Browse Prior Art Database

Barcode Quality - Neural Network Test

IP.com Disclosure Number: IPCOM000112630D
Original Publication Date: 1994-Jun-01
Included in the Prior Art Database: 2005-Mar-27
Document File: 2 page(s) / 59K

Publishing Venue

IBM

Related People

Kishi, GT: AUTHOR

Abstract

A method of implementing a neural network to insure barcode character quality is described. This network orders the bar widths to make all characters appear the same, scales input data, and uses a modified network for performance.

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

Barcode Quality - Neural Network Test

      A method of implementing a neural network to insure barcode
character quality is described.  This network orders the bar widths
to make all characters appear the same, scales input data, and uses a
modified network for performance.

      This article describes a neural network test used to guarantee
the quality of a barcode analyzed on a vision system.  The system
used in this article is a camera-based system, and the barcodes are
read as a series of pixels of varying intensity.

      Prior to applying these tests, one of several methods was used
to determine the location and angle of the barcode.  Then, one of
several methods was applied to determine the size of the bars in
sequence.  Once this has been done, the barcode is analyzed as
characters.  Candidate characters have filtering tests applied to
determine if they are valid or not.  Successfully read characters
have their bar width data saved.

      Once a "valid" barcode has been read, it is still possible that
it is bad.  A neural network is applied to each character of the
"valid" barcode to determine if it is indeed a good "read".  The
neural network is applied at the very end of the process, and only if
the barcode is considered "valid" because the network takes a
(relatively) long time and is only used when necessary.

      The data for each character is processed as follows.  First,
the bar widths are ordered from the widest black bar to the narrowest
black bar.  These are followed by the white bar widths from the
narrowest white bar to the widest white bar.  This ordering of b...