Skip to main content
ARS Home » Research » Publications at this Location » Publication #188691
Title: FISHER LINEAR DISCRIMINANT ANALYSIS FOR IMPROVING FECAL DETECTION ACCURACY WITH HYPERSPECTRAL IMAGES

Author
item Park, Bosoon
item Yoon, Seung-Chul
item Lawrence, Kurt
item Windham, William

Submitted to: Transactions of the ASABE
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 10/1/2007
Publication Date: 12/5/2007
Citation: Park, B., Yoon, S.C., Lawrence, K.C., Windham, W.R. 2007. FISHER LINEAR DISCRIMINANT ANALYSIS FOR IMPROVING FECAL DETECTION ACCURACY WITH HYPERSPECTRAL IMAGES. Transactions of the ASABE. 50(6):2275-2283.

Interpretive Summary: Fisher Linear Discriminant Analysis for Improving Fecal Detection Accuracy with Hyperspectral Images B. Park, S.C. Yoon, K.C. Lawrence, W.R. Windham Transactions of the ASAE (Peer Reviewed Journal) Interpretive Summary Recent trends in food production, processing, distribution and preparation are increasing demand for food safety research in order to ensure a safer food supply. Along with food safety issues, the Food Safety and Inspection Service (FSIS) in USDA are charged with protecting consumers by ensuring that poultry and poultry products are safe and wholesome. FSIS is pursuing a broad and long-term science-based strategy to improve the safety for protecting public health. For science-based inspection system, ARS has developed hyperspectral and multispectral imaging systems. For better system accuracy, the development of image processing software is crucial. For our research accomplishment, dynamic thresholding method to determine optimum threshold values to identify fecal contaminants on broiler carcasses was developed. This new image analysis method can improve the FSIS poultry safety inspection program by incorporating scientific testing and efficacy of fecal detection during poultry processing.

Technical Abstract: Detection of fecal contamination in the visceral cavity of broiler carcasses is important for food safety to protect consumers from food pathogens. The simple ratio of reflectance values of 565-nm image to 517-nm image was effective for fecal detection in the visceral cavity. Since the accuracy of detection algorithms for identifying cecal contaminants varied with fecal threshold values, determination of optimum threshold was crucial for improving the accuracy of fecal detection during poultry processing. The dynamic threshold method using Fisher’s linear discriminant analysis (FLDA), along with simple multispectral image ratio with Gaussian window averaging, performed better (98.9% accuracy with 1.1% omission error) than static threshold method to identify cecal contaminants. The mean and standard deviation of dynamic threshold were 1.025 and 0.027, respectively. Because of uncertainty of fecal threshold and a trade-off between missed contaminants and false positives, the dynamic thresholding method using FLDA was useful for cecal contaminant detection. Also, FLDA can be implemented to determine and update fecal threshold values for on-line inspection at poultry processing plants.