Metaheuristic approach to optical character recognition of Old Slavic letters

Abstract

This paper introduces a metaheuristic approach to the problem of OCR of Old Slavic letters. This approach consists of several sequential stages: filtering, binarization, skeletonization, feature extraction and decision tree classification. The image preprocessing steps proved to provide improvements to the correctness of the letter classification, since the quality of the filtered image influences the feature extraction phase, which is based on the unique characteristics of the Old Slavic letters. Various verifications of this approach were performed and the correctness of this approach has been proven.