模式识别与人工智能
Thursday, Apr. 3, 2025 Home      About Journal      Editorial Board      Instructions      Ethics Statement      Contact Us                   中文
Pattern Recognition and Artificial Intelligence  2024, Vol. 37 Issue (10): 887-909    DOI: 10.16451/j.cnki.issn1003-6059.202410003
Papers and Reports Current Issue| Next Issue| Archive| Adv Search |
Underwater Image Generation Method Based on Contrastive Learning with Hard Negative Samples
LIU Zijian1, WANG Xingmei1,2, CHEN Weijing1, ZHANG Wansong1, ZHANG Tianzi1
1. College of Computer Science and Technology, Harbin Engineering University, Harbin 150001;
2. National Key Laboratory of Underwater Acoustic Technology, Harbin Engineering University, Harbin 150001

Download: PDF (11095 KB)   HTML (1 KB) 
Export: BibTeX | EndNote (RIS)      
Abstract  Image generation is essential to acquire scarce underwater images, and it is typically reliant on paired data. Considering the limitation of practical access to such data distributions in marine environments, a contrastive learning-based generative adversarial network(CL-GAN) is introduced to overcome the constraints of bijection in image domain. However, the model struggles to learn complex content features from noisy images due to the low quality of negative samples resulting from random sampling. To address this issue, a hard negative sample contrastive learning-based feature level GAN(HCFGAN) for underwater image generation is proposed. To improve the quality of negative samples, a hard negative sampling module(HNS) is designed to mine feature similarity between samples. The hard negative samples close to the anchor sample are incorporated into contrastive loss for complex feature learning. To ensure the complexity and comprehensiveness of negative samples, a negative sample generation module(NSG) is constructed. The adversarial training of NSG and HNS ensures the validity of hard negative samples. To enhance feature extraction capability and training stability of the model for underwater fuzzy images, a contextual feature generator and a global feature discriminator are designed. Experiments show that the underwater images generated by HCFGAN exhibit good authenticity and richness with practical value in underwater image generation.
Key wordsUnderwater Image Generation      Generative Adversarial Network(GAN)      Contrastive Learning      Hard Negative Samples      Feature Extraction     
Received: 26 August 2024     
ZTFLH: TP 391  
Fund:Stable Supporting Fund of Acoustic Science and Technology Laboratory(No.JCKYS2024604SSJS006), Fundamental Research Funds for the Central Universities(No.3072024XX0602)
Corresponding Authors: ZHANG Wansong, Ph.D., associate professor. His research interests include simulation and modeling, and pattern recognition.   
About author:: LIU Zijian, Ph.D. candidate. His research interests include image generation and con-trastive learning. WANG Xingmei, Ph.D., professor. Her research interests include artificial intelligence, image processing and computer vision. CHEN Weijing, Master, assistant engineer. His research interests include underwater image translation. ZHANG Tianzi, Master student. Her research interests include underwater image re-storation and image processing.
Service
E-mail this article
Add to my bookshelf
Add to citation manager
E-mail Alert
RSS
Articles by authors
LIU Zijian
WANG Xingmei
CHEN Weijing
ZHANG Wansong
ZHANG Tianzi
Cite this article:   
LIU Zijian,WANG Xingmei,CHEN Weijing等. Underwater Image Generation Method Based on Contrastive Learning with Hard Negative Samples[J]. Pattern Recognition and Artificial Intelligence, 2024, 37(10): 887-909.
URL:  
http://manu46.magtech.com.cn/Jweb_prai/EN/10.16451/j.cnki.issn1003-6059.202410003      OR     http://manu46.magtech.com.cn/Jweb_prai/EN/Y2024/V37/I10/887
Copyright © 2010 Editorial Office of Pattern Recognition and Artificial Intelligence
Address: No.350 Shushanhu Road, Hefei, Anhui Province, P.R. China Tel: 0551-65591176 Fax:0551-65591176 Email: bjb@iim.ac.cn
Supported by Beijing Magtech  Email:support@magtech.com.cn