During the last decade, there has been massive investment in CCTV technology in the UK. Currently, there are approximately four million CCTV cameras operationally deployed. Despite this, the impact on anti-social and criminal behaviour has been minimal.
Although most incidents, alternatively events, are captured on video, there is no response because very little of the data is actively analysed in real-time. Consequently, CCTV operates in a passive mode, simply collecting enormous volumes of video data. For this technology to be effective, CCTV has to become active by alerting security analysts in real-time so that they can stop or prevent the undesirable behaviour. Such a quantum leap in capability will greatly increase the likelihood of offenders being caught, a major factor in crime prevention.
The need therefore, is for the persistent analysis of CCTV video footage in real-time. With the recent explosion in CCTV surveillance systems capable of producing Giga and TeraBytes of data on a daily basis, the security sector has, for the first time, had to confront the problems of how to detect and manage events captured by such enormous video databases. Clearly, human analysis is not a cost-effective option. However, a potential solution is the use of computer vision to automatically analyse the video and detect instances of anti-social behaviour, which we term events. Furthermore, we argue that the event, rather than video, or other sensor, data, should be the primary structure of surveillance systems. Thus, in addition to detecting events, we propose an event management framework to provide local and wide-area situation awareness, which is crucial for deciding the appropriate response.
From a technology viewpoint, our primary focus is on surveillance systems that consist of secure, intelligent mobile ad-hoc sensor networks. One such current project involving this technology is the Intelligent Sensor Information System (ISIS), whose aim is to reduce crime on public transport systems. The following are work packages from ISIS:
• Audio-visual gender profiling
• Multi-camera people tracking in 3D
• Event composition with imperfect information from heterogeneous sources