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

NUCLEAR TECHNIQUES, Volume. 47, Issue 4, 040201(2024)

CUDA-based parallel acceleration algorithm for wavelet denoising of airborne γ-ray spectrometry data

Chao XIONG1, Xin WANG1, Xinjie WANG2, and Hexi WU1、*
Author Affiliations
  • 1School of Nuclear Science and Engineering, East China University of Technology, Nanchang 330013, China
  • 2School of Radiation Medicine and Protection (SRMP) of Soochow University, Suzhou 215123, China
    • show less
    AbstractGet PDF(in Chinese)
    Figures&Tables (13)
    Equations (0)
    References (17)
    Tools
    Paper Information
    Topics
    Figures & Tables(13)
    Principle diagram of wavelet denoising(a) Decomposition process of wavelet transform, (b) Reconstruction process of wavelet transform

    Fig. 1. Principle diagram of wavelet denoising(a) Decomposition process of wavelet transform, (b) Reconstruction process of wavelet transform

    Download full size
    View in Article
    Wavelet threshold denoising process

    Fig. 2. Wavelet threshold denoising process

    Download full size
    View in Article
    Flowchart of threshold denoising algorithm

    Fig. 3. Flowchart of threshold denoising algorithm

    Download full size
    View in Article
    Flowchart of parallel threshold denoising process

    Fig. 4. Flowchart of parallel threshold denoising process

    Download full size
    View in Article
    Variation curves of logarithmic and total logarithmic calculation times of the improved threshold denoising function with block size

    Fig. 5. Variation curves of logarithmic and total logarithmic calculation times of the improved threshold denoising function with block size

    Download full size
    View in Article
    Changes of total time acceleration ratio and acceleration ratio of improved threshold denoising function with data volume

    Fig. 6. Changes of total time acceleration ratio and acceleration ratio of improved threshold denoising function with data volume

    Download full size
    View in Article
    Distribution rendering of artificial white noise

    Fig. 7. Distribution rendering of artificial white noise

    Download full size
    View in Article
    Denoising effects of three thresholding methods with the bior2.4 wavelet basis function(a) Original data, (b) Soft thresholding, (c) Hard thresholding, (d) Improved thresholding

    Fig. 8. Denoising effects of three thresholding methods with the bior2.4 wavelet basis function(a) Original data, (b) Soft thresholding, (c) Hard thresholding, (d) Improved thresholding

    Download full size
    View in Article
    Denoising effects of three thresholding methods with the bior3.1 wavelet basis function(a) Original data, (b) Soft thresholding, (c) Hard thresholding, (d) Improved thresholding

    Fig. 9. Denoising effects of three thresholding methods with the bior3.1 wavelet basis function(a) Original data, (b) Soft thresholding, (c) Hard thresholding, (d) Improved thresholding

    Download full size
    View in Article
    Different denoising effects with various wavelet basis functions (a) Original data, (b) Improved threshold denoising based on bior3.7, (c) Soft threshold denoising based on coif1, (d) Hard threshold denoising based on coif5

    Fig. 10. Different denoising effects with various wavelet basis functions (a) Original data, (b) Improved threshold denoising based on bior3.7, (c) Soft threshold denoising based on coif1, (d) Hard threshold denoising based on coif5

    Download full size
    View in Article

    • Table 1. Calculation time for improved threshold denoising function with different thread sizes for a data volume of 5122

      View table
      View in Article

      Table 1. Calculation time for improved threshold denoising function with different thread sizes for a data volume of 5122

      Block尺寸

      Block size

      改进阈值降噪时间

      Improved threshold denoising time / ms

      总计算时间

      Total calculation time / ms

      2220.221 530.01
      429.74446.08
      824.74181.89
      1625.87186.10
      3226.55131.66
      6423.0035.61
      12823.0334.48
      25623.1734.49
      51223.0134.53

    • Table 2. Acceleration times for different volumes of data on GPU and CPU

      View table
      View in Article

      Table 2. Acceleration times for different volumes of data on GPU and CPU

      数据体量

      Data volumes

      		GPU / msCPU / ms

      总加速时间比

      Total acceleration time ratio

      128×12835.353 243.0291.74
      256×25628.572 342.5181.99
      512×51242.356 382.14150.70
      1 024×1 02460.8613 125.69215.66
      2 048×2 048156.2347 887.57306.52
      4 096×4 096264.6387 289.75329.85

    • Table 3. GPU acceleration time ratios for improved threshold denoising function with different wavelet basis functions

      View table
      View in Article

      Table 3. GPU acceleration time ratios for improved threshold denoising function with different wavelet basis functions

      小波基函数

      Wavelet basis function

      		GPU / msCPU / ms

      总加速时间比

      Total acceleration time ratio

      haar351.2820 539.7058.47
      db1480.6656 125.62116.77
      db2377.6334 995.5792.67
      db3340.4135 170.16103.32
      db4326.0033 039.77101.35
      db5328.2943 708.06133.14
      db6383.2847 283.04123.36
      db7240.6024 907.15103.52
      db8223.8830 142.69134.64
      db9204.0534 520.88169.18
      db10266.7744 674.01167.46
      db11341.1677 962.34228.52
      db12340.6863 993.37187.84
      db13469.88103 128.32219.48
      db14405.7276 217.03187.86
      db15386.0370 866.26183.58
      bior1.194.984 633.6648.79
      bior1.3276.1127 007.2597.81
      bior1.5358.8147 749.07133.07
      boir2.2265.4525 751.5197.01
      bior2.4355.0047 760.17134.54
      bior2.6375.7555 759.64148.40
      bior2.8396.6985 373.38215.21
      bior3.1298.4122 660.6875.94
      bior3.3283.2324 242.8685.59
      bior3.5289.4132 044.64110.72
      bior3.7252.5234 123.10135.13
      bior3.9328.6257 500.67174.98
      bior4.5230.3246 399.65201.46
      bior5.5446.3865 038.18145.70
      bior6.8340.2473 286.82215.40
      coif1417.0239 647.6395.07
      coif2341.8060 120.33175.89
      coif3342.2564 894.79189.61
      coif4440.14122 101.62277.41
      coif5567.05200 433.42353.46
      sym2293.9027 457.1893.42
      sym3259.4028 615.62110.31
      sym4430.4963 204.12146.82
      sym5438.0469 121.81157.80
      sym6416.8470 246.89168.52
      sym7272.3752 153.92191.48
      sym8421.5090 151.20213.88
      sym9336.8167 660.68200.89
      sym10337.0659 812.25177.45
    Tools

    Get Citation

    Copy Citation Text

    Chao XIONG, Xin WANG, Xinjie WANG, Hexi WU. CUDA-based parallel acceleration algorithm for wavelet denoising of airborne γ-ray spectrometry data[J]. NUCLEAR TECHNIQUES, 2024, 47(4): 040201

    Download Citation

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

    Category: Research Articles

    Received: Oct. 19, 2023

    Accepted: --

    Published Online: May. 28, 2024

    The Author Email: WU Hexi (吴和喜)

    DOI:10.11889/j.0253-3219.2024.hjs.47.040201

    Topics

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