A keras implemention of the paper Beyond a Gaussian Denoiser: Residual Learning of Deep CNN for Image Denoising

tensorflow
keras2
numpy
opencv

$ python data.py

Clean patches are extracted from 'data/Train400' and saved in 'data/npy_data'.

$ python main.py

Trained models are saved in 'snapshot'.

$ python main.py --only_test True --pretrain 'path of saved model'

Noisy and denoised images are saved in 'snapshot'.

The average PSNR(dB) results of different methods on the BSD68 dataset.