Browse Prior Art Database

Character Recognition Decision Procedure

IP.com Disclosure Number: IPCOM000095483D
Original Publication Date: 1964-Jan-01
Included in the Prior Art Database: 2005-Mar-07
Document File: 1 page(s) / 11K

Publishing Venue

IBM

Related People

Liu, CN: AUTHOR

Abstract

Pattern recognition is often enhanced by the use of multi-level decision procedures. The first-order measures are selected to provide for high overall pattern separation (discrimination) even though certain closely-related patterns, e.g., characters O and Q, cannot be distinguished.

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 100% of the total text.

Page 1 of 1

Character Recognition Decision Procedure

Pattern recognition is often enhanced by the use of multi-level decision procedures. The first-order measures are selected to provide for high overall pattern separation (discrimination) even though certain closely-related patterns, e.g., characters O and Q, cannot be distinguished.

Second-level measures are designed to resolve the ambiguous situations without adversely affecting the discrimination provided by the first-level measures. When Bayes rule decision techniques are employed, the effect of a second-level measure is restricted to the ambiguous set of patterns by using conditional probabilities of 0.5 for all other patterns.

1