Browse Prior Art Database

A New OCR Error-Correction-Method to Prevent Forgery

IP.com Disclosure Number: IPCOM000123084D
Original Publication Date: 1998-May-01
Included in the Prior Art Database: 2005-Apr-04
Document File: 2 page(s) / 88K

Publishing Venue

IBM

Related People

Sumikawa, M: AUTHOR [+2]

Abstract

Disclosed is a method for OCR error-correction to prevent people from forgery.

This text was extracted from an ASCII text file.
This is the abbreviated version, containing approximately 76% of the total text.

A New OCR Error-Correction-Method to Prevent Forgery

      Disclosed is a method for OCR error-correction to
prevent people from forgery.

   Most conventional OCR error correction methods use field
image or full image to locate miss-recognized data.  In this case an
operator is able to forge them by seeing the relation of fields.  In
addition the correction speed is not fast because an operator has to
visually check the result on a field by field basis.

   A data input method using multiple operators to input the
same fields already exists, but it cannot drastically reduce the
possibility of input errors.  Because people would make similar kind
of errors.

   Our method enables operators to reduce errors effectively
and to prevent them from forgery by using different input methods,
which are keyboard input  by operator and OCR.

   In case of OCR correction, our method mixes all the
characters from multiple forms and sorts them by recognized
character codes (categories).  An operator checks each category
group generated by OCR by his/her eyes.  See the attached figure.
The odd lines are actual character images and the even lines are
recognition results.  An operator cannot see the relationship among
fields or which character belongs to which field.  So an operator
cannot forge the data entry.  Our method improves correction speed
compared with conventional correction methods on a field by field
basis, because grouping characters by category makes it easy...