Page Banner

United States Department of Agriculture

Agricultural Research Service

Research Project: DEVELOPMENT OF NEW AND IMPROVED SYSTEMS TO ENHANCE FOOD SAFETY INSPECTION AND SANITATION OF FOOD PROCESSING Title: Poultry Carcass Inspection by a Fast Line-Scan Imaging System: Results from in-Plant Testing

Authors
item CHAO, KUANGLIN
item Yang, Chun-Chieh - VIS SCI UNIV OF KY
item Chen, Yud
item CHAN, DIANE
item KIM, MOON

Submitted to: Proceedings of SPIE
Publication Type: Proceedings
Publication Acceptance Date: October 4, 2006
Publication Date: November 12, 2006
Citation: Chao, K., Yang, C., Chen, Y.R., Chan, D.E., Kim, M.S. 2006. Poultry carcass inspection by a fast line-scan imaging system: results from in-plant testing. Proceedings of SPIE. 6381:Q1-Q10.

Interpretive Summary: The Agricultural Research Service of U.S. Department of Agriculture has developed two automated poultry inspection systems. The first was a visible/near-infrared reflectance spectroscopy system using a fiber optic assembly that acquired a narrow scan across the breast area of chicken carcasses. This system was tested on a 180 bpm commercial processing line and correctly identified 94% of wholesome and 92% of unwholesome birds. The second was a spectral imaging system. Most recently, the spectral imaging system was upgraded with an electron-multiplying charge-coupled-device (EMCCD) camera which, using an electron multiplying register, allows multiplication of weak signals before readout noise is added by the output amplifier. This system allows more flexibility in adjusting to light conditions, enabling accurate high-speed operation. During in-plant testing, the line-scan imaging system with an electron-multiplying charge-coupled-device (EMCCD) camera and an imaging spectrograph was used for acquisition of online images of wholesome and systemically diseased chickens in December of 2005. The complete image of chicken contained an average of 275 line-scan images for wholesome birds, and 250 line-scan images for systemically diseased birds. Each line-scan image contained 256 pixels and spectral data was collected from 55 channels for each pixel, spanning the region between 389 nm and 753 nm. Based on key wavelengths determined in a previous study, fuzzy logic membership functions were constructed and used to develop an algorithm for identifying wholesome and systemically diseased chicken images. The membership functions were developed using the validation data set, consisting of 543 wholesome and 66 systemically diseased chicken images acquired during the first four-day collection period. The method correctly classified 89.7% of wholesome chicken images and 98.5% of systemically diseased chicken images in the validation set. For the testing data set collected during the second four-day collection period, the method correctly classified 98.2% of 457 wholesome chicken images and 100% of 37 systemically diseased chicken images. In the testing set, 80 images acquired off-shift of systemically diseased chickens also were 100% correctly identified. Further testing to implement the differentiation algorithm for online operation on higher-speed commercial processing lines will be the next step for the in-plant trials.

Technical Abstract: During in-plant testing of a hyperspectral line-scan imaging system, images were acquired of wholesome and systemically diseased chickens on a commercial processing line moving at a speed 70 birds per minute. A fuzzy logic based algorithm using four key wavelengths, 468 nm, 501 nm, 582 nm, 629 nm, was developed using image data from the validation set of images of 543 wholesome and 66 systemically diseased chickens. A classification method using the fuzzy logic based algorithm was then tested on the testing set of images of 457 wholesome and 37 systemically diseased chickens, as well as 80 systemically diseased chickens that were imaged off-shift during breaks between normal processing shifts of the chicken plant. The classification method correctly identified 89.7% of wholesome chicken images and 98.5% of systemically diseased chicken images in the validation set. For the testing data set, the method correctly classified 96.7% of 457 wholesome chicken images and 100% of 37 systemically diseased chicken images. The 80 images acquired off-shift were also 100% correctly identified.

Last Modified: 9/29/2014
Footer Content Back to Top of Page