OCR PREPROCESSING - Enhanced binarization of highly damaged images prior to OCR processing

SATT SUD EST



22 Septembre 2016

Partager sur facebook Partager sur twitter Partager sur linkedin Partager sur google+

Fields

NICT

Sectors

NICT

BACKGROUND


Existing OCR solutions show limits with distorted or degraded supports and require an extra step of binarization that is resources-consuming. OCR PREPROCESSING is a lightning-fast algorithm dedicated to image binarization, which was specifically developed by the LSIS laboratory to address OCR requirements. OCR PREPROCESSING allows fast document image binarization pre-treatment and runs low on few resources.

OCR PREPROCESSING was awarded at multiple International Document Image Binarization Contests: ICFHR 2010 (1st), H-DIBCO 2010 (3rd), DIBCO 2011 (1st) and DIBCO 2013 (2nd).

 

KEY BENEFITS vs. STATE OF THE ART

 Higly efficient with low resolution and/or quality of text, lighting/shadows issues as well as with distortions

Sutiable for mobile applications

Robust

Real-time treatment

 

DEVELOPMENT STATUS


Software layer available
 

APPLICATIONS


OCR mails management, dematerialization, archiving, pattern recognition…
Mobile applications: smartphone, extracting business card information
Assistive technology for blind and visually impaired users
Artificial intelligence
QR Codes, Bar codes

Download the offer Download the offer

Newsletter