摘要
在信息融合领域,利用数据挖掘中的异常检测技术,可以基于目标的多维航迹特征来挖掘目标的异常行为。现有轨迹异常检测方法主要检测目标的位置异常,没有充分利用目标的属性、类型、位置、速度和航向等多维特征,在挖掘目标的异常行为时具有局限性。通过定义多因素定向Hausdorff距离和构造多维度局部异常因子,提出了一种基于多维航迹特征的异常行为检测方法,通过对多维航迹数据的异常检测,实现对目标异常行为的挖掘。在仿真军事场景和真实的民用场景上进行了实验分析,所提方法都能有效的检测出目标的异常行为。
In the information fusion domain, anomalous behaviors could be mined based on multidimensional trajectory characteristics by using the anomalous detection technique in data mining. Previous trajectory anomaly detection algorithms mainly detect the position anomalies, without making full use of the attribute, category, position, velocity, and course characteristics. In order to overcome this limitation, we define the multi-factor Hausdorff distance, construct the multidimensional local outlier factor, and propose a method for detecting anomalous behaviors based on multidimensional trajectory characteristics. The method can mine anomalous behaviors based on detecting multidimensional trajectories. We conducted experiments on simulated military scenario and real civilian scenario, the proposed method can effectively detect the anomalous behavior of the target.
出处
《航空学报》
EI
CAS
CSCD
北大核心
2017年第4期249-258,共10页
Acta Aeronautica et Astronautica Sinica
基金
国家自然科学基金(61531020
61471383
91538201)
山东省科技重大专项基金(2015ZDZX01001)~~
关键词
异常行为
航迹
多维特征
局部异常因子
HAUSDORFF距离
anomalous behavior
trajectory
multidimensional characteristics
local outlier factor
Hausdorff distance
