| Popis: |
The paper discusses Optical Character Recognition (OCR) of historical texts of the 18th–20th century in the Romanian language using the Cyrillic script. We differ three epochs (approximately, the 18th, 19th, and 20th centuries), with different usage of the Cyrillic alphabet in Romanian and, correspondingly, different approach to OCR. We developed historical alphabets and sets of glyphs recognition templates specific for each epoch. The dictionaries in proper alphabets and orthographies were also created. In addition, virtual keyboards, fonts, transliteration utilities, etc. were developed. The resulting technology and toolset permit successful recognition of historical Romanian texts in the Cyrillic script. After transliteration to the modern Latin script we obtain no-barrier access to historical documents. |
