Multi-class colour inspection of baked foods featuring support vector machine and Wilk's λ analysis

An automated, intelligent system for the colour inspection of biscuit products is proposed. In this system, advanced classification techniques featuring Support Vector Machines (SVM) and Wilk's lambda analysis were used to classify biscuits into one of eight distinct groups, corresponding to different degrees of baking. The results of the analyses were compared using standard discriminant analysis employing direct and multi-step classifications. It was discovered that the directed acyclic graph (DAG) and the balanced binary tree (BBT) after Wilk's lambda were more precise in the classification, as compared to other classifiers. In all cases, these methods resulted in the correct classification rate of 87.25% and 86.75% for DAG and BBT, respectively. Since the algorithm was implemented using software, the system could be programmed to inspect other bakery products. (c) 2010 Elsevier Ltd. All rights reserved.