|Delwiche, Stephen - Steve|
Submitted to: Proceedings of the American Society of Agricultural and Biological Engineers International (ASABE)
Publication Type: Proceedings
Publication Acceptance Date: 6/27/2008
Publication Date: 7/29/2008
Citation: Yang, I., Delwiche, S.R., Chen, S., Lo, Y.M. Enhancement of fusarium head blight detection in single free-falling wheat kernel using a multi-spectral inspecton system. Proceedings of the American Society of Agricultural and Biological Engineers International (ASABE). June 29-July 2, 2008.
Interpretive Summary: One of the most prevalent fungal diseases that affect small cereal grain crops during development is Fusarium Head Blight (FHB). In addition to yield reduction, this disease is problematic because it produces a mycotoxin, deoxynivalenol or DON, which is harmful to humans and livestock. Short of eradicating the disease (unlikely in the near term) or developing resistant varieties (an ongoing multi-institutional effort), grain handlers and mills are forced to deal with this problem by a series of costly options: discarding of contaminated lots, blending of contaminated grain with clean grain to reduce the mycotoxin concentration, or physical removal of mold-damaged kernels. The present research addresses the last aforementioned option and attempts to do this by means of high-speed optical inspection, one kernel at a time. A unique form of lighting was used in a two-color, or bichromatic design. Red and green light were flashed in sequence, and reflected energy readings from the kernels in free-fall were captured at high frequency (1,000-10,000 Hz range). The relationship between the red and green reflected light, which is indicative of the kernel’s color, size, and texture, was used to categorize kernels (sound vs diseased) allowing for acceptance or rejection of individual kernels in a high-speed optical sorter. Improvements in the system have resulted in increased accuracy in detection of diseased kernels from approximately 80 percent (as previously reported) to greater than 90 percent. These findings can lead to improvements of high-speed optical commercial sorters, such that mills and grain terminals will have a better means to clean up mold-contaminated grain lots.
Technical Abstract: Since the 1800s, Fusarium Head Blight has deleteriously affected the yield and quality of small grain cereal crops such as wheat. This fungal disease is also a health concern due to the frequent production of the secondary metabolite, deoxynivalenol (DON), which is moderately toxic to humans and non-ruminant animals. Our study is a progress report on the effort to develop more efficient methods for separating Fusarium-damaged kernels from sound wheat kernels. Through the development of a high-powered pulsed LED system, we have demonstrated that Fusarium-damaged and sound individual wheat kernels can be correctly categorized at up to 91 percent average accuracy. The system is bichromatic in the sense that green and red LEDs are pulsed in sequence, similar to our recent work in the past year. In this study, however, the system illumination, signal acquisition, and analysis modules were integrated and improved. Two parameters (slope and r-squared) from a regression analysis of the green response onto the red response were used as input parameters in a linear discriminant analysis (LDA). In agreement with our previous work, slope was the predominant classifier; so much so that it was used exclusively after preliminary analyses. The factors that affect the level of accuracy are the orientation of the optical probe with respect to the LED illumination source that strikes the free-falling kernel, and the color contrast of the two categories, which is observed to vary from sample to sample. Our previous research has shown that commercial high-speed optical sorters are, on average, 50 percent efficient at removing mold-damaged kernels; however, under more carefully controlled, kernel-at-rest, conditions in the laboratory, this efficiency can rise to 95 percent or better. Our present research on free-falling kernels is producing accuracies that are beginning to approach those of the controlled conditions. Knowledge gained from this research will provide design criteria for improvement of high-speed optical sorters for reduction of DON in raw cereal commodities, as well as in finished food products.