Intelligent Perception Algorithms for Sonar Images: A Review

• Published in: 水下无人系统学报 Vol. 33; no. 3; pp. 559 - 572
• Main Authors: Wenpei JIAO, Jie LI, Chunyan ZHANG, Guangming XIE, Wendong XIAO, Jianlei ZHANG
• Format: Journal Article
• Language: Chinese
• Published: Science Press (China) 01.06.2025
• Subjects: deep learning; intelligent perception; ocean exploration; sonar image
• ISSN: 2096-3920
• DOI: 10.11993/j.issn.2096-3920.2024-0127

Intelligent perception algorithms for sonar images are vital in ocean exploration and underwater rescue. In recent years, deep learning has achieved remarkable advancements in intelligent perception tasks related to sonar images. This paper provided a comprehensive review of the field, focusing on sonar image datasets and data augmentation techniques, classical sonar image processing algorithms, and deep learning-based sonar image processing methods. By summarizing open-source datasets and commonly used data augmentation techniques, the paper provided data support for future research efforts. Additionally, this paper systematically analyzed the application and evolution of both classical and deep learning-based algorithms across various tasks, offering researchers an overview of the current state of the field. Finally, based on the research progress, the paper predicted future research directions. It was pointed out that the interpretation ability of sonar images could be further improved by obtaining larger-