|Chao, Kuanglin - Kevin Chao|
Submitted to: Applied Engineering in Agriculture
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 3/9/2005
Publication Date: 4/1/2005
Citation: Chao, K., Chen, Y.R., Ding, F., Chan, D. E. 2005. Characterizing wholesome and unwholesome chickens by CIELUV color diffference. Applied Engineering in Agriculture. 21(4):653-659.
Interpretive Summary: New inspection technologies are needed that can allow poultry plants to meet government food safety regulations and also increase competitiveness and profitability while meeting rising consumer demand. Consequently, the development of accurate, rapid, and non-invasive technologies appropriate for operation on high-speed processing lines is of great importance for the poultry industry. The overall objective of this research was to investigate a quantitative, color-based method suitable for rapid automated online sorting of wholesome and unwholesome chickens. Specific objectives were to characterize wholesome and unwholesome chicken color in CIE color coordinates, and to calculate a simple numerical color difference for the classification of chicken samples as wholesome and unwholesome. Spectral measurements were acquired for fresh chickens on a high-speed kill line. Significant CIELUV color difference values, calculated at each wavelength between the mean wholesome and unwholesome spectra, occurred near wavebands that have been associated with deoxymyoglobin, oxymyoglobin, and metmyoglobin in previous studies. Among all pairwise combinations of 10-nm wavebands between 400 nm and 700 nm, the greatest color difference values occurred for (508 nm, 426 nm), (560 nm, 426 nm), (640 nm, 420 nm). Classification by color difference was calculated as a simple formula in CIELUV color space. Full-spectrum classification resulted in classification accuracies of 85%, 86%, 84%, and 82% for wholesome validation samples, wholesome testing samples, unwholesome validation samples, and unwholesome testing samples, respectively. The (560 nm, 426 nm) waveband combination showed the best classification, with accuracies of 91%, 92%, 90%, and 90% for wholesome validation samples, wholesome testing samples, unwholesome validation samples, and unwholesome testing samples, respectively. Since the numerical color difference is a very simple distance calculation and relatively highly classification accuracies can be achieved without full-spectrum data, this method shows promise for rapid automated online sorting of chicken carcasses. This information is useful to the Food Safety and Inspection Service (FSIS), and poultry equipment and processing plants.
Technical Abstract: Development of an automated inspection system will help the poultry processing industry to provide better chicken products for the consumer while minimizing potential economic losses. The objective of this research was to investigate the potential of a color-based sensing technique suitable for rapid automated inspection for wholesomeness of chickens in the visible region between 400 nm and 700 nm. Spectra of veterinarian-selected carcasses, 400 wholesome and 332 unwholesome, were collected from a high-speed processing plant kill line using a visible/near-infrared spectrophotometer system. CIELUV color differences characterizing wholesome and unwholesome chicken samples were calculated as a simple distance formula and used to classify individual samples. Results showed that the greatest color differences occurred for waveband combinations at (508 nm, 426 nm), (560 nm, 426 nm), and (640 nm, 420 nm). Full-spectrum classification achieved accuracies of 85%, 86%, 84%, and 82% for wholesome validation samples, wholesome testing samples, unwholesome validation samples, and unwholesome testing samples, respectively. Using the (560 nm, 426 nm) waveband combination, classification accuracies of 91%, 92%, 90%, and 90% were achieved for wholesome validation samples, wholesome testing samples, unwholesome validation samples, and unwholesome testing samples, respectively. The potential of using CIELUV color differences to differentiate between wholesome and unwholesome chickens was demonstrated, and the straightforward calculation involved suggest that the method is suitable for rapid automated online sorting of chicken carcasses.