基于视频图像的传统舰船识别方法一般仅能够识别和分类舰船类型，较难对舰船的具体型号做出准确判断，难以达到实际应用要求。本文研究目的在于分析基于声像情报信息的舰船型号识别主要特征与途径，研究主要识别方法与流程。本文研究综合声像信息提取、检测、分析等技术手段，结合人工智能识别模型训练方法，构建识别模型，实现对舰船型号的细分识别。研究结果显示，基于声像情报信息的舰船型号识别在实践中表现出较高的准确度，结合多源声像情报信息的综合利用，使得识别结果更为可靠。此项舰船型号识别方法在装备科技信息领域中有较好的应用前景，能够为声像情报的研究工作提供支撑保障。
Traditional ship identification methods based on video images can generally only identify and classify ship types, and it is difficult to accurately determine the specific models of ships, which cannot meet the requirements of practical application. The purpose of this study is to analyze the main features and approaches of ship model recognition based on audio-visual intelligence information, and to study the main recognition methods and processes. This study combines audio-visual information extraction, detection, analysis and other technical means, combined with artificial intelligence recognition model training methods, to build a recognition model and achieve detailed recognition of ship models. The research results show that the ship model recognition based on audio-visual intelligence information has a high accuracy in practice. The combined use of multi-source audio-visual intelligence information makes the recognition results more reliable. This method of ship model recognition has a good application prospect in the field of equipment and technology information, and can provide support for the research work of audio-visual intelligence.
2024,46(1): 158-163 收稿日期：2023-09-18
DOI：10.3404/j.issn.1672-7649.2024.01.027
分类号：TP391
作者简介：丁禹懿(1992-),男,硕士,工程师,研究方向为声像情报和数字媒体
参考文献：
[1] 杨龙, 苏娟, 黄华, 等. 一种基于深层次多尺度特征融合CNN的SAR图像舰船目标检测算法[J]. 光学学报, 2020, 40(2): 132-140.
[2] 冯雷, 耿英楠. 基于hough变换的倾斜文档图像快速校正[J]. 内蒙古师范大学学报(自然科学汉文版), 2017, 46(2): 219-222.
[3] 赵巍. 基于人工智能技术的图像文字识别程序开发与设计[J]. 电脑编程技巧与维护, 2023(4): 125-126＋161.
[4] 薛亮, 周游. 基于知识图谱的舰船信息分析识别技术研究[J]. 舰船电子工程, 2021, 41(10): 48-51＋145.
[5] 张建丰. 基于视差信息的多重运动图像轮廓提取仿真[J]. 计算机仿真, 2020, 37(9): 214-217.
