多模态信息融合舰船目标识别研究进展
A Review of Ship Target Recognition Based on Multi-modal Information Fusion
吴文静 1王中训 1但波 2邢子杰3
作者信息
- 1. 烟台大学物理与电子信息学院,山东 烟台 264005
- 2. 海军航空大学岸防兵学院,山东 烟台 264001
- 3. 山东工商学院,山东 烟台 264026
- 折叠
摘要
舰船目标识别的信息源主要来自现代高分辨率成像雷达形成的舰船目标信息,包括高分辨距离像、船舶自动识别系统信息以及合成孔径雷达成像.在对海探测环境相对复杂的情况下,基于单模态信息对海上舰船目标识别的能力有限,而利用多模态信息融合将更有益于实现对海上目标高效的侦察监视和识别.首先,对单模态舰船目标识别方法进行梳理和总结,分析目前不同舰船目标识别方法存在的优势和不足;然后对多模态信息融合舰船目标识别常用数据集进行介绍,并对新方法、新模型进行了深入分析;最后对舰船目标识别未来发展趋势进行展望,为后续基于多模态信息融合的舰船目标识别方法研究提供参考.
Abstract
At present,the information source for ship target recognition mainly comes from ship target infor-mation formed by modern high-resolution imaging radar,including high-resolution range profiles,ship automatic recognition system information,and synthetic aperture radar imaging.In the context of relatively complex sea exploration environments,the ability to identify ship targets on the sea based on single mode information is lim-ited,and utilizing multi-modal information fusion will be more beneficial for achieving efficient reconnaissance,monitoring,and identification of sea targets.This article first sorts out and summarizes the methods of single mode ship target recognition,analyzes the advantages and disadvantages of different ship target recognition methods at present,introduces the commonly used datasets of multi-modal information fusion ship target recog-nition,and conducts in-depth analysis of new methods and models.Finally,it looks forward to the future devel-opment trend of ship target recognition.This article can provide reference for future research on ship target rec-ognition methods based on multi-modal information fusion.
关键词
高分辨距离像/船舶自动识别系统/合成孔径雷达/多模态信息融合/舰船目标识别Key words
HRRP/AIS/SAR/multi-modal information fusion/ship target recognition引用本文复制引用
基金项目
国防科技重点实验室项目(2021-JCJQ-LB-018)
出版年
2024
NETL
NSTL
维普
万方数据