HOME
Seeing the Unseen: Infrared computer vision system could help
make meat products safer, tastier, and less costly to produce
By Jane M. Sanders
Consumers can expect that meat products will be safer, tastier, and
less costly to produce within a year or so as the food processing industry
begins to use infrared computer vision scanning systems under development
at the Georgia Tech Research Institute.
|
Pictured left to right: John Stewart, senior research engineer,
and Georgia Tech students Michael Matthews and James Lentini,
are using an infrared computer vision system to measure core
temperature in various meat products.
|
Infrared (IR) camera technology
promises to prevent potentially harmful undercooking and minimize
overcooking – which
diminishes taste – of
ready-to-serve meat products. It is also expected to reduce energy
costs and lower yield loss in the food processing industry.
“IR camera technology is evolving,” says Craig Wyvill,
division chief of the Georgia Tech Research Institute (GTRI) Agricultural
Technology
Research Program. “Today’s camera systems are easier to
use and much more affordable than systems of just a few years back.”
Some
new IR cameras no longer require specialized cooling systems, and
the cost of ownership has dropped significantly during the past
five years. Such advances encouraged GTRI researchers to explore
the use of IR camera technology in screening and controlling thermal
operations
in food processing plants.
“We started with a study designed to use this technology to help
measure product core temperature as it comes out of the oven,” Wyvill
explains. “Now, the research has fanned out to include oven control
and help for technicians on the production line.
“The technology is fueling opportunity in cooking operations
and may make microwave precooking more practical in plants,” he
adds. “It
could eliminate the risk of undercooking while minimizing the level
of overcooking required.”
|
In the foreground, the computer screen displays an infrared
image on the left as it scans chicken nuggets on a conveyor belt.
|
An initial GTRI study funded by Georgia’s
Food Processing Advisory Council (FoodPAC) focused on using IR technology
to measure the mean
surface temperature and then estimate the core temperature of meat
products as they come out of industrial ovens. GTRI senior research
engineer John Stewart led that project, which included a field study
of GTRI’s IR computer vision system at a Gold Kist plant in Boaz,
Ala.
Stewart discovered the system could reasonably estimate core temperature
on uncoated, whole-muscle meat and products with a porous coating,
such as a light layer of breadcrumbs. But in formed and heavily breaded
items, such as chicken fingers, the mean surface temperature is not
well correlated with core temperature, Stewart says.
While imaging
some of the heavily coated products, researchers discovered that
some product had lower surface temperatures than the bulk of product
being imaged. Upon closer inspection, they discovered that most of
these cold areas were caused by ruptures in the outer casing.
“This finding and the additional observation that the variability
of the temperatures measured on the ruptures was lower than the variability
of product core temperatures measured by hand reinforced our belief
that the near-term potential of IR camera technology is in identifying
what is happening as products come out of the oven,” Wyvill explains.
These findings led to a new FoodPAC-funded project that GTRI began
in 2006. Now, Stewart and his colleagues are studying how information
from IR cameras can be used to control product temperature within
emerging microwave precooking technology, which will be used in conjunction
with conventional ovens to shorten overall cooking time.
The food processing
industry has been leery of microwave cooking because its heating
often is non-uniform, Wyvill notes. “But having IR
and visible-light cameras working together in conjunction with a microwave
can help isolate such problems and provide feedback for quick control
adjustments,” he adds.
In the system under development by Stewart
and his colleagues, visible and IR cameras work in sync to deliver
information on product temperature,
color, and size. An algorithm analyzes the data, allowing the system
to adjust microwave oven cooking on the fly – something that
is more difficult to do with conventional ovens.
“By monitoring the surface temperature of the output product,
we can very quickly adapt to variation in the product stream,” Stewart
explains. Such a system will allow processors to cook food quickly,
thus reducing energy costs, an attractive incentive for the food processing
industry.
Companies that design microwave ovens also stand to benefit
from this research at GTRI, Wyvill notes. GTRI is already working
with some manufacturers
on the issue of oven control. “Manufacturers will be able to
build microwave ovens with more consistent heating because of the ability
to get dynamic feedback that allows operators to address the problem
on the fly,” he explains.
Meanwhile, GTRI research is also investigating
the use of IR technology for conventional oven control. “We want
to create a verification step and a feedback loop to the oven to use
the IR signature to adjust
oven temperature,” Wyvill explains.
Infrared computer vision
technology can do several things for food processing plants using
conventional ovens, Stewart says. First, the
system can be easily configured to monitor trends across the width
of the cooking belt.
“Sometimes a heating element inside an oven will fail, or some
other problem will cause one side of the output stream to be significantly
hotter than the other side of the product stream,” Stewart explains. “These
imbalances are hard to detect with a traditional temperature probe,
but a temperature image of the belt can find them quite easily. By
eliminating these wide-area trends, one source of temperature variability
is removed.”
Another source of variability stems from individual
products that are colder than surrounding products on the belt, Stewart
says. “An
individual chicken breast may appear colder than other pieces on the
belt because it was colder than the rest of the product coming into
the oven, or it may have been accidentally stacked under another piece
of product during the cooking process,” he explains. “By
removing these products from the stream, another source of variability
is eliminated.”
In addition, GTRI researchers believe IR computer
vision technology can help oven technicians in food processing plants
as they randomly
pull product from conveyor belts to measure core temperature with
thermal probes. “We hope to combine ‘augmented reality’ feedback
with IR technology to help technicians see product exhibiting temperature
extremes or problems so they can focus their measurements on high-risk
product,” Wyvill says.
The augmented reality system – already
under development in GTRI and the Georgia Tech College of Computing – would
shine laser symbols on products at the temperature extremes or those
judged to
be out of range, informing technicians which products need to be manually
checked.
A lot of fully cooked food products are probably being overcooked
by 10 to 20 degrees Fahrenheit because of inadequate oven control
technology, Wyvill notes.
“Cooking meat to 180 or 190 degrees can ruin its texture, make
it loose its aroma, therefore affecting taste,” Wyvill explains. “And
it wastes energy to cook it longer and then to quickly cool it for
packaging…. Also, these products are sold by the pound, so there
is yield loss when products are overcooked and moisture, and therefore
weight, is reduced.”
The food processing industry is likely to
begin using IR computer vision technology by late 2007, Wyvill says.
Jane M. Sanders is the former editor of
Georgia Tech’s Research
Horizons Magazine.
Reprinted with
permission of the Georgia Tech Research News & Publications
Office.
Photography by Gary Meek.
|
