弹道
异常检测
跟踪(教育)
计算机科学
异常(物理)
车辆跟踪系统
计算机视觉
人工智能
数据挖掘
实时计算
卡尔曼滤波器
心理学
教育学
物理
天文
凝聚态物理
作者
Juan A. Fernández,Jorge García‐González,Rafaela Benítez-Rochel,Miguel A. Molina‐Cabello,Ezequiel López‐Rubio
标识
DOI:10.1007/978-3-031-06527-9_17
摘要
Vehicle trajectories extracted from traffic video sequences can be helpful for many purposes. In particular, the analysis of detected anomalous trajectories may enhance drivers’ safety. This work proposes a methodology to detect anomalous vehicle trajectories by using a vehicle detection, a vehicle tracking and a processing of the tracking information steps. Once trajectories are detected, their velocity vectors are estimated and an anomaly value is computed for each trajectory by comparing its vector with those from its nearest neighbours. The management of these anomaly values allows considering which trajectories are suitable to be potentially anomalous considered. Real and synthetic videos have been included in the experiments to perform the goodness of the proposal.
科研通智能强力驱动
Strongly Powered by AbleSci AI