文件列表:

LICENSE (1063, 2018-03-21)
data.py (2333, 2018-03-21)
data (0, 2018-03-21)
data\Test (0, 2018-03-21)
data\Test\Set12 (0, 2018-03-21)
data\Test\Set12\01.png (38267, 2018-03-21)
data\Test\Set12\02.png (34985, 2018-03-21)
data\Test\Set12\03.png (40181, 2018-03-21)
data\Test\Set12\04.png (42947, 2018-03-21)
data\Test\Set12\05.png (40728, 2018-03-21)
data\Test\Set12\06.png (40985, 2018-03-21)
data\Test\Set12\07.png (39804, 2018-03-21)
data\Test\Set12\08.png (151065, 2018-03-21)
data\Test\Set12\09.png (185727, 2018-03-21)
data\Test\Set12\10.png (177762, 2018-03-21)
data\Test\Set12\11.png (209817, 2018-03-21)
data\Test\Set12\12.png (193637, 2018-03-21)
data\Test\Set12\Thumbs.db (238080, 2018-03-21)
data\Test\Set68 (0, 2018-03-21)
data\Test\Set68\Thumbs.db (246784, 2018-03-21)
data\Test\Set68\test001.png (118864, 2018-03-21)
data\Test\Set68\test002.png (74483, 2018-03-21)
data\Test\Set68\test003.png (86227, 2018-03-21)
data\Test\Set68\test004.png (84485, 2018-03-21)
data\Test\Set68\test005.png (97073, 2018-03-21)
data\Test\Set68\test006.png (71584, 2018-03-21)
data\Test\Set68\test007.png (101018, 2018-03-21)
data\Test\Set68\test008.png (115601, 2018-03-21)
data\Test\Set68\test009.png (93022, 2018-03-21)
data\Test\Set68\test010.png (89721, 2018-03-21)
data\Test\Set68\test011.png (92534, 2018-03-21)
data\Test\Set68\test012.png (103143, 2018-03-21)
data\Test\Set68\test013.png (92986, 2018-03-21)
data\Test\Set68\test014.png (81907, 2018-03-21)
data\Test\Set68\test015.png (105365, 2018-03-21)
data\Test\Set68\test016.png (110448, 2018-03-21)
data\Test\Set68\test017.png (68933, 2018-03-21)
... ...

# DnCNN-keras A keras implemention of the paper [Beyond a Gaussian Denoiser: Residual Learning of Deep CNN for Image Denoising](http://www4.comp.polyu.edu.hk/~cslzhang/paper/DnCNN.pdf) ### Dependence ``` tensorflow keras2 numpy opencv ``` ### Prepare train data ``` $ python data.py ``` Clean patches are extracted from 'data/Train400' and saved in 'data/npy_data'. ### Train ``` $ python main.py ``` Trained models are saved in 'snapshot'. ### Test ``` $ python main.py --only_test True --pretrain 'path of saved model' ``` Noisy and denoised images are saved in 'snapshot'. ### Results #### Gaussian Denoising The average PSNR(dB) results of different methods on the BSD68 dataset. | Noise Level | BM3D | DnCNN-S | DnCNN-keras | |:-------:|:-------:|:-------:|:-------:| | 25 | 28.57 | 29.23 | 29.21 |

近期下载者：

相关文件：

评论：[我要评论] [举报此文件]

收藏者：