Network-Centric Smart Video System and Analytics

Video monitoring and surveillance (VMS) is now a familiar phenomenon in public and private spaces, manifested by wide visibility of cameras in a huge variety of environmental settings. Whether we like it or not, video monitoring/surveillance network infrastructures are here to stay and the growth trend will continue for the foreseeable future. There has been increasing awareness of VMS by society at large judging from attention grabbing media headlines over the past two years such as “Britain is surveillance society”, “Big brother is watching us all”, “Word on the street … they’re listening”, “CCTV could track branded suspects”, “80 per cent of CCTV images ineffective”, etc.

Whilst the number of cameras keeps growing, the projected economic, commercial and societal impact in terms of operational efficiency, cost reduction, effectiveness for crime prevention and detection, and ‘feel safe’ factor, etc, of this wave of investments in VMS cannot be fully realised or sustained, unless the increase in cameras is matched by continuous research and development in ‘behind the scenes’ VMS infrastructures and technologies in a coherent way.

This research theme aims to investigate / encourage a paradigm shift in next-generation value-added video management systems and services, especially smart visual systems for monitoring and surveillance in realistic operational scenarios. One of the two pillars of this paradigm is an IP-based network-centric, configurable and scalable infrastructure capable of securely delivering high-resolution high-quality (live or storage) video content on demand from any point to any point. The other is the co-operative and distributed video and audio sensors and intelligent video analytics algorithms, capitalising on human perception and reasoning capability whilst incorporating scenario- and domain-based knowledge, to automatically interpret the complex environment data from heterogeneous sources for decision making.

The key to all of these is effective acquisition and processing and efficient delivery via the network to the point of consumption. We will look into crucial issues of feasibility, testbed building, robust video analysis algorithms research, prototyping of selected applications and services, pilot demonstrators and trials in connection with customers, including BT stakeholders.

Dr. Li-Qun Xu leads a team of multidisciplinary researchers, in close collaboration with University academics, working on the theme.

-- IanKegel - 17 Jun 2008