Browse Prior Art Database

Character Recognition With Multiple Variable Rules

IP.com Disclosure Number: IPCOM000038750D
Original Publication Date: 1987-Mar-01
Included in the Prior Art Database: 2005-Feb-01
Document File: 3 page(s) / 46K

Publishing Venue

IBM

Related People

Hare, DC: AUTHOR [+2]

Abstract

The mechanism described in this article for character recognition takes advantage of existing matrix reference techniques to preserve sensitivity, under a single reference operation, to all criteria (called "rules") for which a character is recognized. This economy is gained by a new configuration of the matrix (called the "recognition matrix") and a new method for its maintenance. The recognition text (1) is a bit pattern in which each position represents a rule for which the character is recognized. An "on" bit signifies a specific action to be taken upon recognition, and the length of the recognition text is equal to the number of possible actions. A null recognition text has no bits "on" and indicates that a character is not recognized for any rule.

This text was extracted from a PDF file.
At least one non-text object (such as an image or picture) has been suppressed.
This is the abbreviated version, containing approximately 53% of the total text.

Page 1 of 3

Character Recognition With Multiple Variable Rules

The mechanism described in this article for character recognition takes advantage of existing matrix reference techniques to preserve sensitivity, under a single reference operation, to all criteria (called "rules") for which a character is recognized. This economy is gained by a new configuration of the matrix (called the "recognition matrix") and a new method for its maintenance. The recognition text (1) is a bit pattern in which each position represents a rule for which the character is recognized. An "on" bit signifies a specific action to be taken upon recognition, and the length of the recognition text is equal to the number of possible actions. A null recognition text has no bits "on" and indicates that a character is not recognized for any rule. The recognition text results from applying the recognition matrix (2), by associative reference, to each successive character (3) in an input data stream, presenting the character and recognition text to the control program whenever a non-null text results (and otherwise expediting the stream). Thus: Output = Parametric function (input character) Where: Parametric function is the operation represented by - Recognition matrix (input character) - H set of actions Each entry in the recognition matrix is the recognition text for the corresponding member of the character set. For single-byte character sets, the recognition matrix has maximum dimensions (16x16) where each member is (N) bits in size (one bit for each of N rules). The recognition matrix is configured by the operation OR applied to N constituent rule matrices (4) and the zero matrix, collectively. The recognition matrix has the same dimensions as rule matrices, where rule matrix entries have a single a...