METHODS AND SYSTEMS OF OPTICAL IDENTIFICATION SYMBOLS OF IMAGE SERIES Russian patent published in 2018 - IPC G06K9/62 

FIELD: data processing.

SUBSTANCE: invention relates to performing recognition of a series of images containing text characters. It is associated with the use of coordinate transformations of a part of the OCR text with the first cluster of a plurality of clusters of character sequences, where OCR text is obtained by processing the current image, and where character sequences are obtained by processing previously obtained images from a series of images. First median line is displayed, representing the first cluster of a character sequence, based on the first subset of images. First field of the template from the document template is identified corresponding to the first cluster, starting from the first median line representing the first cluster, and marking the text of the current image. Sequence of characters of the first cluster is analyzed to identify suitable sequences of characters, where suitable sequences of characters satisfy the first parameters of the first field of the template. For the first cluster, the second-level median strings corresponding to the cluster of character sequences are identified, based on the set of suitable character sequences. They are obtained using the second-level median string of the resulting OCR text, representing at least part of the first field of the original document template.

EFFECT: technical result is to improve the quality of optical recognition.

20 cl, 9 dwg