雷达与AIS对海上低速目标航迹关联的特征选择  

Feature Selection for Correlation of Radar and AIS with Low Speed Target Tracks at Sea

作　　者：唐裕峰 宋瑶 武浩正 靳标 TANG Yu-feng;SONG Yao;WU Hao-zheng;JIN Biao(School of Electronic Information,Jiangsu University of Science and Technology,Zhenjiang Jiangsu 212100,China)

机构地区：[1]江苏科技大学海洋学院,江苏镇江212003

出　　处：《计算机仿真》2023年第12期193-199,共7页Computer Simulation

基　　金：国家自然科学基金(61701416,61871203,62001194);江苏省基础研究计划(自然科学基金)资助项目(BK20211341);江苏省研究生科研与实践创新计划项目(SJCX21_1749)。

摘　　要：将雷达和自动识别系统(Automatic Identification System, AIS)进行航迹关联能够实现信息互补,提高船舶航行数据的可信度。传统方法利用目标的位置信息进行数据关联,在处理海上低速目标航迹关联问题时,抗噪声性能差,关联精度不高。针对上述问题提出一种航向直方图统计和动态时间规整相结合的关联方法。首先利用时间段关联度对数据预处理,然后综合利用动态时间规整和航向直方图统计提取航迹间的相似度特征,最后将这些特征和距离特征相结合并利用机器学习训练关联模型。实验结果表明:相比其它算法,上述算法充分利用时间段关联度和航迹间的相似度特征,降低了数据的复杂度,具有较好的抗噪声性能,提高了航迹关联精度。Linking radar and Automatic Identification System(AIS)with track correlation can realize information complementarity and improve the credibility of ship navigation data.The traditional method uses the position informa⁃tion of targets for data association,but when dealing with the problem of low-speed targets trajectory association at sea,their anti noise performance is poor and the correlation accuracy is not high.To solve this problem,this paper proposes a correlation method based on the combination of heading histogram statistics and dynamic time warping.Firstly,the data are preprocessed by using the correlation degree of time period;Then the similarity features between tracks were extracted by using dynamic time warping and heading histogram statistics;Finally,these features were combined with distance features and the machine learning was used to train the correlation model.The experimental results show that,compared with other algorithms,this algorithm makes full use of the characteristics of time segment correlation degree and track similarity,reduces the complexity of data,has better anti noise performance,and im⁃proves the accuracy of track correlation.

关 键 词：航迹关联 直方图统计 特征组合 

分 类 号：TN911[电子电信—通信与信息系统]