摘要
目标航迹由一段时间内的点迹组成,按照时间顺序排列,是一种典型的时间序列。航迹数据蕴含着目标的运动特征,在目标跟踪、识别、关联等方向的研究中,实验设计需要大量的航迹数据作为支撑。当前,一部分研究在对实际环境和传感器分析的基础上,探索实现数据的仿真流程,构建与真实世界尽可能相符的仿真数据,并利用仿真数据对方法进行验证;另一部分研究直接采集真实数据,通过对真实数据进行积累和预处理,构建真实数据集进行实验。仿真数据缺乏说服力,难以检验所提方法在真实环境下的性能,而真实数据鲜有公开,增加了研究者获取数据的难度,造成数据闭塞。因此有必要构建并公开专业规范的航迹数据集,为弥补航迹数据集缺少的问题,本文构建并公开了基于自报位航迹数据的海空目标数据集。该数据集包含海面目标航迹数据和空中目标航迹数据两部分,其中海面目标航迹数据来自于船舶自动识别系统数据,空中目标航迹数据来自于广播式自动相关监视数据。通过对两种类型的数据进行长期积累,针对数据特点建立起数据预处理及数据集构建流程,为海空目标跟踪、识别、关联等算法提供了数据支撑。同时本文进一步给出了识别评价指标并利用基线算法进行了实验与分析,结果表明了该数据集的有效性。
The track of a target is a typical time series that contains the points in a period of time arranged in time order.The track data contain the movement characteristics of the target.In the research of target tracking,identification,correlation,etc.,a large amount of track data is needed to support the experimental design.Based on the analysis of the actual environment and sensors,part of the current study explored and realized the simulation process of data to build simulation data that is as consistent as possible with the real world and used simulation data to verify the method.The other part of the study collected real data directly.Real data sets were constructed for experiments by accumulating and preprocessing real data.The simulation data were not convincing,and testing the performance of the proposed method in a real environment was difficult.However,few real data were disclosed,which increased the difficulty for researchers in obtaining real data and resulted in data blocking.Therefore,building and publishing a professionally standardized track data set was necessary.This study constructed and disclosed a sea and air target data set based on self-reported location track data to make up for the lack of a track data set.The data set comprised two parts:ship target track data and air target track data.The ship target track data originated from automatic identification system data,and the air target track data were obtained from automatic dependent surveillance-broadcast data.Through the long-term accumulation of two data types,the data preprocessing and data set construction processes were constructed according to the characteristics of the data,providing data support for ship and air target tracking,identification,correlation,and other algorithms.Moreover,this study provided the identification evaluation index and used the baseline algorithm to conduct the experiment and analysis,and the results showed the effectiveness of the data set.
作者
孔战
崔亚奇
彭煊
熊伟
孙炜玮
顾祥岐
王子玲
夏沭涛
董凯
于洪波
KONG Zhan;CUI Yaqi;PENG Xuan;XIONG Wei;SUN Weiwei;GU Xiangqi;WANG Ziling;XIA Shutao;DONG Kai;YU Hongbo(Naval Aviation University,Yantai,Shandong 264001,China;31092 unit,Beijing 100120,China)
出处
《信号处理》
CSCD
北大核心
2024年第11期2085-2094,共10页
Journal of Signal Processing
基金
国家自然科学基金面上资助项目(62171453)。
关键词
时间序列
自报位航迹
海空目标
数据集
time series
self-reporting track
ship and air target
dataset
