Submitted to: Computer Standards & Interfaces
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: April 6, 2011
Publication Date: N/A
Interpretive Summary: Commercial poultry processing operations are highly dependent on automation, from the entry point of live birds hung at the start of kill lines through evisceration, further processing, and packaging for market. However, safety inspection of chickens on the evisceration lines currently is largely dependent on organoleptic inspection by USDA inspectors who work on the lines. Limited to working at speeds of 35 birds-per-minute, the inspectors examine the organs, body cavity, and exterior of each bird for indicators of disease or defects. With concerns about food safety increasing throughout the food industry, processors and regulators are in need of additional automated methods that can help to address food safety risks for rapid high-throughput processing lines while maintaining product quality and consistency and operation efficiency and also increasing production to meet growing consumer demand for poultry products. A visible/near-infrared inspection interface was developed to enable the use of a visible/near-infrared spectrometer-based safety inspection on poultry processing lines that operate at speeds as high as 140 birds-per-minute. The laboratory-developed inspection system was developed to differentiate wholesome and unwholesome birds on high-speed processing lines. Development of the interface was necessary to test and demonstrate the system on a commercial processing line, allowing individual shackles and birds to be tracked from the kill line inspection point through to the inspection point on the evisceration line. This interface serves as the basis for the spectral inspection system can be implemented for real-world operations in a commercial poultry plant, demonstrating the capacity for data transfer requirements needed to perform functions such as rejection or diversion of unwholesome birds from wholesome. Effective early (upstream) rejection of unwholesome birds can help minimize food safety problems that can result from cross-contamination of birds from processing line equipment. This research will give producers and regulators additional automation methods by which to more efficiently address and minimize food safety risks.
Technical Abstract: A new visible/near-infrared inspection system interface was developed in order to conduct research to test and implement an automated chicken inspection system for online operation on commercial chicken processing lines. The spectroscopic system demonstrated effective spectral acquisition and data processing for real-time classification of chickens on a 140 birds-per-minute processing line. Real-time online testing demonstrated that the system can successfully differentiate between wholesome and unwholesome birds, using a neural network classification model with 20 input nodes that correctly classified 94% and 92% of wholesome and unwholesome birds, respectively. This work demonstrates that the visible/near-infrared inspection interface can effectively allow the spectroscopy-based inspection system to operate on commercial chicken processing lines requiring accurate real-time product and shackle tracking.