The codes and trained model for paper "Color Image Restoration Exploiting Inter-channel Correlation with a 3-stage CNN"

K. Cui, A. Boev, E. Alshina and E. Steinbach, Color Image Restoration Exploiting Inter-channel Correlation with a 3-stage CNN, IEEE-JSTSP, 2020. DOI: 10.1109/JSTSP.2020.3043148

---

Dependencies:

• Python 3
• TensorFlow 1.XX (1.10 or newer)
• NumPy
• Pillow
• NVIDIA GPU + CUDA (if running in GPU mode)

Dataset:

• You need to download the testing datasets to run the demo test for different tasks. We summarize the datasets here. Unzip the datasets and put them into the data folder. If you have your own dataset, please follow the readme in the data folder to organize the dataset.

Usage:

1. There are three subtasks in our paper, color demosaicking (CDM), compression artifacts reduction (CAR), and real-world color image denoising (RIDN). The code and the trained models are in the corresponding folder.

2. For CDM, run python main_py3_tfrecord.py to test the Kodak dataset.
   When testing other datasets, simply add --test_set NAME, e.g., python main_py3_tfrecord.py --test_set McM
   It also supports the ensemble testing mode, run python main_py3_tfrecord.py --phase ensemble

3. For CAR, run python main_py3_tfrecord.py to test the LIVE1 dataset with qp = 10.
   Use --qp XX to test different QPs, e.g., python main_py3_tfrecord.py --qp 100

4. For RIDN, run python main_py3_tfrecord.py to test the SIDD_validation dataset.

5. The codes support both CPU and GPU mode. Default is GPU 0, use --gpu -1 to run on CPU or choose other GPUs.

6. Please read our paper for more details!

7. Have fun!

---

Citation:

Please cite our paper if you find the paper or the code is helpful for your research.

@ARTICLE{CNNCIR-JSTSP-2020,  
  author={K. {Cui} and A. {Boev} and E. {Alshina} and E. {Steinbach}},  
  journal={IEEE Journal of Selected Topics in Signal Processing},  
  title={Color Image Restoration Exploiting Inter-channel Correlation with a 3-stage {CNN}},   
  year={2020},  
  volume={},  
  number={},  
  pages={1-1},  
  doi={10.1109/JSTSP.2020.3043148}}

Related Work

1. K. Cui, Z. Jin and E. Steinbach, Color image demosaicking using a 3-stage convolutional neural network structure, ICIP 2018. [Paper] [Code]
2. K. Cui and E. Steinbach, Decoder Side Image Quality Enhancement exploiting Inter-channel Correlation in a 3-stage CNN: Submission to CLIC 2018, CVPR Workshops 2018. [Paper]
3. K. Cui and E. Steinbach, Decoder Side Color Image Quality Enhancement using a Wavelet Transform based 3-stage Convolutional Neural Network, CVPR Workshops 2019. [Paper]

---

Maintainer:

@Kai Cui ([email protected])
Lehrstuhl fuer Medientechnik (LMT)
Technische Universitaet Muenchen (TUM)
Last modified 06.02.2021

---

License

This project is released under the Apache 2.0 license.