Page Banner

United States Department of Agriculture

Agricultural Research Service

Title: Automated poultry carcass inspection by hyperspectral-multispectral line-scan imaging system

item Chao, Kuanglin - Kevin Chao

Submitted to: Complete Book
Publication Type: Book / Chapter
Publication Acceptance Date: 1/22/2010
Publication Date: 5/15/2010
Citation: Chao, K. 2010. Hyperspectral Imaging for Food Quality Analysis and Control. London: Academic Press, Elsevier. p. 241-272.

Interpretive Summary: Due to increasing production needs and food safety concerns facing the poultry industry in the U.S. and worldwide, automated systems developed for safety inspection of poultry products on high-speed processing lines will be essential. By enabling poultry producers and regulatory agencies to satisfy high-throughput production and inspection requirements more efficiently, science-based automated food inspection systems can help alleviate the pressures on human inspectors, improve production throughput, and grow public confidence in the safety and quality of the food production and distribution system. Development of automated non-destructive food safety inspection methods based on spectroscopy and spectral imaging has been a major ARS research priority over the last decade. This book chapter describes the development of automated chicken inspection techniques by ARS researchers that has led to the latest hyperspectral line-scan imaging system for wholesomeness inspection on commercial high-speed processing lines, now under commercial development for industry use. The current USDA poultry inspection program and the progression of ARS Vis/NIR spectroscopy methods, target-imaging, and hyperspectral/multispectral line-scan imaging for poultry inspection are discussed. The chapter provides engineers and technologists working in research, development, and operations in the food industry with critical and readily accessible information relevant to development of hyperspectral imaging technology for a high-speed commercial application.

Technical Abstract: Visible/near-infrared spectroscopy methods first developed by ARS scientists were demonstrated capable of over 90 percent accuracies on high-speed processing lines in differentiating wholesome chickens from unwholesome birds exhibiting systemic conditions; however, given the lack of spatial information, the field of application for Vis/NIR spectroscopy inspection systems was considered limited. For practical expansion of the spectral techniques to multispectral imaging required the investigation and development of wavelength selection methods in order to reduce the data volumes of full spectral and spatial data acquired for whole bird carcasses. Hyperspectral imaging was first used for spectral analysis to select wavelengths for implementation in automated multispectral imaging systems, and in itself was effective for laboratory-based research. The introduction of EMCCD cameras and their use with imaging spectrographs enabled automated line-scan spectral imaging for high speed processing, and was particularly important for allowing a single imaging system to perform both hyperspectral and multispectral imaging. The transition from target-based imaging to line-scan imaging required the development of algorithms such as line-scan target detection for effective online implementation. Such algorithms streamlined or simplified processing-line imaging operations, for example by eliminating sensors formerly needed to trigger accurate imaging of individual birds on the line, and also provided potential value-added applications that could be performed using the same image data, such as assessing product defects, size, shape, or weight attributes. In-plant testing demonstrated that the ARS line-scan spectral imaging system could successfully inspect chickens on high-speed processing lines. The system can be used for online pre-sorting of birds on commercial poultry processing lines, thereby increasing efficiency, reducing labor and costs, and producing significant benefits for poultry producers and processors.

Last Modified: 10/20/2017
Footer Content Back to Top of Page