The Camera Never Blinks
New technology being developed at UMaine could help in stepped-up
building security efforts
Walk into a bank, government building
or your favorite retail store and chances are they're there. Watching.
Waiting to record your every move.
Security video cameras provide an unblinking record of people's
activities hour after hour, day after day. But it takes a person on the
other end of the camera to monitor, analyze and interpret the images,
and then alert safety personnel or law enforcement officials in the
event of a threat or an emergency.
But what if technology could automatically analyze security video taken
outside busy facilities such as Oklahoma City's Alfred P. Murrah Federal
Office Building or inside Boston's Logan Airport, allowing authorities
to be alerted instantly to safety concerns? At The University of Maine,
a research team led by scientists in the Department of Spatial
Information Science and Engineering (SIE) is developing new automated
image analysis techniques. The work is funded by a half-million-dollar
grant from the National Science Foundation (NSF).
Peggy Agouris, associate professor
and researcher in the University's National Center for Geographic
Information and Analysis.
Peggy Agouris, associate professor and
researcher in the University's National Center for Geographic
Information and Analysis, leads the project with two other UMaine
faculty members, Anthony Stefanidis and Kate Beard. Agouris came to
UMaine in 1995 from the Swiss Federal Institute of Technology in Zurich,
The scientists are collaborating with faculty members at the University
of California at Riverside and Penn State University to harness the
volumes of data collected constantly by satellites, autonomously
operated vehicles, video cameras and other devices.
The objective is to automatically extract information from digital and
video image sequences, and detect changes in behavior patterns without
relying on one or more people to do the work.
"Things change over time, and this change is inherent in the data set,"
says Agouris, who heads SIE's Digital Image Processing and Analysis
Laboratory. "Our goal is to develop a means to make this change explicit
and to communicate it to the people who are interested in this
information. This is not just about detecting the change. It is about
developing (an analytical) framework that is independent of peoples'
views as to what changes."
Such new technology could not only analyze ongoing security video but
also could squeeze more information out of images collected in the past
for military and civilian purposes.
Beyond the security benefits, the automated image analysis technology
has implications for traffic management, agricultural assessments, land-
use planning and environmental monitoring.
In addition, automated techniques could be used in response to disasters
such as the World Trade Center tragedy, Agouris says. By considering
locations and designs of buildings, and the effect of their collapse,
engineers and architects could better understand the risks associated
with densely developed urban areas.
Agouris and her UMaine colleagues soon will be expanding their image
management research with a newly funded $1.5 million multi-year NSF
grant. In that project, scientists will develop the foundation for a
type of "geospatial Internet" that will contribute to making accurate
geospatial information accessible to anyone, anytime, anywhere.
by Nick Houtman
December, 2001/January, 2002
for more stories from this issue of UMaine Today Magazine.