Browse Prior Art Database

Two Level Character Recognition

IP.com Disclosure Number: IPCOM000051358D
Original Publication Date: 1981-Jan-01
Included in the Prior Art Database: 2005-Feb-10
Document File: 2 page(s) / 65K

Publishing Venue

IBM

Related People

Atrubin, AJ: AUTHOR [+4]

Abstract

Character images stored in a computer memory are identified by first processing the image through a progressive "tree" of binary decisions sufficient to provide a reasonably reliable tentative identification. The tentative identification is then verified by further tests by which it is either confirmed or refuted. For the small number of characters refuted, a further different recognition technique is employed.

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 64% of the total text.

Page 1 of 2

Two Level Character Recognition

Character images stored in a computer memory are identified by first processing the image through a progressive "tree" of binary decisions sufficient to provide a reasonably reliable tentative identification. The tentative identification is then verified by further tests by which it is either confirmed or refuted.

For the small number of characters refuted, a further different recognition technique is employed.

Characters, such as the numbers 0 through 9 on a page 1, are captured electronically by optical scanner 2, digitized by converter 3, and are stored as an image pattern in computer 4. The computer 4 is programmed to provide a progressive series of tests 5, 6, 7 and 8, arranged in the fashion of a "tree" on each of the image patterns presented to it. Each test, for example test 6A, determines the subsequent tests required, i.e., 7A or 7B, so that only the tests needed are performed. The tests produce a tentative decision at output nodes 9, representing, for example, identification of the numbers 0 through 9.

A second series of tests, 10, 11, 12, unique to the tentative decisions, are then performed, directed at confirming or refuting the tentative decision. In the example illustrated, there are two different output nodes 9, tentatively indicating recognition of the number 8. Thus, an image that is tentatively indicated as the number 8 (N8) will be tested by test series 10 until a failure is noted. If all tests in series 10 are...