Deep Inverse Halftoning

We run this code under TensorFlow 1.6 on Ubuntu16.04 with python pakage IPL installed.

Network Architecture

TensorFlow Implementation of our paper "Deep Inverse Halftoning via Progressively Residual Learning" accepted to ACCV 2018.

Results

Preparation

• You can run exisitng halftone algorithm (Foyd-Steinberg Error diffusion on 8-bit grayscale image is used in our pretrained model) to generate halftone version of your continuous-tone grayscale or color images, working as training pairs.
• The patch size is set to 256x256 in the model.py (you may change it to any other size as you like).
• Download the pretrained VGG19 model in here.

Run

• Set your image folders and hyperparameters in main.py.

• Start training.

line238: parser.add_argument('--mode', type=str, default='train', help='train, test')

python3 main.py

• Start evaluation. (acess pretrained model)

line 238: parser.add_argument('--mode', type=str, default='test', help='train, test')

python3 main.py 

Copyright and License

You are granted with the license for both academic and commercial usages.

Citation

If any part of our paper and code is helpful to your work, please generously cite with:

@inproceedings{XiaW18,
  author    = {Menghan Xia and Tien-Tsin Wong},
  title     = {Deep Inverse Halftoning via Progressively Residual Learning},
  booktitle = {Asian Conference on Computer Vision (ACCV)},
  year      = {2018}
}