Recognition features for Old Slavic letters: Macedonian versus Bosnian alphabet

This paper compares recognition methods for two Old Slavic Cyrillic alphabets: Macedonian and Bosnian. Two novel methodologies for recognition of Old Macedonian letters are already implemented and experimentally tested by calculating their recognition accuracy and precision. The first method is based on a decision tree classifier realized by a set of rules and the second one is based on a fuzzy classifier. To enhance the performance of the decision tree classifier the extracted rules are corrected according to their accuracy and coverage. The fuzzy classifier consists of rules constructed by fuzzy aggregation of letter features. Both classifiers use the same set of discriminative features, such as number and position of spots in outer segments, presence and position of horizontal and vertical lines and holes, compactness and symmetry. We argue that the same feature set can be used for recognition of Old Bosnian letters. Moreover, due to the similarity of the graphemes and fewer letters we expect better efficiency of the recognition system for Bosnian letters.