Smart machines and machines that can "see" continue
to make inroads into the agricultural industry, and research here
marks OSU as a leader rather than a follower in these fields.
Scientists in the Department of Biosystems and Agricultural Engineering
are using machine vision as part of a "smart sprayer"
that can identify weeds and spot spray them rather than spray
large blocks of a field. The same technology is being used to
develop an applicator that detects when a crop is in need of nitrogen
and apply it as needed rather than blanketing the whole field.
Neural networks have been developed which can identify the
root collar location in digital images of pine seedlings and has
led to a better recognition system to provide inspection on seedlings
as they travel down a moving belt for packaging. Now machine vision
is being investigated as a possible way to improve on meat inspection
methods.
Cooperating with members of OSU's Department of Animal Science
and the Food and Agricultural Products Center, the agricultural
engineers connected a color video camera to an image processing
microcomputer to acquire and analyze images of beef steaks. They
developed image processing software to measure ribeye marbling
and color, and to assign quality grade according to applicable
USDA standards. The imaged steaks also were evaluated by expert
beef graders for comparison.
This project demonstrated the potential of video image analysis
(VIA) technology for accurate, objective, and rapid evaluation
of quality characteristics in beef cuts. Ribeye color was measured
and scored according to a recognized beef color guide. Marbling
was measured and calculated as a percentage of ribeye area and
assigned a numerical value compatible with USDA marbling score
cards.
Results suggest it would be beneficial to augment current on-line grading by providing USDA graders with immediate, objective values for marbling and color. VIA implementation could significantly enhance accuracy and consistency of quality grade assignments. Extended further, the researchers believe the results target eventual replacement of current graders. Costs could be significantly reduced, since USDA grading service charges now exceed $1,000 per day in major packing plants.