Imaging system for fecal and ingesta detection on poultry carcasses

A hyperspectral imaging system was developed to detect surface contaminants on poultry carcasses. The system consists of a transportable stand, two quartz halogen line lights, a prism-grating-prism imaging spectrograph connected to a 1280x1024 pixel silicon CCD camera, and a computer with a frame-grabber card. The imaging system is capable of collecting spectral reflectance information from 430 to 900 nm with 1-nm resolution for every pixel of a carcass image. The resulting three-dimensional image cube is typically reduced to 320x340 pixels of spatial information with 512 pixels of spectral information for each spatial pixel. This paper reports the results of this system for the detection of ingesta from the crop or gizzard and feces from the duodenum, ceca, and colon on the surface of a poultry carcass. Sixteen six-week old male birds on a corn/soybean diet were meal-fed, subjected to an 8-hour feed withdrawal, cooped, slaughtered, hard scalded (57.5 C for 2 min.), picked, eviscerated, and feces and ingesta was collected. Uncontaminated carcasses were then immediately imaged. Next, feces and ingesta were applied to varying locations on the breast, wings, and legs of the carcass, typically three spots per contaminant, 12 spots per bird, for a total of 190 contaminant spots, and second images were taken. Results are presented for several data analysis techniques including principal component analysis, which includes all measured wavelengths, and the wavelength ratio of the 565-nm image divided by the 517-nm image. The wavelength-ratio images were further processed with background masking, thresholding, and histogram stretching. The threshold and histogram stretching values were fixed for all birds. Linear and square root histogram stretches were performed. Results indicated that 97.3 and 100 percent of the contaminates were detected with the linear and square-root histogram stretches, respectively. The research shows the feasibility of a real-time system for fecal and ingesta detection from two wavelength images at typical poultry processing line speeds.