Accurate prediction of nutritional value of sorghum grain using image analysis

Abstract

ABSTRACT1. This study evaluated the application of L (lightness)*a (redness) and *b (blueness) colour analysis and chemical compositions to predict the nutritional value of sorghum grain.2. A total of 12 varieties of sorghum grain were analysed for L*a*b colours, chemical composition, energy and total and digestible amino acid content. Regression models based on the linear, non-linear and the interaction effects of inputs were applied to predict the nutritional value of sorghum grains either using L*a*b colour or chemical composition, as the model inputs.3. The results illustrated a significant relationship between a*b and/or chemical compositions with energy content in the samples of sorghum grain. The provided estimation equations presented high goodness of fit in terms of R2adj ranging from 0.744 to 0.999.4. Total and digestible amino acids of sorghum grain were estimated based on a*b and chemical compositions data with the goodness of fit ranging from 0.641 to 0.999 (R2adj).5. In conclusion, the L*a*b colour analysis may be used for developing equations to predict nutritional value of sorghum grain as an alternative approach to the conventional time-consuming and costly chemical and bioassay methods.