Acta Optica Sinica, Volume. 41, Issue 23, 2310001(2021)

Blind Separation Algorithm of X-Ray Polychromatic Projections Based on Material Composition Prior

Yihong Li1, Zhaoyan Qu2, Xiaojie Zhao1, Jiaotong Wei1, and Ping Chen1、*
Author Affiliations
  • 1Shanxi Key Laboratory of Signal Capturing & Processing, North University of China, Taiyuan, Shanxi 0 30051, China;
  • 2Department of Physics and Electronic Engineering, Yuncheng University, Yuncheng, Shanxi 0 44000, China
  • show less
    Figures & Tables(11)
    Simulation setting. (a) Simulation body; (b) simulation energy spectrum generated by Spekpy v2.0; (c) attenuation coefficient of material (from NIST)
    Direct reconstruction results of multi-energy spectrum projection images in 17 intervals. (a) Reconstructed image; (b) attenuation coefficient curve
    Partially reconstructed images obtained by decomposing 5th, 7th, 9th, and 11th projection images by NMF-GN algorithm and BSS-BE algorithm
    Attenuation coefficient curve and normalized root mean square error of reconstructed image. (a) Attenuation coefficient of 5th reconstructed image processed by NMF-GN algorithm; (b) attenuation coefficient of 7th reconstructed image processed by NMF-GN algorithm; (c) change curve of normalized root mean square error with sequence number of reconstructed image
    Direct reconstruction results of multi-energy spectrum projection images in 18 intervals. (a) Reconstructed image images; (b) attenuation coefficient curve
    Partially reconstructed images obtained by decomposing 7th, 8th, 9th and 16th projection images by NMF-GN algorithm and BSS-BE algorithm
    Attenuation coefficient curves decomposed by NMF-GN algorithm and BSS-LV-BE algorithm. (a) Attenuation coefficient of 8th reconstructed image at line in Fig. 5(a); (b) attenuation coefficient of 16th reconstructed image at line in Fig. 5(a)
    Change curves of average attenuation coefficient of different materials with sequence number of reconstructed image. (a) Mg; (b) Al
    Reconstructed images and attenuation coefficient curves of sample 2. (a) Directly reconstructed image under 60 kV voltage; (b) 7th reconstructed image obtained by NMF-GN algorithm (corresponding energy is 38 keV); (c) 7th reconstructed image obtained by BSS-LV-BE algorithm; (d)(e)(f) attenuation coefficient of reconstructed image (a)(b)(c) at line of Fig.(a)
    • Table 1. Time cost of three experimentsunit: s

      View table

      Table 1. Time cost of three experimentsunit: s

      AlgorithmSimulationSample 1Sample 2
      NMF-GN641725203940
      BSS-LV-BE553113385156
    • Table 2. Gray mean, standard deviation, and coefficient of variation of each material in partially reconstructed images of sample 1

      View table

      Table 2. Gray mean, standard deviation, and coefficient of variation of each material in partially reconstructed images of sample 1

      ParameterDirect reconstructed image (power is 90 kV)8th CT image16th CT image
      MgAlMgAlMgAl
      Mean /mm-10.110000.173000.095600.161000.033800.05490
      STD /mm-10.012900.014000.006910.006820.002620.00291
      CV0.118000.081100.071900.042400.078500.05500
    Tools

    Get Citation

    Copy Citation Text

    Yihong Li, Zhaoyan Qu, Xiaojie Zhao, Jiaotong Wei, Ping Chen. Blind Separation Algorithm of X-Ray Polychromatic Projections Based on Material Composition Prior[J]. Acta Optica Sinica, 2021, 41(23): 2310001

    Download Citation

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

    Category: Image Processing

    Received: Jun. 21, 2021

    Accepted: Aug. 20, 2021

    Published Online: Nov. 29, 2021

    The Author Email: Chen Ping (pc0912@163.com)

    DOI:10.3788/AOS202141.2310001

    Topics