Browse Prior Art Database

Greyscale Assist for Machine Recognition of Courtesy Amounts On Checks

IP.com Disclosure Number: IPCOM000121998D
Original Publication Date: 1991-Oct-01
Included in the Prior Art Database: 2005-Apr-04
Document File: 4 page(s) / 131K

Publishing Venue

IBM

Related People

Bedell, RE: AUTHOR [+4]

Abstract

Disclosed is a method by which greyscale data at a lower resolution is combined with binary data at a higher resolution to improve recognition reliability. This provides all the benefits from the greyscale information while reducing the data handling requirements.

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

Greyscale Assist for Machine Recognition of Courtesy Amounts On Checks

      Disclosed is a method by which greyscale data at a lower
resolution is combined with binary data at a higher resolution to
improve recognition reliability.  This provides all the benefits from
the greyscale information while reducing the data handling
requirements.

      Personal and business checks have convenience amount areas
where the amount of the check is handwritten or printed in Arabic
numbers.  The convenience amount data is viewed and manually keyed by
key-entry operators.  Machine recognition reduces this keying effort
significantly.  In order to maximize the recognition rate, the
characters must be reliably segmented before applying the recognition
logics.

      Traditionally, a dynamic thresholding algorithm converts the
captured greyscale data to a black/white or binary image.  This
binary image is used for character segmentation and recognition.
Convenience amount boxes and sometimes the background security
pattern interfere with the data to be recognized.  The thresholding
parameters are normally optimized over the entire document and are
not optimized for recognition of the convenience amount.  The typical
problems encountered are:
   . the box is not uniformly maintained,
   . the box fragments interfere with the character and, hence,
making segmentation harder and unreliable, and
   . breaks in character strokes.

      The objectives for preprocessing steps prior to recognition are
as follows:
   1.  First, recognize the $ symbol and the box in order to locate
the convenience amount area.
   2.  Next, filter the $ and the box out so that the amount field
can be segmented without any interference, but maintain the desired
character pieces overlapping the box.
   3.  Maintain the characters without breaking up the character
strokes.
   4.  Use serpentine segmentation well known in the prior art for
segmenting the characters written with tilt and overlap.

      This article explains a method of combining low-resolution
greyscale with high resolution binary data to reliably perform the
preprocessing steps.  For handwriting recognition, the preferred
resolution for binary data is 240 pels/inch.  It has been observed
that the full advantages of the greyscale information can be realized
using data at one-third this resolution (80 pels/inch).  While this
is the preferred implementation, the method is equally applicable for
other binary to greyscale resolution ratios.

      Fig. 1 shows the ty...