А. В. Фролов The authors examine the issues of text machine readability in the age of digital technologies. The optical character recognition (OCR) of printed texts and manuscripts offer solutions. The main tasks of the proposed method is to analyze existing recognition systems and algorithms, own code design and testing for various fonts The authors examine the computer vision function for handwritten and printed text processing. Besides, recognition can be improved by dividing images into the black and white, and highlighting symbol parts. Many systems recognize printed text at low error rate, however recognition of handwriting is a challenge for many global languages. Not every handwritten text recognition system can be applied to printed texts, especially with use of neural networks. Most often, such systems operate with feature- or template-driven methods.
Научная статья 
код