Automatic Classification of Herbal Substances Enhanced with an Entropy Criterion

Abstract

This paper presents a novel automatic pattern recognition system for the classification of herbal substances, which comprises the analysis of chemical data obtained from three analytical techniques such as Thin Layer Chromatography (TLC), Gas Chromatography (GC) and Ultraviolet Spectrometry (UV), composed of the following stages. First, a preprocessing stage takes place that ranges from the TLC plate image conversion into a spectrum to the normalization and alignment of spectral data for all techniques. Then, a hierarchical clustering procedure is applied for each technique with the goal of discovering groups or classes that provide evidence concerning the different existing types. Next, an entropy-based template selection step for each group was introduced to exclude the less significant samples, thus allowing to improve the quality of the training set for each technique. In this manner, each class is now described by a set of key prototypes that allows the field expert to have a more accurate characterization and understanding of the phenomenon. Moreover, an improvement of the computational complexity for training and prediction tasks of the Support Vector Machines (SVM) is also achieved. Finally, a SVM classifier is trained for each technique. The experiments conducted show the validity of the proposal, showing an improvement of the classification results on each technique.