Dismiss
InnovationQ will be updated on Sunday, Oct. 22, from 10am ET - noon. You may experience brief service interruptions during that time.
Browse Prior Art Database

Computing Character Match From Stroke Matches in Online Handwriting Recognition

IP.com Disclosure Number: IPCOM000101240D
Original Publication Date: 1990-Jul-01
Included in the Prior Art Database: 2005-Mar-16
Document File: 3 page(s) / 126K

Publishing Venue

IBM

Related People

Ellozy, HA: AUTHOR [+2]

Abstract

Linear and elastic template matching are common techniques for online character recognition (1-6). One technique uses inter-point Euclidean distances to linearly 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). For this technique we have derived a formula that decomposes a character-match distance into the sum of stroke-match distances and stroke relationship terms. A stroke is the writing from pen-down to pen-up.

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

Computing Character Match From Stroke Matches in Online Handwriting Recognition

       Linear and elastic template matching are common
techniques for online character recognition (1-6).  One technique
uses inter-point Euclidean distances to linearly 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).  For this technique we have derived a formula that
decomposes a character-match distance into the sum of stroke-match
distances and stroke relationship terms.  A stroke is the writing
from pen-down to pen-up.

      The derivation of the decomposition formula follows.
Inter-point Euclidean distances are used and the number of coordinate
points in strokes to be matched is assumed equal. For simplicity,
computations are presented only for x coordinates; they are similar
for y coordinates.

      Stroke k is a sequence of coordinates, The x coordinate of the
center-of-gravity of the stroke is

      .  The x coordinate of the center-of-gravity of a character is
similarly defined, and can be expressed as the weighted sum of the
centers of gravity of the strokes, that is, where n is the number of
points and 1 the number of strokes in the character.

      We now present expressions for computing stroke and character-
match distances between an unknown and prototype character.  The x
component of the match distance of unknown u against prototype p for
stroke k is defined as the sum of the squared differences between the
normalized (by the center-of-gravity) unknown and prototype
coordinate values. This can be expressed as the sum of the squared
differences between the unnormalized unknown and prototype coordinate
values, corrected by a weighted squared difference of the centers of
gravity, that is,

      The x component of the character-match distance of unknown u
against prototype p is similarly defined.  This can be expressed as
the sum of stroke-match distances and, for each pair of strokes in
the character, a stroke relationship term, that is,

                            (Image Omitted)

      If the number of points is the same for all strokes, for
example, (2,3), the formula can be further simplified. The
character-match distance is thus decomposed into stroke-match
distances and terms involving stroke center-of-gravity differences.
The later terms account for the relative positions of strokes in a
character, permitting, for example, T to be distinguished from +.

      In our present runon handwriting recognition system, strokes
are first matched against stroke prototypes, and then combinations of
strokes likely to be characters are matched against character
prototypes.  A large percentage of computation time is used in the
character matching.  This time can be essentially eliminated by
computing the character matches directly from the stroke matches as
follows.  As...