Submitted to: Transactions of the ASAE
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 3/15/2001
Publication Date: 10/25/2001
Interpretive Summary: Government agencies and industries have committed cooperative efforts to improve the overall quality and safety of foods produced in the United States. Food products are required to be inspected to assure high quality and for safe consumption. Rapid and noninvasive methods that can be implemented for the assessment of hazardous food conditions in the steps of food production would be substantially beneficial. Spectroscopic imaging techniques may allow on-line measurements with such specificity. On-line imaging applications may only require several essential spectral bands (multispectral) to extract unwholesome conditions present in food commodities. However, the optimal spectral bands may depend on the food commodity of interest and a number of factors, such as types of contaminations and the state of the food commodity. In order to refine optimal spectral bands and to further develop spectral algorithms for on-line implementations, more research is needed when considering the myriad of food commodities available. Thus, an imaging system in conjunction with spectroscopic capabilities (hyperspectral) is desirable. We developed a laboratory-based hyperspectral imaging system. This system is capable of both reflectance and fluorescence measurements in the visible to near-infrared regions of the spectrum. This information will be of great interest to research scientists/engineers who are applying imaging techniques for food safety inspection and quality grading.
Technical Abstract: This paper presents a laboratory-based hyperspectral imaging system designed and developed by the Instrumentation and Sensing Laboratory, United States Department of Agriculture, Beltsville, Maryland. The spectral range is from 430 to 930 nm with spectral resolution of approximately 10 nm (full width at half maximum) and spatial resolution better than 1 mm. Our system is capable of reflectance and fluorescence measurements with the use of dual illumination sources where fluorescence emissions are measured with ultraviolet (UV-A) excitation. Calibrations and image correction procedures for the system artifacts and heterogenous responses caused by the optics, sensor and lighting conditions throughout the spectrum region for reflectance and fluorescence are presented. The results of the fluorescence correction method showed that the system responses throughout the spectrum region were normalized to within 0.5% error. The versatility of the hyperspectral imaging was demonstrated with sample fluorescence and reflectance images of a normal apple and one with fungal contaminated and bruised spots. The primary use of the imaging system in our laboratory is to conduct food safety and quality research. However, we envision that this unique system can be used in a number of scientific applications.