OCR systems in language specific environments

Abstract

This paper is focused on describing methods which can be used for improving optical recognition systems. Nowadays optical recognition systems are based on neural networks. These systems have respectable success rate in handwritten recognition. Main problem of these OCRs is in recognition of handwritten text in language specific format. Success rate is affected by specifics of these languages like diacritics etc. The main goal of this work is to discover new ways or solutions of the language specific handwritten text recognition.