The Computer Vision group is one of the largest in the UK and internationally leading in its work on the extraction of object behaviour models and dynamic face models from image sequences and live video. The work has been widely applied to vehicle and people detection, object tracking, counting and recognition in public space CCTV, human gesture recognition for visually mediated interaction and abnormal behaviour recognition in visual surveillance. A current significant focus is in crime prevention, with work on real-time surveillance and biometrics.
Current research activities
- Statistical machine learning
- Time series analysis
- Dynamic Bayesian graph models
- Multi-view geometry
- Multi-modal data fusion
- Neurobiologically inspired vision
- Image compression
- Multi-camera networks
- Image analysis
- Video tracking
Areas of expertise
Core expertise includes dynamic scene analysis, mathematical modelling, multi-view geometry, pattern recognition and learning, biologically inspired vision and image compression. An additional new research line concerns the extraction of 3D information from image sequences using geometric information. In particular, research is being carried out into self-calibration of cameras and 3D metric reconstruction of scenes viewed by uncalibrated cameras. Work is also being undertaken to develop novel computer vision algorithms and hardware based on neurobiological principles.
The group’s research attracts significant interest from the industry and the government. Over the years, the group has been funded by the EU ESPRIT, HCM Networks, UK EPSRC, BBSRC, DTI, Ministry of Defence, the Wellcome Trust, the Royal Society, BAA, BT Labs, BBC R&D, and Qinetiq.
Significant projects the group has been involved with in past years include:
- ICONS developing incident recognition techniques for surveillance and security;
- VIGOUR, an Integrated Vision System integrating face detection, head tracking, human body modelling, feature extraction, behaviour interpretation;
- INSIGHT working on semantic content analysis of CCTV recordings for automatic semantic video tagging, search and pro-active sampling.
These research projects see the group collaborate with several government and industry partners and end-users include the Ministry of Defence, Home Office, Petards Vision Ltd and Safehouse International Ltd.
The group want to continue looking at man and machine interaction and using cameras to interpret the environment as humans would do.
The group organises its own seminar series. The specialised, research-led Masters course in Intelligent Imaging Systems exploits the expertise within the group, while members also contribute to other advanced Masters programmes in the Department.