Authors:
Cveta Martinovska
1
;
Igor Nedelkovski
2
;
Mimoza Klekovska
2
and
Dragan Kaevski
3
Affiliations:
1
University Goce Delcev, Macedonia, The Former Yugoslav Republic of
;
2
University St Kliment Ohridski, Macedonia, The Former Yugoslav Republic of
;
3
University St Cyril and Methodius, Macedonia, The Former Yugoslav Republic of
Keyword(s):
Handwritten Character Recognition, Historical Manuscripts Recognition, Fuzzy Decision Techniques, Feature Extraction, Recognition Accuracy and Precision.
Related
Ontology
Subjects/Areas/Topics:
Advanced Applications of Fuzzy Logic
;
Artificial Intelligence and Decision Support Systems
;
Enterprise Information Systems
Abstract:
This paper presents a fuzzy methodology for classification of Old Slavic Cyrillic handwritten characters. The main idea is that the most discriminative features are extracted from the outer character segments defined by intersections. Prototype classes are formed using fuzzy aggregation techniques applied over the fuzzy rules that constitute the descriptions of the characters. Recognition methods use features like number and position of spots in outer segments, compactness, symmetry, beams and columns to assign a pattern to a prototype class. The accuracy and precision of the fuzzy classifier are evaluated experimentally. This fuzzy recognition system is applicable to a large collection of Old Church Slavic Cyrillic manuscripts.