Browse Prior Art Database

Character Recognition Analysis

IP.com Disclosure Number: IPCOM000121593D
Original Publication Date: 1991-Sep-01
Included in the Prior Art Database: 2005-Apr-03
Document File: 3 page(s) / 104K

Publishing Venue

IBM

Related People

Arnold, TW: AUTHOR [+2]

Abstract

Disclosed is a method of analyzing the performance of character recognition systems. The method is especially useful with magnetic codelines on financial documents to determine the characteristics of MICR failures in reading the codelines.

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

Character Recognition Analysis

      Disclosed is a method of analyzing the performance of
character recognition systems.  The method is especially useful with
magnetic codelines on financial documents to determine the
characteristics of MICR failures in reading the codelines.

      Various parameters affect the ability of a character
recognition system to accurately recognize a character. Print
quality, document condition, document skew (tracking), and algorithm
performance can all affect the final result of the recognition
system.  A true analysis of the system requires that the input to the
system be known along with the output, since an incorrect output
offers little help to correcting the problem unless the input
conditions are known.  This disclosure offers a method whereby the
codeline characters, the results of the character recognition
process, and an intermediate step in the process, are all displayed
for viewing by an operator.

      Fig. 1 illustrates a display containing the three different
items available for inspection by the operator or analyzer.   The
display can be produced on a video terminal screen.  The actual
document image with the magnetic codeline shown, in this example, is
a deposit slip 10.  The magnetic codeline 12 is read by the MICR
device in the associated equipment to enter data into the codeline
buffer 14.  The codeline buffer is a pushdown buffer wherein the
latest data is on the top line and previously read codeline data is
pushed down one line in the buffer for every subsequent reading.  The
image matrix 16 is displayed to indicate the dot pattern used by the
character recognizing algorithm in performing its recognition
process.  The particular character displayed in the image matrix 16
is selected by moving a cursor to the desired character in the
codeline buffer 14.  In this example, the cursor is represented by
the arrow 18 and is pointing to the zero near the end of the first
line of data.  The image matrix is a 7...