|
In this project, the following activities are pursued:
Video Surveillance Commercial R&D
The group undertakes research in automated visual surveillance
and analysis for commercial applications. Some previous work has
been patented and licensed. Contract work is undertaken in the
realm of detection of suspicious events, reduction of false alarms,
etc. Our partner for work in Background Change detection is iOmniscient
Pty Ltd (see iOmniscient.com).
This commercialised work has recently won prestigious awards (see
http://www.cs.adelaide.edu.au/~mjb/awards/index.html).
Networked Video Surveillance
Interpreting imagery generated by a network of cameras viewing
an extended area poses many important problems. A system has been
developed that records suspicious events across a network of sensors,
uploading image sequences to a server repository that facilitates
browsing. A key problem is then "how may useful information
be retrieved from the video repository?". Another is "how
may information be tracked across a network of cameras?".
The following example problem highlights some of the challenges
in this area. Suppose that a city has numerous outdoor cameras
that monitor their surroundings, generating a network of video-image
repositories. If a robbery takes place at a certain location and
a known vehicle is involved in the getaway, how might the network
of repositories be searched in order to trace the path the vehicle
took through the city?
Fundamental Research
As a means of supporting the above endeavours, we research new
algorithms and parameter estimation techniques for the analysis
of videos. Examples are our work on automatically estimating the
synchrony of videos taken of the same scene, and our work on automated
geometric methods for post-production processing of movies.
PUBLICATIONS found here
|
