Browse Prior Art Database

Correction Method for OCR Results

IP.com Disclosure Number: IPCOM000116422D
Original Publication Date: 1995-Sep-01
Included in the Prior Art Database: 2005-Mar-30
Document File: 2 page(s) / 53K

Publishing Venue

IBM

Related People

Amano, T: AUTHOR

Abstract

This article describes a user interface for error correction of OCR results, which uses a history of inputs from a keyboard. Correcting character recognition errors is an indispensable task in most OCR applications, and it sometimes strongly affects a throughput of an OCR system.

This text was extracted from an ASCII text file.
This is the abbreviated version, containing approximately 62% of the total text.

Correction Method for OCR Results

      This article describes a user interface for error correction of
OCR results, which uses a history of inputs from a keyboard.
Correcting character recognition errors is an indispensable task in
most OCR applications, and it sometimes strongly affects a throughput
of an OCR system.

      To make this task easy, it is useful to provide a dedicated
user interface such that an operator can enter correct character code
by selecting the one from a possible list, not by typing with a
keyboard and a Kana-Kanji conversion program.  Conventionally, the
second or below candidates of each recognition result are used as the
items of this list.  As far as the accumulated recognition accuracy
is high, this user interface works very well.  However, due to
various reasons such as deterioration of printing quality, use of
special fonts, and appearance of non-supported characters, a correct
character is not always included in even the candidates.  If such
severe errors happen in words which are frequently used in a target
document, the error correction task requires much time because an
operator has to enter correct characters with a keyboard one by one.

      To avoid this inefficiency, a history of characters entered
with a keyboard was used.  The Figure shows an example of user
interface (a dialog box) which implements the idea.  This dialog box
is used to enter correct character when an operator detects a
recognition error.  As i...