Submitted to: ASAE Annual International Meeting
Publication Type: Proceedings
Publication Acceptance Date: 8/1/2004
Publication Date: 8/1/2004
Citation: Yang, C., Chao, K., Chen, Y.R., Kim, M.S. 2004. Application of multispectral imaging for identification of systemically diseased chicken.(Abstract). ASAE Annual International Meeting. ASAE Paper No. 04-3034. Interpretive Summary: The Poultry Products Inspection Act (PPIA) requires Food Safety and Inspection Service (FSIS) inspectors of the United States Department of Agriculture (USDA) to conduct post-mortem inspection for wholesomeness of all chickens intended for sale to U.S. consumers. FSIS has completed a 3-year transformation of its traditional inspection system to a Hazard-Analysis-and-Critical-Control-Point (HACCP) inspection system. Under HACCP, increasing consumer demand and line speeds will continue to increase the need for and pressure on inspectors. Thus, to address food safety concerns and meet growing consumer demand, there is an urgent need to develop automated inspection systems that can operate on-line in real-time in the slaughter plant environment. In this research, image processing algorithms were developed to differentiate septicemic chickens from wholesome ones. Based on spectral measurements, the CIELAB values showed that septicemic chickens had lower intensity values than wholesome chickens. Also based on spectral measurements, the wavelengths of 580 nm, 540 nm, and 488 nm were selected for the wave band filters because the significant difference between wholesome and septicemic chickens could be found in these wavelengths. Another filter was selected in the wavelength of 610 nm for masking in image processing. These four filters were used for multispectral imaging analysis. From multispectral images, the contour maps of wholesome and septicemic chickens showed that the intensity from the abdomen part of a carcass between wholesome and septicemic chickens was more different than from other parts of a carcass. It is necessary to focus on this region instead of the whole carcass for image classification. An image processing algorithm was developed to define this part as the region of interest. Using the average intensity in 580 nm wavelength from the region of interest, a single threshold was generated by the decision tree method and tested by independent data from different collection time frame. The results showed with sufficient training data, the threshold could successfully differentiate 98.6% of septicemic chickens and 96.3% wholesome chickens from each other. It indicated that, comparing to other wavelengths, the wavelength of 580 nm is essential and unique for differentiation between wholesome and septicemic chickens. This information is useful to the Food Safety and Inspection Service (FSIS), and poultry processing plants.
Technical Abstract: A multispectral imaging system for automated inspection of wholesome and systemically diseased chickens was developed and demonstrated. The disease of septicemia was selected as the detection target because it is the most common chicken disease in the United States.From visible/near-infrared reflectance spectra of poultry carcasses, average CIELAB L*(lightness), a*(redness), and b*(yellowness) values were analyzed. The difference of lightness between wholesome and septicemic chickens was significant. The multispectral imaging system consisted of a back-illuminated CCD camera and a spectrometer with four narrow-band interference filters for 488, 540, 580, and 610 nm wavelengths, respectively. The 16-bit multispectral images of chicken carcasses were collected for image processing and analysis. Image processing algorithms, including image registration, flat-field correction, image segmentation, region of interest identification, feature measurement, and symptom recognition, were developed to differentiate septicemic chickens from wholesome ones. The image from 610-nm wavelength was used to create a mask to extract chicken images from background. The average reflectance intensities at 488, 540, 580, and 610 nm from different parts of the carcass in the front side were calculated. Four normalization methods and four normalized differentiation methods between two wavelengths were also calculated for comparison. Decision tree was applied to generate thresholds for differentiation between wholesome and septicemic chickens. Images were collected at three time frames. The images from the first time frame were used to generate first thresholds that were tested by the images from the second time frames. Then, the images from the first and second time frames were used together to generate second thresholds. The first and second thresholds were tested by the images from the third time frame, respectively. The results showed that using average intensity at 580 nm from the region of interest, 98.6% of septicemic chickens and 96.3% of wholesome chickens could be differentiated from each other. More training data could help to generate more appropriate thresholds used at different time frames.