Browse Prior Art Database

Discrimination of Characters Impinging on Form Boundaries Based on Partial Character Features

IP.com Disclosure Number: IPCOM000116957D
Original Publication Date: 1995-Dec-01
Included in the Prior Art Database: 2005-Mar-31
Document File: 4 page(s) / 85K

Publishing Venue

IBM

Related People

Kato, Y: AUTHOR

Abstract

An algorithm is disclosed that enables the discrimination of characters impinging on field boundaries of a form based on partial character features. Unlike traditional algorithms, this algorithm does not require he line removal process.

This text was extracted from an ASCII text file.
This is the abbreviated version, containing approximately 52% of the total text.

Discrimination of Characters Impinging on Form Boundaries Based on
Partial Character Features

      An algorithm is disclosed that enables the discrimination of
characters impinging on field boundaries of a form based on partial
character features.  Unlike traditional algorithms, this algorithm
does not require he line removal process.

      The forms processing needs to handle the registration error
which results in so me character data impinging on field boundaries.
Unless the field boundary lines are removed from the character data,
the Optical Character Recognition (OCR) process cannot be performed
to recognize this character data.  This line removal is a complex
process for achieving good quality.  Because character shapes have
some degree of redundancies as shown in Fig. 1, human eyes can
distinguish each digit even if the lower halves are not shown.

      Fig. 2 is a schematic illustration of this algorithm.  The
Segmentation (1) analyzes the input image and generates character
boxes circumscribing each character.  The Partial Feature Extraction
(2) determines the areas usable for the character recognition by
checking the intersections of field boundary lines and character
boxes, and then extracts the partial feature data.  The Partial
Feature Recognition (3) performs the character discrimination process
based on the partial feature data, and then generates the candidate
characters.  The Detailed Discrimination (4) performs the further
character discrimination process if the candidate characters are
found to be confu...