Assessing the quality level of corn tortillas with inductive characterization and digital image analysis

Abstract

Characterization and classification of corn tortillas turns out to be an extremely delicate and difficult process when dealing with regulations for import/export and production process certification. In this paper we present a method for non-invasive feature extraction, based on digital imaging and a series of procedures to characterize different qualities of corn tortillas for their later classification. The novelty in this whole method lies in the extremely reduced set of features required for the characterization with only geometrical and color features. Nonetheless, this set of features can assess diverse quality elements like the homogeneity of the baking process and others alike. Experimental results on a sample batch of 600 tortillas show the presented method to be around 95% effective. © 2013 Springer-Verlag Berlin Heidelberg.