Problems of digital conversion and automatic translation of text

The paper deals with the classification methods used in digital processing of bi-tonal image of the text, obtained by scanning or digital photography. For the currently known classification algorithms, including well-established - an algorithm JB2, given the quantitative characteristics of the classification - the number of classes obtained by these algorithms to image the standard page of text. Thus, it was shown that the smallest possible number of classes in the classification of images of text characters can reduce the recognition error text optical character recognition systems. Minimizing OCR errors can improve the quality of automatic translation into a foreign language.