DnCNN-tensorflow

A tensorflow implement of the TIP2017 paper Beyond a Gaussian Denoiser: Residual Learning of Deep CNN for Image Denoising

This is a fork of https://github.com/crisb-DUT/DnCNN-tensorflow

The main differences from the original version are:

• It supports RGB images.
• PIL was replaced with OpenCV.
• You don't need to generate patches in advance, which require a lot of disk and memory space.
• The noise is not added within the programme but you need an original and a noisy image, so you can train the model even with different types of noise.
• The provided checkpoint/model was trained for blind gaussian denoising with sigma values 0-50.

Model Architecture

Results

• BSD68 Average Result

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

Noise Level	BM3D	WNNM	EPLL	MLP	CSF	TNRD	DnCNN-S	DnCNN-B	DnCNN-tensorflow
25	28.57	28.83	28.68	28.96	28.74	28.92	29.23	29.16	29.17

• Set12 Average Result

Noise Level	DnCNN-S	DnCNN-tensorflow
25	30.44	30.38

Requirements

tensorflow >= 1.4
numpy
opencv

Dataset

I used the BDS500 dataset for training, you can download it here: http://www.eecs.berkeley.edu/Research/Projects/CS/vision/grouping/BSR/BSR_bsds500.tgz It contains 500 RGB images, 400 for training and 100 for testing.

Data preprocessing and noise generation

Before training, you have to rescale the images to 180x180 and adding noise to them. The folder structure is supposed to be:

./data/train/original  for the 180x180 original train images
./data/train/noisy  for the 180x180 noisy train images
./data/test/original  for the 180x180 original test images
./data/test/noisy  for the 180x180 noisy test images

You need the original files for testing just to calculate the PSNR. You can denoise without original files: just put the noisy files also in ./data/test/original .

Train

$ python main.py
(note: You can add command line arguments according to the source code, for example
    $ python main.py --batch_size 64 )

Test

$ python main.py --phase test