Author Affiliations
1School of Information Engineering, Southwest University of Science and Technology, Mianyang 621010, Sichuan , China2Robot Technology Used for Special Environment Key Laboratory of Sichuan Province, Mianyang 621010, Sichuan , Chinashow less
Fig. 1. Noise example in γ nuclear radiation scene. (a) Noise map; (b) bright patch noise; (c) dark patch noise
Fig. 2. Variation of image quality in γ radiation scene with increase of irradiation time. (a) Variation of PSNR; (b) variation of SSIM
Fig. 3. Detection results of bright patch noise in different γ radiation scene images. (a) Noise map in scene 1; (b) detection results of bright patch noise in [Fig.3(a)]; (c) binarization of [Fig.3(b)]; (d) noise map in scene 2; (e) detection results of bright patch noise in [Fig.3(d)]; (f) binarization of [Fig.3(e)]
Fig. 4. Detection results of dark patch noise in different γ radiation scene images. (a) Scene 1 pixel flip; (b) detection results of dark patch noise in [Fig.4(a)]; (c) binarization of [Fig.4(b)]; (d) scene 2 pixel flip; (e) detection results of dark patch noise in [Fig.4(d)]; (f) binarization of [Fig.4(e)]
Fig. 5. Time series correlation method for transient noise removal
Fig. 6. Denoising effect of different number of near frame images with different irradiation time. (a) PNSR value after denoising; (b) SSIM value after denoising
Fig. 7. Results of denoising and enhancement in different γ radiation scene images. (a) Noise map in scene 1; (b) denoising results of [Fig.7(a)]; (c) enhancement results of [Fig.7(b)]; (d) noise map in scene 2; (e) denoising result of [Fig.7(d)]; (f) enhancement result of [Fig.7(e)]
Fig. 8. Quality of noiseless image changes with average number of frames increase. (a) Variation of PSNR value; (b) variation of SSIM value
Fig. 9. Comparison of denoising results in 200 Gy/h γ radiation scene images. (a) Noise map; (b) median; (c) wavelet; (d) anisotropy;(e) PDE; (f) BM3D; (g) NLM; (h) guide; (i) TV; (j) proposed algorithm
Fig. 10. Comparison of denoising results in 20 Gy/h γ radiation scene images. (a) Noise map; (b) median; (c) wavelet; (d) anisotropy;(e) PDE; (f) BM3D; (g) NLM; (h) guide; (i) TV; (j) proposed algorithm
Lt | r=0.25 | r=0.5 | r=0.75 |
---|
n=1 | n=2 | n=3 | n=1 | n=2 | n=3 | n=1 | n=2 | n=3 |
---|
5 | 32.56 | 32.78 | 32.81 | 31.5 | 31.96 | 32.23 | 24.48 | 29.5 | 30.23 | 10 | 31.59 | 31.68 | 32.2 | 26.52 | 31.2 | 31.71 | 23.29 | 26.22 | 29.93 | 15 | 30.43 | 30.73 | 31.48 | 24.55 | 29.43 | 30.03 | 22.48 | 23.45 | 28.48 |
|
Table 1. PSNR of denoising results in γ radiation scene images with change
Dose /(Gy·h-1) | Index | Noise | Median | Wavelet | Anisotropy | PDE | BM3D | NLM | Guide | TV | Proposed algorithm |
---|
200 | PSNR /dB | 16.72 | 20.89 | 19.68 | 16.80 | 21.78 | 16.95 | 16.93 | 21.95 | 18.86 | 32.66 | SSIM | 0.28 | 0.47 | 0.36 | 0.32 | 0.44 | 0.31 | 0.31 | 0.47 | 0.36 | 0.88 | 20 | PSNR /dB | 23.48 | 24.22 | 23.85 | 23.22 | 26.25 | 23.52 | 23.65 | 26.20 | 25.50 | 33.43 | SSIM | 0.65 | 0.67 | 0.65 | 0.71 | 0.79 | 0.65 | 0.68 | 0.79 | 0.76 | 0.94 |
|
Table 2. Quantitative comparison of denoising results
Image size | Index | Noise | Median | Wavelet | Anisotropy | PDE | BM3D | NLM | Guide | TV | Proposed algorithm |
---|
500×500 | PSNR /dB | 16.90 | 20.90 | 20.05 | 16.80 | 19.16 | 16.95 | 16.90 | 21.96 | 24.16 | 32.73 | SSIM | 0.30 | 0.47 | 0.41 | 0.32 | 0.41 | 0.31 | 0.31 | 0.50 | 0.67 | 0.87 | 600×600 | PSNR /dB | 16.87 | 20.73 | 19.95 | 16.83 | 19.15 | 16.92 | 16.88 | 21.86 | 23.54 | 32.68 | SSIM | 0.31 | 0.47 | 0.40 | 0.32 | 0.41 | 0.31 | 0.31 | 0.50 | 0.64 | 0.88 | 700×700 | PSNR /dB | 16.83 | 20.64 | 19.74 | 16.84 | 19.02 | 16.88 | 16.84 | 21.67 | 23.00 | 32.68 | SSIM | 0.31 | 0.47 | 0.39 | 0.33 | 0.41 | 0.31 | 0.31 | 0.50 | 0.64 | 0.88 | 800×800 | PSNR /dB | 16.91 | 20.77 | 19.81 | 16.94 | 19.13 | 16.96 | 16.91 | 21.77 | 22.75 | 1 | SSIM | 0.31 | 0.48 | 0.40 | 0.34 | 0.42 | 0.32 | 0.32 | 0.50 | 0.64 | 0.88 | 900×900 | PSNR /dB | 16.98 | 20.89 | 19.91 | 17.06 | 19.21 | 17.03 | 16.99 | 21.85 | 22.71 | 32.62 | SSIM | 0.32 | 0.48 | 0.40 | 0.34 | 0.42 | 0.323 | 0.33 | 0.50 | 0.65 | 0.87 | 1000×1000 | PSNR /dB | 16.94 | 20.87 | 19.90 | 17.06 | 19.17 | 16.99 | 16.95 | 21.85 | 22.86 | 32.65 | SSIM | 0.31 | 0.47 | 0.39 | 0.33 | 0.41 | 0.31 | 0.32 | 0.50 | 0.66 | 0.87 |
|
Table 3. Quantitative comparison of denoising results in different size images