Browse Prior Art Database

Elastic Matching by Decomposition in Online Character Recognition

IP.com Disclosure Number: IPCOM000120383D
Original Publication Date: 1991-Apr-01
Included in the Prior Art Database: 2005-Apr-02
Document File: 2 page(s) / 91K

Publishing Venue

IBM

Related People

Tappert, CC: AUTHOR

Abstract

Linear and elastic template matching are common techniques for online character recognition (1). One technique uses inter-point Euclidean distance to match the same number of coordinate points in corresponding strokes, where the points are normalized by the center of gravity of the character, for example (2,3). A stroke is the writing from pen-down to pen-up. For this technique, a formula was previously disclosed that decomposes the character-match distance into the sum of stroke-match distances and stroke-relationship terms, hereafter referred to as the decomposition formula (4). The decomposition formula is exact for linear matching of equal number of points in corresponding strokes but is not exact for elastic matching. It is well known, however, that elastic is superior to linear matching (1).

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

Elastic Matching by Decomposition in Online Character Recognition

      Linear and elastic template matching are common
techniques for online character recognition (1).  One technique uses
inter-point Euclidean distance to match the same number of coordinate
points in corresponding strokes, where the points are normalized by
the center of gravity of the character, for example (2,3).  A stroke
is the writing from pen-down to pen-up.  For this technique, a
formula was previously disclosed that decomposes the character-match
distance into the sum of stroke-match distances and
stroke-relationship terms, hereafter referred to as the decomposition
formula (4).  The decomposition formula is exact for linear matching
of equal number of points in corresponding strokes but is not exact
for elastic matching.  It is well known, however, that elastic is
superior to linear matching (1).  Therefore, it is important to
determine whether and, if so, how the decomposition formula can be
used with elastic matching.

      Disclosed herein is a new method of elastic matching of
characters that uses the decomposition formula.  The new
elastic-match distance of a character is defined as the sum of the
elastic-match distances of the strokes plus the stroke-relationship
terms of the decomposition formula.  We call this elastic matching by
decomposition.  Where the formula requires the number of points in
the k-th stroke, we use the number of points in the k-th stroke of
the unknown because that is the number of point distances summed to
compute the stroke distance.

      Compared to earlier Euclidean-distance formulations of elastic
matching, elastic matching by decomposition gives comparable results
and similar, but not identical, distance scores.  Thus, elastic
matching by decomposition has all the advantages provided by the
decomposition formula.  For example, it provides a fast response
because stroke matching distances can be computed as the strokes are
entered, important in the recognition of run-on characters
(characters can touch or overlap one another).  It may also be useful
for stroke-order-independent algorithms (3,5,6), since the stroke
relationship terms can be computed after...