Dismiss
InnovationQ will be updated on Sunday, Oct. 22, from 10am ET - noon. You may experience brief service interruptions during that time.
Browse Prior Art Database

Video Thresholding for Character Recognition

IP.com Disclosure Number: IPCOM000081518D
Original Publication Date: 1974-Jun-01
Included in the Prior Art Database: 2005-Feb-28
Document File: 2 page(s) / 38K

Publishing Venue

IBM

Related People

Siverling, MM: AUTHOR

Abstract

In applications such as mail-address reading, recognition systems frequently encounter character fields which contain spurious printing or images, or in which printing on the reverse side shows through. Additionally, the desired characters in such applications are frequently degraded and often obscured by semiopaque window coverings. There exist certain techniques for reliably excluding spurious background data when the density of the desired characters is high, as within single words. But such techniques must necessarily be capable of lowering the threshold level rapidly toward white, in order to accommodate shaded printing, badly struck characters, and so forth. But then when character groups are separated, the threshold level can decrease to the point where background data is no longer excluded.

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 2

Video Thresholding for Character Recognition

In applications such as mail-address reading, recognition systems frequently encounter character fields which contain spurious printing or images, or in which printing on the reverse side shows through. Additionally, the desired characters in such applications are frequently degraded and often obscured by semiopaque window coverings. There exist certain techniques for reliably excluding spurious background data when the density of the desired characters is high, as within single words. But such techniques must necessarily be capable of lowering the threshold level rapidly toward white, in order to accommodate shaded printing, badly struck characters, and so forth. But then when character groups are separated, the threshold level can decrease to the point where background data is no longer excluded.

Thresholder 10, Fig. 1, alleviates this problem with a "soft peak detector" 16. An input video signal on line 11 is compared with a composite threshold in comparator 12 to produce a digital output signal on line 13. A dynamic threshold generator 14 uses any of a number of known methods to produce a first threshold level, which is only required to be accurate within groups of closely spaced characters. Line 15 provides a fixed minimum level to limit threshold excursions toward white.

Soft peak detector 16 is enabled at the beginning of each print line, or other read field, by a signal on line 17. Another input 18 then causes detector 16 to seek the peak black value of the input signal on line 11, by moving toward the maximum threshold leve...