Hyperspectral Imaging to Catch Poultry Production Line Pollution

November 26, 2018

The November 2018 issue of Poultry Digital magazine features a "Foreign Intelligence" article (pages 10-13) on how imaging technology can reduce or eliminate poultry production line pollution. Headwall hyperspectral cameras can detect the spectral "fingerprint" of contamination such as fecal matter or the hallmarks of diseases such as septicemia and toxemia.

Poultry processing involves acquiring, analyzing, and exploiting large amounts of data very quickly, instantaneously if possible. Up to 400 line-scan images are taken per carcass. A Headwall Hyperspec® sensor and software algorithm accurately measures reflectance of 55 unique spectral bands (wavelengths).  The relative reflectance measurements are layered into a Headwall algorithm with the output being a determination of wholesome (healthy) versus diseased (unhealthy) carcasses.

In terms of disease conditions, Headwall embeds algorithms for 1) diseases such as septicemia and toxemia and also for 2) various different states of quality such as woody breast syndrome or fecal contamination.  The goal is to leverage the sensor platform by layering as many different algorithms onto the sensor to establish the necessary data analytics to drive food quality and supply chain decisions throughout the process line.

The same reflectance measurements are used as input for other algorithms which drive safety and quality decisions within the factory. In this way, poultry processors are able to leverage the investment in these inline sensors across a much broader set of applications/conditions to optimize their return on investment.

Headwall has worked very closely and partnered with the USDA Agricultural Research Service to develop and implement this method on a high speed processing line, and a number of Headwall customers in Canada are working with the Canada Poultry Association on other applications.