智能视频监控系统异常行为检测算法研究综述

随着公共安全需求的快速增长,监控摄像头数量不断增多,视频监控数据呈爆炸式增长;传统的视频监控系统难以对如此海量的数据进行理解分析,因此智能视频监控系统应运而生;作为一个跨学科的研究领域,智能视频监控系统异常行为检测技术迎来重大机遇的同时也面临不少挑战;为了更好地研究智能视频监控系统异常行为检测算法,梳理了相关研究并从原理上对不同算法进行分类,对基于能量、基于聚类、基于重构、基于推断以及基于深度学习几个不同依据的算法进行对比分析,归纳了各类算法的分支研究方向,接着简要介绍了异常行为检测常用的公开数据集,最后讨论了目前异常行为检测算法所面临的挑战并针对性地提出了未来智能视频监控系统异常行为检测算法的可行研究方向。

A Survey of Detection Algorithms for Abnormal Behaviors in Intelligent Video Surveillance System

With the rapid growth of public security demand, the number of surveillance cameras is increasing, and video surveillance data is growing explosively. The traditional video monitoring system is difficult to understand and analyze such a large amount of data, so the intelligent video monitoring system came into being. As an interdisciplinary research field, intelligent video surveillance system abnormal behavior detection technology is facing great opportunities and challenges. In order to study the abnormal behavior detection algorithm of intelligent video surveillance system better, this paper combed the related research and divides different algorithms into energy based, clustering based, reconstruction based, inference based and depth learning based, and made comparative analysis and summarizes the branch research directions of various algorithms. Then, the common public data sets of abnormal behavior detection are introduced. Finally, this paper discussed the challenges of the current abnormal behavior detection algorithm and put forward the feasible research direction of the future intelligent video monitoring system abnormal behavior detection algorithm.