IDENTIFYING IMAGE TRANSFORMATIONS FOR IMPROVING OPTICAL CHARACTER RECOGNITION QUALITY

摘要

FIELD: data processing; optical system.;SUBSTANCE: invention relates to identification of transformations, which can be used at least part of document image to increase quality of optical character recognition (OCR). For this purpose, example of method involves using computer system of ordered list transformations, which should be applied to image containing string of characters, each conversion corresponds to hypothesis with respect to one or several image characteristics, use of image of next transformation from list to obtain converted image quality assessment of transformed image, obtaining quality assessment and updating list taking into account of obtained quality assessment.;EFFECT: higher quality of optical character recognition, which directly depends on effective determination of sequence of one or more transformations.;16 cl, 4 dwg