论文部分内容阅读
对拥挤地区人群行为的检测对于多种安全任务都至关重要。这个问题包括2个部分:对拥挤程度的估计和对异常的检测。本文定义了一个运动特征来进行群体建模和实时检测。在此基础上,通过分析大量的运动信息。提出了一种通过自定义的能量函数来评估观测区域内的拥挤程度和异常事件检测的新方法。此方法被用于地铁的视频监控系统中,具有良好的效果。
The detection of crowd behavior in crowded areas is crucial for a variety of safety tasks. This question consists of two parts: the degree of congestion and the detection of abnormalities. In this paper, a motion feature is defined for population modeling and real-time detection. On this basis, by analyzing a large number of sports information. A new method of assessing the degree of congestion and detecting anomalous events in the observation area by using a custom energy function is proposed. This method is used in the subway video surveillance system, with good results.