Non-destructive determination of crude fiber in food and agricultural by-products

The projected increases in human populations and increasing and sometimes conflicting demand on land and water resources necessitates the livestock industry to increase its productivity using existing or fewer resources. A resilient livestock industry will require cost-effective feed rations as feed accounts for between 60 to 80% of production costs. Recently, there has been exploration of food and agriculture by-products in cost-effective livestock feed production. However, there is a huge variation in the nutritional content of these by-products including crude fiber which is a vital component in feed formulation. This necessitates regular analyses of their nutritional content. The current available methods are destructive, requires sample preparation and are time-consuming. Consequently, a hyperspectral imaging system was used to acquire spectral data. Second derivatives and normalization were used as data preprocessing methods. A prediction model was developed using partial least square regression. Results showed a correlation between measured and predicted values with an r-square of 0.76 for calibration and 0.70 for prediction. A model for estimation of crude fiber in mixed byproducts that is rapid and non-destructive was developed using hyperspectral imaging.