Poultry carcass scanner passes first test
October 09, 2009
by Bryan Salvage
WASHINGTON — Agricultural Research Service (A.R.S.) scientists have developed technology that automatically scans poultry carcasses for contamination and it has been successfully tested in a commercial poultry plant. The system was developed by A.R.S. scientists Kurt Lawrence, Bosoon Park, Bob Windham and Seung-Chul Yoon, who work for the agency's Quality and Safety Assessment Research Unit in Athens, Ga.
Researchers have improved the hyperspectral imaging system so it can detect small amounts of fecal contamination. Creating individual wavelengths of light that pinpoint contaminants, the hyperspectral imaging technique combines digital imaging with spectroscopy.
In order to commercially test the technology, a prototype was placed in a poultry plant to detect contaminated carcasses. The system was developed through a research agreement with Stork Food Systems in Gainesville, Ga. According to Ms. Durham, carcasses were imaged after evisceration and before washing at a rate of 150 birds per minute. The system ran for several days without hardware or software problems and demonstrated its feasibility, according to one scientist.
The scientists are collaborating with agricultural engineer Kevin Chao and biophysical scientist Moon Kim at the A.R.S. Environmental Microbial and Food Safety Laboratory in Beltsville, Md. The Beltsville scientists developed an on-line imaging system to differentiate systemically diseased poultry carcasses from wholesome ones using the same hyperspectral imaging technology but at different wavelengths.
The fecal-detection and diseased-carcass-detection systems are being merged onto a common platform by the researchers and their industry partner that includes a line-scan hyperspectral imaging camera, lighting and operating and detection software. Merging both systems will aid in commercialization of the technology by creating one interchangeable imaging system that can be installed in different locations of the processing line to solve two separate and significant processing problems, Ms. Durham writes.
Combining both systems will also allow processors to more easily integrate such a system into their operations, the researchers added. The team plans to have a new prototype tested by the end of 2009. Researchers in Athens have also developed and implemented a new image-processing method to identify and remove false-positive readings.