Skip to main content
ARS Home » Research » Publications at this Location » Publication #119251
Title: LEARNING VECTOR QUANTIZATION FOR COLOR CLASSIFICATION OF DISEASED AIR SACS IN CHICKEN CARCASSES

Author
item IBARRA, JUAN - UNIVERSITY ARKANSAS
item TAO, YANG - UNIVERSITY MARYLAND
item NEWBERRY, LISA - UNIVERSITY ARKANSAS
item Chen, Yud

Submitted to: Transactions of the ASAE
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 1/14/2002
Publication Date: N/A
Citation: N/A

Interpretive Summary: Each poultry carcass sold in the US must be visually inspected for its wholesomeness by an USDA inspector. The Instrumentation and Sensing Laboratory (ISL) is developing an industrial poultry inspection system that consists of a visible/near-infrared (Vis/NIR) subsystem and an imaging subsystem that can separate wholesome and unwholesome carcasses on-line. In ncollaboration with engineers at the University of Arkansas, Fayetteville, Arkansas, a color imaging technique was developed for detection of air-sacculitis lesions in chicken carcasses induced by secondary infection with Escherichia coli (E. coli). The method used intensity normalized color components with proper transformation and a modified competitive neural network to achieve a 96.7% classification of infected and normal tissues. This information is very important the FSIS who may want to adopt this new technology. It is also very important to the poultry industry, particularly poultry processors and equipment manufacturers. The research engineers who are interested in developing on-line inspection systems for agricultural products would also be interested in these findings.

Technical Abstract: The variation in color features observed during the evolution of air sacculitis in chicken carcasses is exploited to classify the disease using digital imaging and neural networks. For the experiments, air sacculitis was induced by secondary infection of E. coli via direct inoculation of challenge bacteria. Mild and severe infection samples were obtained and imaged. For the supervised classification, a knowledge base set of normalized RGB values, corresponding to negative, mild, and severely infected air sac images, was obtained. Statistical data exploration indicated no significant difference between the color features of mild, and severely infected sacs, but a significant difference was found between infected and negative tissues. A neural network classified the data in infected and negative categories, using the Learning Vector Quantization algorithm. Resubstitution and hold-out errors were calculated, giving an overall accuracy in the classification of 96.7%. Each poultry carcass sold in the US must be visually inspected for its wholesomeness by an USDA inspector, with air sacculitis being the major cause of condemnation in poultry processing plants. The method presented here has the potential for integration in a computer-assisted inspection of wholesomeness in poultry processing lines.