Underwater Image Enhancement via Adaptive Group Attention-Based Multiscale Cascade Transformer

被引:90
作者
Huang, Zhixiong [1 ]
Li, Jinjiang [2 ]
Hua, Zhen [1 ]
Fan, Linwei [3 ]
机构
[1] Shandong Technol & Business Univ, Sch Informat & Elect Engn, Yantai 264005, Shandong, Peoples R China
[2] Shandong Technol & Business Univ, Sch Comp Sci & Technol, Yantai 264005, Shandong, Peoples R China
[3] Shandong Univ Finance & Econ, Sch Comp Sci & Technol, Jinan 250014, Shandong, Peoples R China
基金
中国国家自然科学基金;
关键词
Image color analysis; Image restoration; Imaging; Attenuation; Transformers; Image enhancement; Estimation; Adaptive group attention (AGA); multiscale cascade; Swin Transformer; underwater image enhancement; COLOR CORRECTION; RESTORATION; MAP;
D O I
10.1109/TIM.2022.3189630
中图分类号
TM [电工技术]; TN [电子技术、通信技术];
学科分类号
0808 ; 0809 ;
摘要
The absorption and scattering caused by the underwater medium degrade the quality of underwater optical imaging, which limits the further development of underwater tasks. Recently, transformer-based methods have shown the same excellent performance as convolutional neural networks (CNNs) in various vision tasks, but the huge parameters of such networks hinder their application deployment. In this article, we propose novel adaptive group attention (AGA), which can dynamically select visually complementary channels based on the dependencies, reducing the number of further attention parameters. The AGA is applied in the Swin Transformer module and used to design an end-to-end underwater image enhancement network. The network also introduces the multiscale cascade module and the channel attention mechanism. This article conducted ablation study and qualitative and quantitative comparisons on public datasets, and the results show that the application of AGA significantly compresses the model size while ensuring performance, and other application components have the significant gain on the network. Compared with other advanced methods, the network in this article has outstanding performance.
引用
收藏
页数:18
相关论文
共 68 条
[1]   Color Balance and Fusion for Underwater Image Enhancement [J].
Ancuti, Codruta O. ;
Ancuti, Cosmin ;
De Vleeschouwer, Christophe ;
Bekaert, Philippe .
IEEE TRANSACTIONS ON IMAGE PROCESSING, 2018, 27 (01) :379-393
[2]   Natural-based underwater image color enhancement through fusion of swarm-intelligence algorithm [J].
Azmi, Kamil Zakwan Mohd ;
Ghani, Ahmad Shahrizan Abdul ;
Yusof, Zulkifli Md ;
Ibrahim, Zuwairie .
APPLIED SOFT COMPUTING, 2019, 85
[3]   Underwater Single Image Color Restoration Using Haze-Lines and a New Quantitative Dataset [J].
Berman, Dana ;
Levy, Deborah ;
Avidan, Shai ;
Treibitz, Tali .
IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE, 2021, 43 (08) :2822-2837
[4]  
Bychkovsky V, 2011, PROC CVPR IEEE, P97
[5]   End-to-End Object Detection with Transformers [J].
Carion, Nicolas ;
Massa, Francisco ;
Synnaeve, Gabriel ;
Usunier, Nicolas ;
Kirillov, Alexander ;
Zagoruyko, Sergey .
COMPUTER VISION - ECCV 2020, PT I, 2020, 12346 :213-229
[6]   Underwater Image Enhancement by Wavelength Compensation and Dehazing [J].
Chiang, John Y. ;
Chen, Ying-Ching .
IEEE TRANSACTIONS ON IMAGE PROCESSING, 2012, 21 (04) :1756-1769
[7]  
Child R, 2019, Arxiv, DOI arXiv:1904.10509
[8]   Color Channel Compensation (3C): A Fundamental Pre-Processing Step for Image Enhancement [J].
Codruta, Ancuti O. ;
Ancuti, Cosmin ;
De Vleeschouwer, Christophe ;
Sbert, Mateu .
IEEE TRANSACTIONS ON IMAGE PROCESSING, 2020, 29 :2653-2665
[9]  
Dehghani M, 2019, Arxiv, DOI [arXiv:1807.03819, 10.48550/ARXIV.1807.03819]
[10]   RUIG: Realistic Underwater Image Generation Towards Restoration [J].
Desai, Chaitra ;
Tabib, Ramesh Ashok ;
Reddy, Sai Sudheer ;
Patil, Ujwala ;
Mudenagudi, Uma .
2021 IEEE/CVF CONFERENCE ON COMPUTER VISION AND PATTERN RECOGNITION WORKSHOPS, CVPRW 2021, 2021, :2181-2189