Scalable Robust Video Surveillance over Constrained Networks

对大量人员的实时监控对于诸如高效的服务提供和针对普通犯罪和恐怖主义的安全等应用变得越来越重要。由于收集的数据量很大，使用人工操作员完成这些任务是不可行的。现有的自主视频监控系统容易出现大量误报，而且往往需要昂贵的硬件。该提案试图通过使用严格的统计信号处理方法来最佳地融合来自低成本摄像机网络的信息来解决这两个困难。

Real-time monitoring of large numbers of people is becoming increasingly important for applications such as efficient service delivery and security against both common crime and terrorism. The use of human operators for such tasks is infeasible due to the large amount of data collected. Existing autonomous video surveillance systems are prone to high numbers of false alarms and often require expensive hardware. This proposal seeks to address both difficulties by using rigorous statistical signal processing methods to optimally fuse information from a network of low-cost cameras.