Multi-domain fusion network for underwater image enhancement
Junbin Zhuang, Jiajia Zhou, Yan Zheng, Yasheng Chang, Suleman Mazhar
Abstract
A novel network for underwater optical images enhancement is proposed in this paper. It contains a Reinforcement Fusion Module for Haar wavelet (RFM-Haar) images based on Reinforcement Fusion Unit (RFU), which is used to fuse an spatial image and some frequency information. Fusion is achieved to enrich available underlying information for better enhancement. As this network make Haar Images into Fusion Images, it is called HIFI-Net. The experimental results at Mean Square Error (0.1952, 0.1512 and 0.4683), Peak Signal to Noise Ratio (25.2241, 26.3342 and 21.4253) and Structure Similarity Index Measure (0.8265, 0.8819 and 0.8012) on three public datasets (EUVP, UFO-110 and UIEB) show the proposed HIFI-Net performs best among some state-of-the-art methods.
BibTeX
@inproceedings{icassp2025_multidomainfusio,
title = {Multi-domain fusion network for underwater image enhancement},
author = {Junbin Zhuang and Jiajia Zhou and Yan Zheng and Yasheng Chang and Suleman Mazhar},
booktitle = {ICASSP 2025},
year = {2025}
}