首页 | 期刊介绍 | 编委会 | 道德声明 | 投稿指南 | 常用下载 | 过刊浏览 | In English
引用本文:王飞,刘梦婷,刘雪芹,秦志亮,马本俊,郑毅.基于YOLOv3深度学习的海雾气象条件下海上船只实时检测[J].海洋科学,2020,44(8):197-204.
【打印本页】   【HTML】   【下载PDF全文】   查看/发表评论  下载PDF阅读器  关闭
←前一篇|后一篇→ 过刊浏览    高级检索
本文已被:浏览 42次   下载 25 本文二维码信息
码上扫一扫!
分享到: 微信 更多
基于YOLOv3深度学习的海雾气象条件下海上船只实时检测
王飞1,2,3, 刘梦婷1,2,3, 刘雪芹1,2,3, 秦志亮1,2,3, 马本俊1,2,3, 郑毅1,2,3
1.哈尔滨工程大学 水声技术重点实验室, 黑龙江 哈尔滨 150001;2.哈尔滨工程大学 工业和信息化部 海洋信息获取与安全工信部重点实验室, 黑龙江 哈尔滨 150001;3.哈尔滨工程大学 水声工程学院, 黑龙江 哈尔滨 150001
摘要:
海雾气象条件下船只高精度检测识别面临较大困难,传统的目标识别、定位方法效果差强人意。作者围绕海雾气象条件下不同类型船只的实时检测问题,提出一种基于YOLOv3深度学习的实时海上船只检测新思路。首先构建清晰图片和模糊图片(海雾、雨)的判别方法,实现图片清晰度分类处理;其次为提高海雾气象条件下海上船只的实时检测精度,消除海雾遮挡对目标识别的影响,运用暗通道先验去雾方法对含有海雾的图像实行去雾;最后基于YOLOv3深度学习算法对精细处理后的图像进行船只实时检测。实验结果表明该方法能够在海雾气象条件下高效、准确地检测到船只,对海上复杂环境条件下的船只实时检测研究具有一定的理论指导意义。
关键词:  船只检测  暗通道先验去雾  深度学习  YOLOv3
DOI:10.11759/hykx20200106001
分类号:TP391.4
基金项目:国家自然科学基金资助项目(No.41304096,41876053);声学科学与技术实验室基金资助项目(GK2050260214,GK2050260217,GK2050260218,KY10500180084,KY10500190031)
Real-time detection of marine vessels under sea fog weather conditions based on YOLOv3 deep learning
WANG Fei1,2,3, LIU Meng-ting1,2,3, LIU Xue-qin1,2,3, QIN Zhi-liang1,2,3, MA Ben-jun1,2,3, ZHENG Yi1,2,3
1.Acoustic Science and Technology Laboratory, Harbin Engineering University, Harbin 150001, China;2.Key Laboratory of Marine Information Acquisition and Security, Harbin Engineering University, Ministry of Industry and Information Technology, Harbin 150001, China;3.College of Underwater Acoustic Engineering, Harbin Engineering University, Harbin 150001, China
Abstract:
High-precision detection and recognition of vessels under sea fog weather conditions encounter numerous difficulties, and traditional target recognition methods rarely achieve satisfactory results. This study focuses on detection of different types of vessels under complex sea fog weather condition and proposes a new real-time marine vessel detection method based on YOLOv3 deep learning. First, a discriminative method for clear and fuzzy pictures (e.g., sea fog and rain) is constructed to realize the classification and processing of picture sharpness. Then, to improve the detection precision of marine vessels under complex sea fog weather condition, the dark channel prior dehazing algorithm is used to suppress the effect of sea fog occlusion on target recognition. Finally, real-time detection of vessels is performed on the finely processed images on the basis of the YOLOv3 deep learning algorithm. The experimental results show that the proposed method can be used to efficiently and accurately detect vessels under sea fog weather condition and has certain theoretical guidance significance for research on target vessel recognition under complex marine conditions.
Key words:  vessel detection  dark channel prior dehazing algorithm  deep learning  YOLOv3
版权所有 《海洋科学》编辑部 Copyright©2008 All Rights Reserved
主管单位:中国科学院 主办单位:中国科学院海洋研究所
地址:青岛市南海路七号  邮编:266071  电话:0532-82898755  E-mail:marinesciences@qdio.ac.cn
技术支持:北京勤云科技发展有限公司