Browse Prior Art Database

Word Level Redundancy Thresholding

IP.com Disclosure Number: IPCOM000041940D
Original Publication Date: 1984-Mar-01
Included in the Prior Art Database: 2005-Feb-03
Document File: 1 page(s) / 12K

Publishing Venue

IBM

Related People

Rosenbaum, WS: AUTHOR

Abstract

The key concept on which a word level match algorithm's performance performance and reliability for facsimile is predicted is taking advantage of the redundancy inherent at the word level in making such decisions. Because of anomalies in video, the likelihood that a character geometry will be aliased is inherently much greater than a similar level of distortion affecting the correct identification at the word level. This article describes a method for establishing thresholding to reflect word level considerations in order to maximize the utility of a word level match algorithm. The word level match threshold provides an approach of thresholding the word match operation to reflect the redundancy of a subject word. The approach is as follows: 1.

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 51% of the total text.

Page 1 of 1

Word Level Redundancy Thresholding

The key concept on which a word level match algorithm's performance performance and reliability for facsimile is predicted is taking advantage of the redundancy inherent at the word level in making such decisions. Because of anomalies in video, the likelihood that a character geometry will be aliased is inherently much greater than a similar level of distortion affecting the correct identification at the word level. This article describes a method for establishing thresholding to reflect word level considerations in order to maximize the utility of a word level match algorithm. The word level match threshold provides an approach of thresholding the word match operation to reflect the redundancy of a subject word. The approach is as follows: 1. Each word is a high frequency word list is examined against a lexicon to find words that differ by only one character. These words represent the most likely mismatched aliasing candidates for the respective word in the high frequency list. Therefore, a word such as "about" will be related with candidate aliases such as "abort". Similarly, the word "man" will be related to a set of words such as "men," "can," "ran," "ban," etc. 2. The character positions that can lead to a word alias are noted for each word in the high frequency word list after examination against the respective sets of candidate word aliases. Hence, for example in the word "about" the fourth position, "u," is the "aliasing sensitive" character position. 3. A. For those character positions whose misrecognition can alias the wo...