Skip to main content
ARS Home » Research » Publications at this Location » Publication #93723
Title: ON-LINE TRIALS OF A CHICKEN CARCASS INSPECTION SYSTEM USING VISIBLE/ NEAR-INFRARED REFLECTANCE

Author
item Chen, Yud
item Hruschka, William

Submitted to: Journal of Food Process Engineering
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 8/25/1999
Publication Date: N/A
Citation: N/A

Interpretive Summary: Presently, slaughter plant poultry inspection is conducted by human inspectors on-line. A machine vision inspection system on-line can help to reduce the workload of the inspectors and to improve the speed and the effectiveness of inspection. A machine vision system can accurately screen for wholesome carcasses and allow the inspectors to visually inspect only carcasses rejected by the machine vision system. The Instrumentation and Sensing Laboratory (ISL) has developed such a system. The ISL system consists of a visible/near-infrared (Vis/NIR) subsystem and a multispectral imaging subsystem. This paper reports the results of a two-week plant trail of the Vis/NIR subsystem in an actual production environment. The equipment operated well under the adverse conditions of humidity, water, and other processing substance splashing that are commonly found in poultry processing plants. The results show that a calibration model based on the spectral data of the beginning 3 days could predict the following days' chickens consistently with an accuracy of 95%, which is the same accuracy as in an off-line work previously reported. The results show that the instrument has great promise for on-line automated inspection of chicken carcasses. This information is most important to the Food Safety and Inspection Service (FSIS) who may adopt this new technology. It is also very important to the poultry industry, particularly poultry processors and equipment manufacturers. Any research engineers who are developing on-line inspection systems for agricultural products would also be interested in these findings.

Technical Abstract: On-line trials of a visible/near-infrared chicken carcass inspection system were performed during an 8-day period. Spectra (470-960 nm) of 1750 (1174 normal and 576 abnormal) chicken carcasses were measured. The spectra were obtained with a diode array instrument using fiber optic illumination and collection. The instrument measured the spectra of veterinarian-selected carcasses as they passed on a processing line at a speed of 70 birds per minute. Classification models using principal component analysis as a data pretreatment for input into neural networks were able to classify the carcasses from the spectral data with a success rate of 95%. Thus, the method shows promise for separation of abnormal from normal carcasses in a partially automated inspection system, reducing the visual inspection load. Details of the models using various training regimens are discussed.