Quality Assessment for Tone-Mapped HDR Images Using Multi-Scale and Multi-Layer Information

Description

Code for the following paper:

• Qin He, Dingquan, Tingting Jiang, and Ming Jiang. "Quality Assessment for Tone-Mapped HDR Images Using Multi-Scale and Multi-Layer Information." ICMEw, 2018.

Requirement

Framework: Caffe 1.0 (with CUDA 8.0) + MATLAB 2016b Interface

Download the ResNet-50-model.caffemodel from https://github.com/KaimingHe/deep-residual-networks and paste it into the directory "models/" before using the code! It's about 100MB which is too large to upload to this repo. If you have difficulty, you can also download the ResNet-50-model.caffemodel in my sharing on BaiduNetDisk with password u8sd.

Feature Extraction

The features are extracted from the DCNN models pre-trained on the image classification task.

Remember to change the value of "im_dir" and "im_lists" in data info！

Run ExtractFeatures.m to get the features. For features of images from the ESPL-LIVE HDR dataset, you can also download from my sharing on BaiduNetDisk with password 3aj0.

Quality Prediction by PLSR

All we need to train is a PLSR model, where the training function is plsregress.m in MATLAB.

Run QualityPrediction.m to conduct the experiments on ESPL-LIVE HDR.

Citation

Please cite our paper if it helps your research:

@inproceedings{he2018quality,
  title={Quality Assessment for Tone-Mapped HDR Images Using Multi-Scale and Multi-Layer Information},
  author={He, Qin and Li, Dingquan and Jiang, Tingting and Jiang, Ming},
  booktitle={ICMEw},
  year={2018}
}

Contact

Dingquan Li, [email protected].