A post-classification scheme for an OCR system for the notation of the Orthodox Hellenic Byzantine Music

In this paper we present, for the first time, the development of a new OCR system for the off-line optical recognition of the characters of the Orthodox Hellenic Byzantine Music Notation, that has been established for use since 1814. We describe the structure of the new system, and propose algorithms for the recognition of the 71 distinct character classes, based on structural and statistical features. For the classification, a tree-structured classification schema and a simple Nearest Neighbor classifier was used. In this paper, an emphasis is given for the description of the post-classification part, which is used in order to increase the overall performance of the system. The final accuracy achieved by the OCR system, tested on a data base of 18,000 characters is 99.3%. The development of such a system is of great importance to musicologists, especially in our days, which are marked by an increased interest, world wide, for the study and understanding of Eastern type musical forms.