Three-stage Binarization

Three-stage binarization of color document images based on discrete wavelet transform and generative adversarial networks

Stage-1 Flowchart

Stage-2 Flowchart

Stage-3 Flowchart

Citation

If you find our paper useful in your research, please consider citing:

Conference version (accepted by PRICAI 2023)

@inproceedings{ju2023ccdwt,
  title={CCDWT-GAN: Generative Adversarial Networks Based on Color Channel Using Discrete Wavelet Transform for Document Image Binarization},
  author={Ju, Rui-Yang and Lin, Yu-Shian and Chiang, Jen-Shiun and Chen, Chih-Chia and Chen, Wei-Han and Chien, Chun-Tse},
  booktitle={Pacific Rim International Conference on Artificial Intelligence},
  pages={186--198},
  year={2023},
  organization={Springer}
}

Journal version (accepted by KBS):

@article{ju2024three,
  title={Three-stage binarization of color document images based on discrete wavelet transform and generative adversarial networks},
  author={Ju, Rui-Yang and Lin, Yu-Shian and Jin, Yanlin and Chen, Chih-Chia and Chien, Chun-Tse and Chiang, Jen-Shiun},
  journal={Knowledge-Based Systems},
  pages={112542},
  year={2024},
  publisher={Elsevier}
}

Requirements

• Linux (Ubuntu)
• Python >= 3.6 (Pytorch)
• NVIDIA GPU + CUDA CuDNN

Installation

• Install segmentation_models

    pip install segmentation-models-pytorch

• Install pytesseract

    pip install pytesseract

• Download tesseract data

  For Conda users, you can create a new Conda environment using conda env create -f environment.yaml

Dataset

You can download the dataset used in this experiment from Dropbox.

Usage

• Preprocess

    python ./Base/image_to_224.py
    python ./Base/image_to_512.py
  

• Train the model

  • Stage2

    python train_stage2.py
  

  • Before train left part of Stage3

    python predict_for_stage3.py
  

  • left part of Stage3 (need train predict_for_stage3.py first)

    python train_stage3.py
  

  • right part of Stage3 (independent training)

    python train_stage3_resize.py
  

• Evaluation the model

    python3 eval_stage3_all.py
  

Related Works

Expand

• https://github.com/RuiyangJu/Efficient_Document_Image_Binarization