Browse Prior Art Database

Online Form Free Japanese Character Segmentation

IP.com Disclosure Number: IPCOM000121745D
Original Publication Date: 1991-Sep-01
Included in the Prior Art Database: 2005-Apr-03
Document File: 3 page(s) / 91K

Publishing Venue

IBM

Related People

Kitamura, K: AUTHOR [+3]

Abstract

Disclosed is a method of online Japanese handwritten character segmentation for use in character recognition. In this method, rough segmentation process is added before precise character segmentation (Fig. 1). The rough segmentation, which extracts groups of strokes containing more than one character, allows not only writing-box- and guide-line-free segmentation but also on-the-fly recognition.

This text was extracted from an ASCII text file.
This is the abbreviated version, containing approximately 56% of the total text.

Online Form Free Japanese Character Segmentation

      Disclosed is a method of online Japanese handwritten
character segmentation for use in character recognition.  In this
method, rough segmentation process is added before precise character
segmentation (Fig. 1).  The rough segmentation, which extracts groups
of strokes containing more than one character, allows not only
writing-box- and guide-line-free segmentation but also on-the-fly
recognition.

      The candidate character lattice method is one of a few existing
methods of writing-box-free recognition.  The unsolved problem of
such methods is that processing is lengthy when many characters are
written at one time.  Rough segmentation is an effective way of
reducing for the processing time for on-line form-free Japanese
character recognition.

      A general flowchart of this method is shown in Fig. 2. Rough
segmentation divides the stroke sequence where a new character
obviously begins.  Rough segments contain strokes that form more than
one character.  Features used for rough segmentation are:
  - the time difference between the end of one stroke and the
beginning of the next, and
  - the overlapping portion of one stroke and the next.

      That is, for the two successive pen strokes S(i) and S(i+1), if
  T(S(i), S(i+1)) > Trough
     and
  L / W(S(i))   < Lrough
     and
  L / W(S(i+1)) < Lrough
then do rough segmentation between S(i) and S(i+1), where T(S(i),
S(i+1)) is the...