计算机与现代化

• 模式识别 • 上一篇    下一篇

 基于船舶融合点迹行为识别的雷达监视系统

  

  1. (江苏科技大学计算机学院,江苏镇江212003)
  • 收稿日期:2017-10-09 出版日期:2018-06-13 发布日期:2018-06-13
  • 作者简介:陈晓利(1989-),女,江苏徐州人,江苏科技大学计算机学院硕士研究生,研究方向:聚类分析,软件应用; 祁云嵩(1967-),男,江苏如皋人,教授,博士,研究方向:机器学习,图像处理; 林嘉炜(1993-),男,江苏靖江人,硕士研究生,研究方向:图像处理,软件应用。
  • 基金资助:
    国家自然科学基金资助项目(61471182); 2017年江苏省研究生实践创新计划项目(SJCX17_0607)

Radar Surveillance System Based on Ship’s Plot-data-fusion Action Recognition

  1. (School of Computer Science, Jiangsu University of Science and Technology, Zhenjiang 212003, China) 
  • Received:2017-10-09 Online:2018-06-13 Published:2018-06-13

摘要: 针对近海监控管理的需求,将电子海图、雷达监控、AIS数据与CCTV技术相结合,通过集成平台对数据的融合处理,实现海洋交通的宏观、动态、实时、立体化的综合智能监控。同时提出一种基于地域信息位置特征点提取(Regional Information Feature Points Extraction, RIFPE)的点迹段划分方法。以某雷达基站为实验点,对已有船只的各项数据运用向量自回归和因子分析进行建模得到区域划分后的轨迹段的轨迹阈值,基于k最近邻算法(kNN)得到对轨迹阈值训练后的结果,最终对测试集进行轨迹行为判别。

关键词: 近海船舶, 点迹段划分, 异构网络, 轨迹聚类, 行为识别

Abstract: In view of the demand of offshore monitoring and management, the integrated intelligent monitoring of marine traffic is realized by combining the electronic chart, radar monitoring, AIS data and CCTV technology, and integrating the data through the integrated platform. At the same time, a segmentation method of Regional Information Feature Points Extraction(RIFPE) based on location information is proposed. Taking a radar base station as the experimental point, using vector autoregression and factor analysis to model the trajectory threshold of the track segment after the region division, based on the k nearest neighbor algorithm(kNN), the result of the trajectory threshold training is obtained. Finally, the trajectory behavior of the test set is judged.

Key words: coasting ship, plot section division, heterogeneous network, trajectory clustering, activity recognition

中图分类号: