SYSTEM AND METHOD OF USING PREVIOUS FRAME DATA FOR OPTICAL CHARACTER RECOGNITION OF FRAMES OF VIDEO MATERIALS

摘要

FIELD: optics.SUBSTANCE: group of inventions relates to technologies of optical character recognition (OCR) of frames of video materials in order to detect therein text in natural languages. Disclosed is a method for optical character recognition (OCR) in frame of video material. Method includes a step of obtaining a first frame of video material by hardware processor. Further, method comprises performing OCR of at least part of first frame for generation of data of first frame. Performing OCR of at least part of first frame includes detection of linked components in at least part of first frame for adding at least one description of linked components to first frame, as well as detection of candidate characters in at least part of first frame for adding at least one candidate symbol description to data of first frame. Method also includes detection of candidate text in at least part of first frame for adding at least one candidate text description to data of first frame, and detection of text lines in first part of first frame for adding at least one description text line to said first frame.EFFECT: optimising OCR of video materials.44 cl, 9 dwg