Hyperspectral imaging is a promising technology for real-time monitoring of feed and litter quality, and mycotoxin detection

Abstract

Monitoring the quality of feed and litter moisture in the poultry industries is necessary to maintain/increase production and improve chicken health and welfare. Wet chemistry methods are time consuming and near infrared spectroscopy (NIR) depends on sampling methods (same as wet chemistry) that introduce uncertainties. Hyperspectral imaging (HSI) scans and analyses a large portion of a load, which minimises sampling error. In contrast to wet chemistry and NIR, HSI can display the variability within materials (feed and litter) and perform qualitative (identifying and sorting) and quantitative measurements simultaneously, which can be used in automatic quality monitoring systems using machine vision. In this study, we used wheat samples (ground and kernels) to investigate the possibility of using HSI combined with machine learning for quantifying carbon and nitrogen concentrations, identifying impurities and mouldy grains and quantifying deoxynivalenol (DON) infection (artificially added). We also investigated the possibility of using HSI for quantifying moisture contents in three poultry litter types (pine shavings, hardwood and re-used hardwood litter). The results showed that HSI was able to accurately quantify wheat carbon and nitrogen concentrations; identify impurities and moist/mouldy grains; identify, quantify and visualise mycotoxins in grains; and quantify/visualise moisture contents in different types of poultry litter. These capabilities, while partially dependent on future development and adoption of AgTech, could be applied to reduce the variabilities and impurities in feed ingredients, and contribute to improving the in-shed environment by accurately monitoring litter moisture.