A structured-illumination reflectance imaging dataset for woody breast assessment of broiler meat.

Journal: Data in brief
Published Date:

Abstract

Wood breast (WB) myopathy is an economically important muscular defect that downgrades poultry meat quality and currently requires manual assessment for identifying and removing affected products at processing lines. An image dataset was created to assess WB conditions in broiler breast fillets using structured-illumination reflectance imaging (SIRI) as a non-destructive, objective means for WB assessment. A custom-assembled SIRI platform was used for sample imaging, and it mainly consisted of a broadband quartz tungsten halogen light source, a digital micro-mirror-device-based projector that shined phase-shifted sinusoidal patterns of light over samples, a monochromatic camera with a resolution of 2048 × 2048 pixels, and a computer, operating in an enclosed chamber. A total of 168 broiler breast fillets were collected from a commercial poultry processing plant and categorized by trained personnel into 72 "Normal" (WB-free) and 96 "Defective" (WB-affected) fillets based on tactile palpation and visual inspection. Sinusoidal illumination patterns at eight different spatial frequencies (0.015-0.150 cycles/mm) were sequentially projected onto the samples, and the reflectance pattern images were captured under the sinusoidal illumination of three phase-shifted patterns at each spatial frequency, yielding a set of 24 images acquired per sample. Hence the dataset consists of a total of 4032 raw pattern images, each of which is of size 2048 × 2048 pixels and saved as a 16-bit grayscale image in .tif format. Through demodulation, direct component (DC), amplitude component (AC), and phase difference images can be readily obtained from the three phase-shifted raw pattern images at each spatial frequency, and these images, especially the phase difference image that depicts the surface geometry, are useful for WB assessment and sample classification. In addition to the raw pattern images, the demodulated image (DC, AC, and phase difference) data is also included in the dataset. This dataset represents the first publicly available SIRI dataset and is expected to be a valuable resource for advancing SIRI for poultry quality assessment and beyond.

Authors

  • Yuzhen Lu
    Department of Agricultural and Biological Engineering, Mississippi State University, Starkville, MS, United States.
  • Hamed Sardari
    Department of Biosystems and Agricultural Engineering, Michigan State University, East Lansing, MI 48824, USA.

Keywords

No keywords available for this article.