Browse Prior Art Database

Image Matching Character Recognition System

IP.com Disclosure Number: IPCOM000095225D
Original Publication Date: 1965-Oct-01
Included in the Prior Art Database: 2005-Mar-07
Document File: 3 page(s) / 62K

Publishing Venue

IBM

Related People

Stockdale, NS: AUTHOR

Abstract

As characters on a document are scanned, the video is quantized into binary bits. These are entered into storage matrix 10. In order to recognize the unknown scanned character, the data in 10 is then compared with references contained in storage 15. The references each consist of ternary bits representing known characters. The three states of the reference bits are black, white and boundary. The boundary bit is indicative of a don't care condition and this accounts for possible variations in line width of the characters being scanned.

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

Image Matching Character Recognition System

As characters on a document are scanned, the video is quantized into binary bits. These are entered into storage matrix 10. In order to recognize the unknown scanned character, the data in 10 is then compared with references contained in storage 15. The references each consist of ternary bits representing known characters. The three states of the reference bits are black, white and boundary. The boundary bit is indicative of a don't care condition and this accounts for possible variations in line width of the characters being scanned.

The quantized video data representing the unknown character is compared with the reference data. The number of mismatches resulting from the comparison is entered into a counter. The reference most similar to the unknown character produces the least number of mismatches during the comparison operation. This reference is then considered as identifying the unknown character. By including the don't care conditions in the references, mismatches do not develop due to variations in line width of the unknown characters being scanned. In other words, if the unknown character is a thin or thick F, for example, then the reference for the character F still results in the least number of mismatches and identifies the unknown character as an F. If variations in line width were not accounted for, through the facility of the don't care conditions, then a reference other than the true reference for that character might result in a lesser number of mismatches and the unknown character would be misidentified.

After the quantized video data from each scan is loaded into matrix 10, a scanner timing signal is applied to terminal 20 to start the storage address ring and clock circuits 25. The outputs of 25 are connected to Or 30. The latter's output is connected to set latch 35 which, upon being set,...