Fast Line-Scan Imaging System For Broiler Carcass Inspection

Authors: Chao, Kuanglin - Kevin Chao; YANG, CHUN-CHIEH - UNIV OF KY, VIS SCI; Chen, Yud; Kim, Moon; Chan, Diane

Submitted to: Sensing and Instrumentation for Food Quality and Safety
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 2/12/2007
Publication Date: 4/20/2007
Citation: Chao, K., Yang, C., Chen, Y.R., Kim, M.S., Chan, D.E. 2007. Fast Line-Scan Imaging System For Broiler Carcass Inspection. Jrnl of Sensing and Instrumentation for Food Quality and Safety. 1:62-71.

Interpretive Summary: The Agricultural Research Service of U.S. Department of Agriculture has developed a fast line-scan imaging system for on-line poultry carcass inspection. The fast line-scan imaging system was used to acquire images of 250 fresh chickens in two sets: 70 wholesome and 76 systemically diseased chickens for the first image set, and 60 wholesome and 44 systemically diseased for the second image set. The chicken carcasses were hung on a line of shackles moving at a speed of 70 birds per minute for imaging. A method to locate the ROI within each line-scan image was developed and through spectral analysis of the 103 available wavelengths between 395 nm and 1138 nm, four wavelengths at 424 nm, 465 nm, 515 nm, and 546 nm were selected as key wavelengths for differentiation. The wavelength of 689 nm was selected as a reference wavelength for calculating band ratios with the key wavelengths. A fuzzy logic based algorithm was developed to differentiate between images of wholesome and systemically diseased chickens. For each scanned line, image features calculated for single pixels were used as inputs to the fuzzy logic algorithm to obtain a discrete decision output, indicating the existence of systemic disease. Two sets of membership functions were evaluated, one using the four key wavelengths and the second using band ratios of the four key wavelengths. For model development using the key wavelengths classification accuracies for wholesome and systemically diseased chicken images were 98% and 95%, respectively, and for model testing were 98% and 93%, respectively. For model development using the wavelength ratios, classification accuracies for wholesome and systemically diseased chicken images were 94% and 95%, respectively, and for model testing were 95% and 95%, respectively. The use of ratios decreased variation within each category of chicken condition but did not improve the classification accuracies significantly, as the accuracies from using key wavelengths were already high. Using the four key wavelengths, the fast line-scan imaging system is suitable for online multispectral differentiation of wholesome and systemically diseased chickens on high-speed processing lines. This information is useful to the Food Safety Inspection Service (FSIS) and poultry processing plants.

Technical Abstract: The USDA Agricultural Research Service has developed a fast line-scan imaging system for differentiating wholesome and systemically diseased fresh chickens. The imaging system was used to acquire hyperspectral line-scan images of 250 chicken carcasses on a laboratory processing line moving at 70 birds per minute. A method appropriate for line-scan imaging was developed for automated sensing of birds and locating the Region of Interest (ROI) within the line-scan images most suited for differentiation. From analysis of wholesome and systemically diseased chicken spectra in the ROI, four key wavelengths for differentiating between wholesome and systemically diseased chickens were selected: 424 nm, 465 nm, 515 nm, and 546 nm. The key wavelengths and their ratios with a reference wavelength (689 nm) were investigated for a fuzzy logic based differentiation algorithm. Classification using the key wavelengths correctly identified 98% and 95% of wholesome and systemically diseased chickens for model development, and 98% and 93% of wholesome and systemically diseased chickens for model testing. Although band ratios reduced variation within each chicken category, the resulting classification accuracies were not significantly improved over those for classification by key wavelengths.