• Infrared and Laser Engineering
  • Vol. 49, Issue 8, 20200001 (2020)
Hanyu Hong1, Jun Zhao2, Yu Shi1、*, and Shikang Wu1
Author Affiliations
  • 1Hubei Key Laboratory of Optical Information and Pattern Recognition, Wuhan 430205, China
  • 2School of Electrical and Information Engineering, Wuhan Institute of Technology, Wuhan 430205, China
  • show less
    DOI: 10.3788/IRLA20200001 Cite this Article
    Hanyu Hong, Jun Zhao, Yu Shi, Shikang Wu. Blur kernel region estimation and space variant restoration based on weighted L1 norm measure[J]. Infrared and Laser Engineering, 2020, 49(8): 20200001 Copy Citation Text show less
    Comparison of gradient
    Fig. 1. Comparison of gradient
    8 neighborhood direction codes of point P
    Fig. 2. 8 neighborhood direction codes of point P
    Determined region for kernel estimation
    Fig. 3. Determined region for kernel estimation
    Schematic diagram of image expansion
    Fig. 4. Schematic diagram of image expansion
    Multiple region for kernel estimation
    Fig. 5. Multiple region for kernel estimation
    Flowchart of space-invariant image restoration
    Fig. 6. Flowchart of space-invariant image restoration
    Flowchart of space-variant image restoration
    Fig. 7. Flowchart of space-variant image restoration
    Comparison of restoration results of degraded images of ground targets
    Fig. 8. Comparison of restoration results of degraded images of ground targets
    Comparison of PSNR and SSIM
    Fig. 9. Comparison of PSNR and SSIM
    Comparison of restoration results of degraded images of real air and surface targets
    Fig. 10. Comparison of restoration results of degraded images of real air and surface targets
    Results of space-variant image restoration
    Fig. 11. Results of space-variant image restoration
    ImagesShanKrishnanProposed
    Church0.91730.91870.9664
    Airplane0.95110.94540.9615
    Harbor0.96820.94410.9729
    Power station0.90400.93610.9660
    Intersection0.83630.94070.9711
    Table 1.

    Comparison of the kernel similarity

    核相似性对比

    ImagesShanKrishnanProposed
    256×25613.284 s5.161 s3.350 s
    320×24016.362 s5.773 s5.621 s
    512×51219.726 s17.901 s9.694 s
    630×46026.213 s22.533 s9.885 s
    1 080×1 44083.221 s53.417 s28.925 s
    Table 2.

    Comparison of blur kernel estimation time

    模糊核估计耗时对比

    Hanyu Hong, Jun Zhao, Yu Shi, Shikang Wu. Blur kernel region estimation and space variant restoration based on weighted L1 norm measure[J]. Infrared and Laser Engineering, 2020, 49(8): 20200001
    Download Citation