Welcome to ICONS project

     This project is related to security systems and human behaviour modelling
          Most of the computer models are subject of future security systems.
Home | Industrial software | Golden eye | Clouds | Robot Vision | Aims | Icons | SPOT | CV | Pictures | Contact     
 

DTI / EPSRC LINK Project

 

ICONS: Incident Recognition for Surveillance and Security

Incident Recognition for Surveillance and Security using Computer Vision (Automatic detection of 'significant events' or 'unusual behaviour') In today's society, the need for reliable and accurate safety and asset monitoring is essential. The science of Computer Vision can now address these needs utilising contemporary computers and cameras provided suitable algorithms can be developed to understand image content. The ICONS project was formed between Queen Mary - University of London, Safehouse Technologies Ltd, British Airport Authority plc, BT Exact Ltd, & Heritage Protection Ltd to research the techniques needed to recognize significant events occurring in CCTV images that relate to abnormal or undesirable behaviour in a scene, as well as detecting 'zero motion' events that include medium to long term changes in the scene such as objects being placed or removed in or from secure areas. Safehouse has developed a software platform incorporating the early outcomes of this research and is actively trialling the applications depicted.

July 2000-July 2003

About the Project

The ICONS project is a 3 year endeavour funded by DTI and EPSRC, and links QMW college with several industry partners, namely Safehouse technologies, BAA plc, BT and Heritage Protection Ltd.   The aim of the project is to advance and exploit state-of-the-art research into:

  • zero motion detection: detection of medium- to long-term visual changes in a scene
    • eg: deployment of a parcel bomb, theft of a precious item
  • behaviour recognition: characterise and detect undesirable behaviour in video data
    • eg: trespassing, patterns of theft, violence


Zero Motion Detection

The task is to distinguish background from foreground, even though the scene appearance is changing due to varying illumination and transitory movements, such as human activity or a swaying tree.

By learning pixel variations and adapting to slow changes over time, sudden changes in the scene can be detected.  In the movie example below, a box is placed in the room.  Temporary changes such as the person walking are ignored, but after the box has been present for a while it is detected as a foreground object.

     mpeg demonstration (285 kB)
 

Behaviour Recognition

In dynamic scenes,  the task is to separate normal from abnormal behaviour.  To do this, pixel phase variations are learned over time for "normal" behaviours.  When a variation is observed that does not fit the model of allowable variations, it is flagged as an abnormal occurrence.

In the movie example below, the system has been trained on sequences of the subject entering the room and walking up and down.  During testing, the same behaviour is ignored because it is "normal".  However when the subject stops moving or jumps, that behaviour is detected as abnormal.
 

 

     mpeg demonstration (1.3 MB)
 
 
Current progress

A major component of the project is the detection of changes in the background of scenes in which there is a lot of activity.

Fig 1. Raw picture from CCTV camera

Fig 2. The result after processing

he first photo (Fig. 1) is a still shot from a video sequence taken by a CCTV camera looking at the forecourt of terminal 2 at Heathrow airport. The image in the camera is constantly changing due to the wind moving trees, clouds moving across the sun and camera shake as well as the constantly moving traffic.
The second photo (Fig. 2) is the same shot after processing (in real time) using the algorithms being developed. We have detected that two vehicles have been parked in unexpected places. Notice that the standard background has been successfully eliminated.

Scene splitting according to background behavoir: You can see the road and building splitting from pixel learning algorithm. To represent the areas with different behavoir they are marked with different colours.

This algorithm learns pixel behavoir and splits the road form the scene

These algorithms bellow slit the scene in different activity zones

 

 

Organisations Involved in the ICONS Project 

The ICONS project is funded and managed by DTI and EPSRC, and involves the following academic and industrial collaborators:
 

 

 QMW College (project leader) 
Safehouse Technologies (lead industrial partner)

 

 BAA plc
British Telecom
 

 

 Heritage Protection Ltd.

 
1