基于机器视觉的海上目标智能化预警观测过程模型与技术方法

详细信息    查看全文 | 推荐本文 |

英文篇名：A General Approach for Automatic Marine Target Monitoring and Tracking Based on Machine Vision 作者：应文 ; 杨志霞 ; 符亚明 ; 杨建平 英文作者：YING Wen;YANG Zhi-xia;FU Ya-ming;YANG Jian-ping;China Academy of Electronic and Information Technology;China University of Geosciences;CETC Ocean Information Co.,Ltd; 关键词：海上态势感知 ; 智能化管控 ; 机器视觉 ; 光电监控系统 英文关键词：Maritime State Awareness;;Intelligent Management and Control;;Machine Vision;;Photoelectric Monitoring System 中文刊名：中国电子科学研究院学报 英文刊名：Journal of China Academy of Electronics and Information Technology 机构：中国电子科学研究院;中国地质大学(武汉);中电科海洋信息技术研究院有限公司; 出版日期：2019-08-20 出版单位：中国电子科学研究院学报 年：2019 期：08 语种：中文; 页：84-93 页数：10 CN：11-5401/TN ISSN：1673-5692 分类号：TP391.41;P715

摘要

海洋态势感知与目标观测技术正向着自动化、智能化发展。光电监控系统作为海洋目标活动态势感知中不可或缺的必要手段,在面向长期无人值守的持续海上预警观测应用场景中,其自主运行能力受到极大考验。本文基于多年研究以及相关项目实践,提出了基于机器视觉的海上目标智能化预警观测的一般过程模型与相关的技术实现方法,通过基于深度学习的目标识别模型和一体化的联动跟踪机制驱动光电传感器智能自主运行,在全自动托管的情况下实现目标发现、跟踪、识别、取证、再发现闭环持续工作,为海上目标活动视觉特征提取与信息服务提供有力支撑。
        Maritime state awareness and marine object observation technology are developing towards automation and intellectualization. For the photoelectric monitoring system,which is the core means of maritime state awareness and target activity forensics,the ability of autonomous operation is greatly tested under the actual situation of long-term unattended observation. Based on many years of research and related project practice,this paper presents the general process and technical method of intelligent early warning and observation of marine targets based on machine vision. The intelligent autonomous operation of photoelectric sensors is driven by the target recognition model based on deep learning and the integrated linkage tracking mechanism,which provides strong support for visual feature extraction and information service of marine target activities.

引文

[1]陈质二，俞建成，张艾群．面向海洋观测的长续航力移动自主观测平台发展现状与展望．海洋技术学报，2016,35(1):122-130.
    [2] Uzunoglu C,Guedes S. A system for the hydrodynamic design of tension leg platforms of floating wind turbines[J]. Ocean Engineering,2019,171.
    [3]王冰，职秦川，张仲选，耿国华，周明全．灰度图像质心快速算法．计算机辅助设计与图形学学报，2004,16(10).
    [4]罗钟铉，刘成明．灰度图像匹配的快速算法[J]．计算机辅助设计与图形学学报，2005,17(5):966-970.
    [5]冉洪亮．基于模型预测控制算法的无人驾驶车辆路径识别与跟踪控制[D]．重庆大学，2018.
    [6]李娟．基于数据融合的危险驾驶行为识别与车辆跟踪算法研究[D]．武汉工程大学，2017.
    [7] COCO-Common Objects in Context. http://cocodataset.org/#overview
    [8] AG,Zhu ML,Chen B,et al. MobileNets:Efficient convolutional neural networks for mobile vision applications.Howard,2017.
    [9] Szegedy C,Vanhoucke V,Ioffe S,et al. Rethinking the inception architecture for computer vision[C]//2015IEEE Conference on Computer Vision and Pattern Recognition(CVRR). IEEE. 2015.
    [10] Jifeng D,Yi L,et al. R-FCN:Object Detection via Region-based Fully Convolutional Networks. Conference on Neural Information Processing Systems(NIPS),2016.
    [11] Abadi M,Agarwal A,Barham P,et al. Tensor Flow:Largescale machine learning on heterogeneous distributed systems. 2016.
    [12] Krotkov,Paul E. Active Computer Vision By Cooperative Focus And Stereo[M]. Springer-Verlag,1989.