USDA Uses Remote Sensing and ENVI to Detect Food Contaminants
Anonym
Customer Challenge
Food contaminants are a problem for both food distributors and consumers. Poultry can be particularly dangerous, as contaminants can cause serious health issues. The USDA needed a way, using remotely sensed imagery, to detect contaminated poultry before it ever leaves the production line.
Solution Achieved
When most people think of remote sensing, spacecraft and orbiting satellites might be what come to mind. But researchers at the USDA Agricultural Research Service have taken the potential of remotely sensed imagery a step further, and into an application that directly affects consumer health and safety. The USDA’s Poultry Processing Research Unit, uses remotely sensed imagery and ENVI® to inspect and assess the quality and safety of poultry products at the processing level, an application that has the potential not only to save the processing plants money, but to increase the quality and safety of the food we buy.
The need for testing system grew out of the implementation of a zero tolerance policy for poultry contaminants, mandated by the federal government. Each poultry processing plant has to achieve that goal by law, and the USDA developed the imaging system as one method by which that goal can be met. Currently, processing plants have no choice but to use massive amounts of water to meet the mandate. Once adopted, testing with high tech cameras and an easy-to-use online system based on ENVI will be an economical and environmentally friendly alternative to excessive water usage.
The USDA’s Poultry Processing Research Unit began the food testing initiative as a collaborative effort with the Institute for Technology Development at Stennis Space Center, which built the hyperspectral imaging instruments used in the development of the detection process. The group used the hyperspectral images and processed them with ENVI to determine a subset of three wavelengths that were optimal for contaminant detection. They then used common aperture cameras that captured those three bands simultaneously, and ultimately perfected a system that includes a portable imaging instrument mounted inside of a food-grade stainless steel enclosure,
Project lead Bob Windham, Jerry Heitschmidt, and Agricultural Engineers Kurt Lawrence and Bosoon Park used ENVI and its underlying programming language, IDL®, to test the viability of the system, develop algorithms, to process the imagery, and to verify final results. Self-taught ENVI and IDL users, the group , along with four other ENVI users in the facility, developed the process primarily using ENVI’s Decision Tree method of classifying images. ENVI and IDL’s batch processing capabilities were essential for perfecting a process where the operator only has ¼ second to take an image, process it, and make a decision about its cleanliness. “We always start with ENVI and build from there,” said Heitschmidt. “If ENVI doesn’t have the functionality we need, we use IDL to develop it, since IDL is a lot more intuitive than other languages.”
Now, they are able to not only detect contaminants as mandated by federal law, but are using ENVI to detect foreign objects in poultry fillets, such as bones which is also of concern to the industry, government, and consumers. “ENVI is the workhorse image processing package that we use for all of these [food quality] applications," said Heitschmidt. "When you look at the functionality, especially with respect to hyperspectral imaging, and user friendliness, and price, it is clearly head and shoulders above the others.”
Although the system has not yet been used in a commercial plant, the USDA believes it will be adopted by commercial food distributors in the near future. In addition, they will be continuing work on eggs and fertility prediction. Said Heitschmidt, “Of course there is traditional remote sensing, from airborne and space borne platforms, and you can do land cover classifications and other processes, but that is, I think, a myopic view of what remote sensing and ENVI can be used for.”
Benefits
- With the help of ENVI, the USDA has produced a system that increases quality and safety in a consumer product
- Since ENVI was easily customized to their needs, the team was able to develop and implement the proper algorithms for a successful system
- ENVI provides more of the functionality needed in a single package, which saved them money and resources.