摘要
针对不同应用场景给出了不同的徘徊定义,并提出一种基于三维模型的行人徘徊行为检测算法。该方法利用融合前景检测和行人检测的方法获取目标,并利用摄像机标定信息和图像信息建立目标的三维模型。针对传统算法无法获取目标真实轨迹的问题,利用三维模型对分块目标跟踪方法进行改进,以提取目标的真实运动轨迹。根据不同的应用场景对目标轨迹进行分析,判断目标是否有徘徊行为。实验结果表明,该方法能够获得较为准确的目标三维模型,得到目标的真实运动轨迹,从而鲁棒地、快速地检测目标的徘徊行为。
Different wandering definitions are proposed for different application scenarios, and a pedestrian wandering behavior detection algorithm based on 3D model is proposed. The method uses the method of fusion foreground detection and pedestrian detection to obtain the target, and uses the camera calibration information and the image information to establish the 3D model of the target. Aiming at the problem that the traditional algorithm can not obtain the true trajectory of the target, the 3D model is used to improve the segmented target tracking method to extract the true trajectory of the target. According to the different application scenarios to analyze the target trajectory to determine whether there is wandering behavior of the target. The experimental results show that the proposed method can obtain accurate three-dimensional model of the target and obtain the true trajectory of the target, so as to detect the wandering behavior of the target robustly and quickly.
出处
《计算机应用与软件》
2017年第4期149-156,共8页
Computer Applications and Software
基金
国家科技支撑计划项目(2013BAH09F01)
上海市科委科技创新行动计划项目(14511106900)
关键词
智能视频监控
徘徊行为检测
三维重建
行人跟踪
行人检测
Intelligent video surveillance
Pedestrian wandering detection
3D reconstruction
Pedestrian tracking
Pedestrian detection
