SYSTEM FOR OPTICAL CHARACTER RECOGNITION (OCR)

A system (1) for optical character recognition (OCR) from an image containing text shall minimize transcription errors and improve the quality of the resulting machine-encoded text. To this end, the system comprises:- a text area dispatcher module (6), connected to a collection (8) of OCR engines (a...

Full description

Saved in:
Bibliographic Details
Main Authors Dr. Schreiber, Gerald, Kolodziej, Richard, Bauer, Joachim, Lissowski, Thomas, Dinu, Ciprian
Format Patent
LanguageEnglish
French
German
Published 19.02.2020
Subjects
CALCULATING
COMPUTING
COUNTING
PHYSICS
Online AccessGet full text

Cover

More Information
Summary:A system (1) for optical character recognition (OCR) from an image containing text shall minimize transcription errors and improve the quality of the resulting machine-encoded text. To this end, the system comprises:- a text area dispatcher module (6), connected to a collection (8) of OCR engines (aTR1, aTR2,...aTRn), and configured to allocate said image as input to a plurality of said OCR engines (aTR1, aTR2,...aTRn) and receive machine-encoded texts as output from each of said plurality of said OCR engines (aTR1, aTR2,...aTRn), and- an OCR validation engine module (10), configured to receive said machine-encoded texts and assign a confidence level to each of said machine-encoded texts,wherein said text area dispatcher module (6) is further configured to receive said confidence levels from the OCR validation engine module (10), wherein said plurality of said OCR engines (aTR1, aTR2,...aTRn) is a proper subset of said collection (8) of OCR engines (aTR1, aTR2,...aTRn), and wherein said text area dispatcher module (6) is further configured to choose said plurality of said OCR engines (aTR1, aTR2,...aTRn) based on previously received confidence levels.
Bibliography:Application Number: EP20180189225