 3D covolution; (b) factorized 3D convolution.

    Fig. 2. The sketches of 3D covolution and factorized 3D convolution: (a) 3D covolution; (b) factorized 3D convolution.

    Download full size
    View in Article
    The example of different error distributions: (a) the ground truth; (b) 1th predicted result with uniform error; (c) 2th predicted result with concentrated error.

    Fig. 3. The example of different error distributions: (a) the ground truth; (b) 1th predicted result with uniform error; (c) 2th predicted result with concentrated error.

    Download full size
    View in Article
    The function of intensity weighting parameter.

    Fig. 4. The function of intensity weighting parameter.

    Download full size
    View in Article
    The example of samples: (a) the target samples; (b) the background samples.

    Fig. 5. The example of samples: (a) the target samples; (b) the background samples.

    Download full size
    View in Article
    The original image and results of different methods for 1th Background.: (a) the original input; (b) the result of our method; (c) the result of Lin’s method; (d) the result of Max-Mean; (e) the result of TopHat; (f) the result of STDA.

    Fig. 6. The original image and results of different methods for 1th Background.: (a) the original input; (b) the result of our method; (c) the result of Lin’s method; (d) the result of Max-Mean; (e) the result of TopHat; (f) the result of STDA.

    Download full size
    View in Article
    The original image and results of different methods for 2th Background.: (a) the original input; (b) the result of our method; (c) the result of Lin’s method; (d) the result of Max-Mean; (e) the result of TopHat; (f) the result of STDA.

    Fig. 7. The original image and results of different methods for 2th Background.: (a) the original input; (b) the result of our method; (c) the result of Lin’s method; (d) the result of Max-Mean; (e) the result of TopHat; (f) the result of STDA.

    Download full size
    View in Article
    The original image and results of different methods for Target 1: (a) the original input; (b) the result of our method; (c) the result of Lin’s method; (d) the result of Max-Mean; (e) the result of TopHat; (f) the result of STDA.

    Fig. 8. The original image and results of different methods for Target 1: (a) the original input; (b) the result of our method; (c) the result of Lin’s method; (d) the result of Max-Mean; (e) the result of TopHat; (f) the result of STDA.

    Download full size
    View in Article
    The original image and results of different methods for Target 2: (a) the original input; (b) the result of our method; (c) the result of Lin’s method; (d) the result of Max-Mean; (e) the result of TopHat; (f) the result of STDA.

    Fig. 9. The original image and results of different methods for Target 2: (a) the original input; (b) the result of our method; (c) the result of Lin’s method; (d) the result of Max-Mean; (e) the result of TopHat; (f) the result of STDA.

    Download full size
    View in Article
    The display of Target 2: (a) the original gray image; (b) the standard deviation in the time domain.

    Fig. 10. The display of Target 2: (a) the original gray image; (b) the standard deviation in the time domain.

    Download full size
    View in Article
    The ROC curves of different methods.

    Fig. 11. The ROC curves of different methods.

    Download full size
    View in Article
    The result of different input size: (a) the input image with 35×35 pixels; (b) the result of image with35×35 pixels; (c) the input image with 45×45 pixels; (b) the result of image with45×45 pixels.

    Fig. 12. The result of different input size: (a) the input image with 35×35 pixels; (b) the result of image with35×35 pixels; (c) the input image with 45×45 pixels; (b) the result of image with45×45 pixels.

    Download full size
    View in Article
    The ROC curves with different jitters.

    Fig. 13. The ROC curves with different jitters.

    Download full size
    View in Article
    The ROC curves with different mean original SCRs.

    Fig. 14. The ROC curves with different mean original SCRs.

    Download full size
    View in Article

    • Table 1. Computation comparison of different convolutions.

      View table
      View in Article

      Table 1. Computation comparison of different convolutions.

      ItemFlops（G）Parameters（K）
      3D Conv.0.0628.67
      Factorized 3D Conv.0.039.38

    • Table 2. The simulation environment.

      View table
      View in Article

      Table 2. The simulation environment.

      ItemParameter
      CPUIntel i7，2.8GHz×12
      GPUNvidia-1080Ti
      RAM64GB
      SystemUbuntu 18.04
      Disk2TB
      SoftwarePytorch 1.1
      LanguagePython 3.6

    • Table 3. Background suppression comparison by SCR in output.

      View table
      View in Article

      Table 3. Background suppression comparison by SCR in output.

      ProposedLin’sMax-MeanTopHatSTDA
      Target 120.406116.825320.930820.039419..8378
      Target 219.095311.92283.44103.062316.5872
      Mean19.750714.374112.185911.550918.2125

    • Table 4. Background suppression comparison by BSF.

      View table
      View in Article

      Table 4. Background suppression comparison by BSF.

      ProposedLin’sMax-MeanTopHatSTAD
      Target 11.35271.02931.24771.12101.3375
      Target 27.18154.32181.25201.04315.5936
      Mean4.26712.67551.24981.08213.4656

    • Table 5. Average runtime comparison.

      View table
      View in Article

      Table 5. Average runtime comparison.

      ProposedLin’sMax-MeanTopHatSTDA

      Average

      runtime（s）/sample

      		5.84×10-43.38×10-43.30×10-31.33×10-24.51×10-3
    Tools

    Get Citation

    Copy Citation Text

    Miao LI, Zai-Ping LIN, Jian-Peng FAN, Wei-Dong SHENG, Jun LI, Wei AN, Xin-Lei LI. Point target detection based on deep spatial-temporal convolution neural network[J]. Journal of Infrared and Millimeter Waves, 2021, 40(1): 122

    Download Citation

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

    Category: Research Articles

    Received: Jan. 9, 2020

    Accepted: --

    Published Online: Aug. 30, 2021

    The Author Email: Miao LI (lm8866@nudt.edu.cn)

    DOI:10.11972/j.issn.1001-9014.2021.01.017

    Topics

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