Laser & Optoelectronics Progress, Volume. 57, Issue 12, 121005(2020)
Analysis of an Algorithm for Repairing Image Defects
Figures & Tables(16)
Fig. 1. Image interior block map. (a) Gradient amplitude image; (b) segmentation image
Fig. 2. Sub-items relation graph of the two algorithms. (a) Criminisi algorithm; (b) proposed algorithm
Fig. 3. Repair at the 350th time of Criminisi algorithm. (a) Repair piece; (b) result
Fig. 4. Comparison of image restoration effect for blocks with different sizes. (a) 9 pixel×9 pixel; (b) 7 pixel×7 pixel; (c) 5 pixel×5 pixel; (d) 3 pixel×3 pixel
Fig. 5. Comparison of repair effect for image with artificial tiny acratches. (a) Original image; (b) image of artificial tiny scratches; (c) Ref. [2]; (d) Ref. [3]; (e) proposed algorithm
Fig. 6. Comparison of repair effect for image with artificial large area scratches. (a) Original image; (b) image of artificial big scratch; (c) Ref. [2]; (d) Ref.[3]; (e) proposed algorithm
Fig. 7. Comparison of image repair effect of eight algorithms. (a) Original image; (b) Criminisi algorithm; (c) Ref. [8]; (d) Ref. [9]; (e) Ref. [12]; (f) Ref. [13]; (g) Ref. [14]; (h) Ref. [15]; (i) proposed algorithm
Fig. 8. Comparison of restoration effect for image with tiny scratches. (a) Original image; (b) Criminisi algorithm; (c) Ref. [3];(d) Ref. [12]; (e) Ref. [13]; (f) Ref. [14]; (g) Ref. [15]; (h) proposed algorithm
Fig. 9. Comparison of restoration effect for image with large area defect. (a) Original image; (b) Criminisi algorithm; (c) Ref. [3]; (d) Ref. [12]; (e) Ref. [13]; (f) Ref. [14]; (g) Ref. [15]; (h) proposed algorithm
Fig. 10. Comparison of restoration effect of moving scene in image. (a) Original image; (b) Criminisi algorithm; (c) Ref. [3]; (d) Ref. [12]; (e) Ref. [13]; (f) Ref. [14]; (g) Ref. [15]; (h) proposed algorithm
Fig. 11. Comparison of restoration effect for damaged inscription. (a) Original image; (b) Criminisi algorithm; (c) Ref. [12];(d) Ref. [13]; (e) Ref. [14]; (f) Ref. [15]; (g) proposed algorithm
Fig. 12. Comparison of repair effects of different algorithms for removing aircraft from image. (a) Original image1; (b) original image2; (c) composite image; (d) Criminisi algorithm; (e) Ref. [12]; (f) Ref. [13]; (g) Ref. [14]; (h) Ref. [15]; (i) proposed algorithm
Table 1. Comparison of image repair effects of three algorithms (tiny scratch)
View tableTable 1. Comparison of image repair effects of three algorithms (tiny scratch)
Algorithm SSIM /% PSNR /dB Ref. [2] 0.59 33.83 Ref. [3] 0.69 48.75 Proposed 0.80 84.05
Table 2. Comparison of image repair effects of three algorithms (big scratch)
View tableTable 2. Comparison of image repair effects of three algorithms (big scratch)
Algorithm SSIM /% PSNR /dB Ref. [2] 0.71 36.34 Ref. [3] 0.78 39.10 Proposed 0.80 52.62
Table 3. Image repair time of 7 algorithmsunit: s
View tableTable 3. Image repair time of 7 algorithmsunit: s
Algorithm Fig. 8 Fig. 9 Fig. 10 Fig. 11 Criminisi 29.48 178.50 44.40 2.36 Ref. [3] 434.96 680.79 252.60 40.51 Ref. [12] 34.09 198.78 48.91 2.57 Ref. [13] 33.62 188.73 47.35 2.59 Ref. [14] 33.79 193.86 47.83 2.59 Ref. [15] 28.52 162.13 37.90 2.23 Proposed 9.64 97.89 7.14 1.33
Table 4. Analysis of repair effects of six algorithms
View tableTable 4. Analysis of repair effects of six algorithms
Algorithm Repair time /s PSNR /dB SSIM /% Criminisi 47.85 50.22 0.59 Ref. [12] 52.37 50.18 0.59 Ref. [13] 52.12 49.88 0.59 Ref. [14] 52.74 50.12 0.59 Ref. [15] 40.78 48.84 0.49 Proposed 6.91 51.39 0.59
Tools
Get Citation
Copy Citation Text
Yongfei Wang. Analysis of an Algorithm for Repairing Image Defects[J]. Laser & Optoelectronics Progress, 2020, 57(12): 121005
Paper Information
Category: Image Processing
Received: Sep. 23, 2019
Accepted: Oct. 29, 2019
Published Online: Jun. 3, 2020
The Author Email: Wang Yongfei (121086842@qq.com)
© Copyright 2018-2021 | Chinese Laser Press.
All Rights Reserved 沪ICP备15018463号-20