SUPERVISED MACHINE LEARNING FOR HANDWRITTEN TEXT RECOGNITION IN CYRILLIC SCRIPTS: A CASE STUDY FOR RUSSIAN AND KYRGYZ LANGUAGES
Handwritten text recognition is still hard. The difficulty comes from the fact that each person writes in a way the scanning process adds noise. There are few data sets for each language. This study uses a machine learning method to read text, in Cyrillic scripts focusing on Russian and Kyrgyz. The supervised machine learning method uses image representation. Compares each image with a database of reference character patterns. In my view handwritten text recognition can improve with this method. First, we turn each character into a matrix. Then we classify the matrix with distance-based similarity measures. These measures include distance and vector similarity metrics. The system runs a training phase. During the training phase the system stores handwriting variants, for each handwritten character. The system does this to make the recognition more robust. We test the system with experiments. The experiments show that the approach gives recognition for handwritten characters and short words in both languages. The results show that adding entries to the reference database and using a matrix resolution both raise recognition accuracy. The developed system provides a practical and interpretable solution for handwritten text recognition in underrepresented languages and can be extended to other Cyrillic-based scripts.
Сведения
- Опубликовано
- 2026 · Т. 10-741, № 1/2, С. 48-55
- Язык
- Английский
- ISSN
- 1694-7851 · 1694-8505
- Издатель
- Kyrgyz State University named after I. Arabaev
- Аффилиации
- Kyrgyz State University named after I. Arabaev