Author Affiliations
1 Department of Fundamentals, Shaanxi Institute of International Trade & Commerce, Xianyang, Shaanxi 712046, China;2 School of Science, Xi'an University of Architecture and Technology, Xi'an, Shaanxi 710055, Chinashow less
Fig. 1. Results of scale-perception and edge-protection filter acting on one-dimensional signals. (a)(b) Partial structures of one-dimensional signal I; (c)(d) result after one filtering process, R1; (e)(f) result after two filtering processes, R2; (g)(h) result after three filtering processes, R3
Fig. 2. Network structure of residual learning
Fig. 3. Framework of denoising algorithm
Fig. 4. Denoising results for image Butterfly under different algorithms. (a) Image Butterfly; (b) noise image; (c) DWT algorithm; (d) CNC algorithm; (e) NLM algorithm; (f) BM3D algorithm; (g) EPLL algorithm; (h) proposed algorithm
Fig. 5. Denoising results for image Lena under different algorithms. (a) Image Lena; (b) noise image; (c) DWT algorithm; (d) CNC algorithm; (e) NLM algorithm; (f) BM3D algorithm; (g) EPLL algorithm; (h) proposed algorithm
Fig. 6. Denoising results for image Man under different algorithms. (a) Image Man; (b) noise image; (c) DWT algorithm; (d) CNC algorithm; (e) NLM algorithm; (f) BM3D algorithm; (g) EPLL algorithm; (h) proposed algorithm
Fig. 7. Denoising results for image Pepper under different algorithms. (a) Image Pepper; (b) noise image; (c) DWT algorithm; (d) CNC algorithm; (e) NLM algorithm; (f) BM3D algorithm; (g) EPLL algorithm; (h) proposed algorithm
Algorithm | Image |
---|
Butterfly | Lena | Man | Pepper |
---|
DWT | 16.8817 | 19.1079 | 18.1423 | 19.0933 | CNC | 22.9421 | 27.5587 | 24.3194 | 27.8372 | NLM | 23.6660 | 25.2807 | 23.9621 | 25.8773 | BM3D | 24.7856 | 28.7919 | 25.1842 | 29.1196 | EPLL | 25.1088 | 28.3879 | 25.3142 | 28.8953 | Proposed | 28.0179 | 30.8781 | 27.2887 | 31.2700 |
|
Table 1. Comparison of PSNR values of test images under different algorithmsdB
Algorithm | Image |
---|
Butterfly | Lena | Man | Pepper |
---|
DWT | 0.5945 | 0.7397 | 0.6245 | 0.7784 | CNC | 0.7506 | 0.8085 | 0.7445 | 0.8430 | NLM | 0.6881 | 0.7140 | 0.7435 | 0.7507 | BM3D | 0.8221 | 0.8261 | 0.7900 | 0.8463 | EPLL | 0.8248 | 0.8047 | 0.7915 | 0.8365 | Proposed | 0.8757 | 0.8783 | 0.8760 | 0.9002 |
|
Table 2. Comparison of SSIM values of test images under different algorithmsdB