多视角视频异常检测算法研究

Research on Multi-view Video Anomaly Detection Algorithms

异常检测在计算机视觉领域有很多研究,在视频监控、活动识别和场景理解等方面有很多潜在应用。一个异常检测系统能够在很大程度上减少工作量和投入的时间。笔者提出了一种基于稀疏编码的多视角视频异常检测算法,整合多视角下的视频信息,利用字典同时在外观和动作上编码规则模式,并稀疏重构正常事件相对应的特征,减少重构错误,提高异常检测的准确性。

Anomaly detection has a lot of research in the field of computer vision,and has many potential applications in video surveillance,activity recognition and scene understanding.An anomaly detection system can greatly reduce the workload and investment time.This paper proposes a multi-view video anomaly detection algorithm based on sparse coding,which integrates multi-view video information,uses dictionary to encode rule patterns both in appearance and action,and sparsely reconstructs the corresponding features of normal events,reduces reconstruction errors and improves the accuracy of anomaly detection.