Submitted to: Journal of Food Engineering
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: October 22, 2004
Publication Date: February 6, 2005
Citation: Kim, M.S., Lefcourt, A.M., Chen, Y.R., Tao, Y. 2005. Automated detection of fecal contamination of apples based on multispectral flourescence image fusion. Journal of Food Engineering. 71(1):85-91.
Interpretive Summary: Researchers at the Instrumentation and Sensing Laboratory (ISL), Agricultural Research Service (ARS), United States Department of Agriculture (USDA) in Beltsville, Maryland have been developing instrument-based on-line inspection system and methodologies to provide a rapid, nondestructive means to detect animal fecal contamination on fruits and vegetables. A recently developed, field portable multispectral fluorescence imaging system was used to acquire fluorescence images of apples artificially contaminated with animal feces. We developed simple fluorescence band ratios in conjunction with histogram-based automated threshold classification, and demonstrated 100% detection of animal feces contamination spots on apples. The results of this investigation are of interest to scientists developing machine vision imaging based detection methods for fecal contamination of foods, and companies interested in commercial systems to detect fecal contaminated apples.
Fluorescence techniques have shown great potential for detecting animal fecal contamination on foods. A recently developed field portable multispectral fluorescence imaging system was used to acquire fluorescence images of feces-contaminated apples. Twenty Red Delicious apples, encompassing the natural color variation of that cultivar, were each empirically contaminated on 5 spots with dairy cow feces. The contamination spots were not clearly visible to the human eye. Multispectral fluorescence images, at wavebands centered near the red emission peaks of cow feces and apples, in addition to blue and green bands, were evaluated to determine an optimal red band for detection of feces contamination spots on apples. The results show that fluorescence emission bands at 670 nm provided the greatest potential for the detection of fecal contamination on apples. In addition, a two-band ratio as a multispectral fusion method along with unsupervised histogram-based threshold classification was devised for automated detection of cow feces contamination on apples. Using the band ratio of 670 nm to 450 nm, the automated detection of fecal contamination spots on the apples achieved a 100% success rate, regardless of apple color variation.