/RNTIEQA-dataset

We constructed a new RNTIEQA dataset for night-time image enhancement, and carried out human subjective studies to compare the quality of ENTIs obtained by a set of representative night-time image enhancement algorithms.

Benchmark Dataset and Pair-WiseRanking Method for Quality Evaluation of Enhanced Night-Time Images

RNTIEQA-dataset

Table of content: 📎Paper Link 💡Abstract 📃Requirement 📖Usage 🍎Noting ✨Statement 🔍Citation

📎Paper Link:https://ieeexplore.ieee.org/document/10506553

💡Abstract: The previously published works still suffer from the following limitations: 1) The subjective studies only focused on the ability of night-time image enhancement (NTIE) algorithms in boosting the intensity of low-light regions, which neglected analyzing the impact of night-time scenes under different lighting conditions (e.g., extreme-low light and uneven light) on the quality of enhanced night-time images (ENTIs) generated by different NTIE algorithms; 2) Only traditional image enhancement algorithms were used to generate ENTIs without considering the emerging deep learning-based image enhancement algorithms, which may make the dataset cannot fully reflect the quality degradation factors; 3) The performance of the existing IQA metrics for ENTIs were not robust enough for real applications, which called for effective and reliable objective quality evaluation metrics to automatically assess the quality of ENTIs. We constructed a new RNTIEQA dataset for night-time image enhancement, and carried out human subjective studies to compare the quality of ENTIs obtained by a set of representative night-time image enhancement algorithms.We proposed a new objective ranking method that comprehensively considering image intrinsic and impairment attributes for automatically predicting the quality of enhanced night-time images.

📃Requirement:

📖Usage:

You can download our RNTIEQA-dataset and code from BaiduYun Disk：https://pan.baidu.com/s/10XZXEm3MS2PNy30Z0vrDOw?pwd=uzst password：uzst or Goolgle Drive: https://drive.google.com/file/d/14_p9YFQu9iFk5M72izADakHUKjOP2vKR/view?usp=sharing

🍎Noting:

✨Statement: This project is for research purpose only, please contact us for the licence of commercial use. For any other questions please contact 3512068622@qq.com,wxj_2024@126.com.

🔍Citation: If you find this work useful for you. Please cite:X. Wang, L. Huang, H. Chen, Q. Jiang, S. Weng and F. Shao, "Benchmark Dataset and Pair-Wise Ranking Method for Quality Evaluation of Night-Time Image Enhancement," in IEEE Transactions on Multimedia, doi: 10.1109/TMM.2024.3391907. keywords: {Measurement;Image enhancement;Lighting;Feature extraction;Image quality;Distortion;Benchmark testing;Enhanced night-time image;Image quality evaluation;Deep learning;Subjective assessment;Pair- wise ranking},