News & Events
NEW SOFTWARE APPLICATIONS INCREASE LANDMINE DETECTION RELIABILITY
November 13, 2003
GAINESVILLE, Fla. --- Princess Diana. Landmines. Afghanistan. We all saw the media images of the popular princess with landmine victims. As a spokesperson, she brought the humanitarian issue to the forefront of the publics' attention. More recently, in the latest 007 movie, James Bond required a high-tech hovercraft to pass through an area scattered with landmines. Paul McCartney and James Taylor are presenting a duet performance to benefit the "Adopt a Minefield" project. Angelina Jolie reported in an interview with Barbara Walters that she fears there are landmines on the property she purchased in Cambodia. (Jolie has good reason to worry-statistics show that 1 in every 245 Cambodians are amputees as a result of landmine explosions.)
Behind celebrity news, there are a myriad of scientists, researchers, and product engineers working to solve the problem of landmine detection. An injury or death from landmines occurs once every 20 minutes. There are up to 100 million active buried landmines around the world and that number is growing. Those seeking a solution in creating efficient minefield clearance have a huge challenge because the current means of landmine detection are often time-consuming, expensive, and give inefficient findings of non-landmine materials that have to be investigated before they can be ruled out.
Dr. Paul Gader, a professor at the Department of Computer and Information Science and Engineering
(CISE) at the University of Florida, is one of the behind-the-scenes researchers tackling the
problem; in his case, from a computer science perspective. Gader spearheads a collaborative
research effort with government and industry that is involved in signal and processing algorithms
developed to detect landmines and unexploded ordnance using a variety of sensor technologies.
The pressing problem is that even though technology in landmine detection systems has advanced dramatically since the late 1990s, all current detection systems continue to produce a high level of false alarms. "The plastic mines can be as small as 1.5 to 2 inches in diameter, so the sensor systems that pick these up also pick up small pockets of water and all kinds of other clutter (non-landmine materials)," Gader explains. "We've got to achieve low false alarm rates (FAR's) but high levels of detection and the two are dichotomous. Furthermore, we need to be able to do this in any type of soil and weather, at an efficient speed." At current clearance rates, it would be centuries before mines in place could be removed.
Gader, who has received over $2 million from the U.S. Military in research funding, and his team of Joe Wilson and Gerhard Ritter at CISE specialize in theoretical and applicable interpretation of data which is gathered from industry-built prototype detection systems such as ground penetrating radar, acoustics and lasers, infrared and hyper-spectral imaging, and electro-magnetic induction. Gader utilizes computer pattern recognition to sort through the abundance of data collected by the sensor systems with a goal of achieving an order of magnitude reduction in false alarm rates by being able to distinguish landmines from clutter.
In recent blind tests of a ground penetrating radar system, Gader and his colleagues' new
pattern-recognition models reduced previous false alarm rates by a factor of 10. "It's a
breakthrough order of magnitude reduction in false alarms." Gader explains. The detection tests
took place over two weeks at two different locations, covering approximately 10,000 square meters
dotted with buried plastic and metal landmines. In September, 2003, Gader and his team presented their
recent developments at the European Union Conference of De-mining held in Brussels, Belgium.
While the landmine issue predominantly focuses on foreign locales, there is a need for detection of unexploded ordinances here in the United States. Military bases have long been the site of weapons testing and occasionally a test explosive does not discharge. A potentially harmful situation surfaces when military test lands are open for building and development; the sites must be cleared of unexploded ordinances before building can commence.
Interdisciplinary applications for innovations in pattern recognition algorithms developed by Gader and his team extend beyond landmine detection. For example, the medical field could use the hyper-spectral detection process for identifying characteristics of blood, such as bilirubin and glucose levels, without having to draw a sample. In archeology, ground penetrating radars have been used successfully to detect ancient dwellings, sink holes, and buried bodies.
When asked if landmine detection is close to achieving 100% reliability and efficiency, Gader shakes his
head. "While we'd love to work our way out of a job by achieving perfection in landmine detection, I
don't think we can solve the problems within the next decade. There is still much work to be done, both
from a military and humanitarian perspective."
Writer: Terri Bailey, 352-373-1041, TBailey567@aol.com
Source: Paul Gader, 352-392-1526 pgader@cise.ufl.edu
