Authors of scientific publications and books use images to present a wide spectrum of information. Despite the richness of the visual content of scientific publications the figures are usually not taken into consideration in the context of text mining methodologies towards the automatic indexing and retrieval of scientific corpora. In this work, we present a system for automatic categorization of figures from scientific literature to a set of predefined classes. We have employed a wide range of visual features that achieve high discrimination ability between the adopted classes. A real-world dataset has been compiled and annotated in order to train and evaluate the proposed method using three different classification schemata.