Browse Prior Art Database

ADAPTIVE BINARISER

IP.com Disclosure Number: IPCOM000024718D
Original Publication Date: 1981-Oct-31
Included in the Prior Art Database: 2004-Apr-02
Document File: 4 page(s) / 133K

Publishing Venue

Xerox Disclosure Journal

Abstract

In converting the video signal from a line copy type document in-a scanning system into digital form, the video signal is usually compared to a fixed threshold, and a binary output is obtained depending upon whether the video signal is greater than or less than the threshold. This method, however, does not handle variations in background and text densities well, resulting in inaccurate reproduction of line widths or missing of the lines completely. Furthermore, a similar problem is encountered with small signal amplitudes due to system Modulation Transfer Function (MTF) for narrow lines.

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 58% of the total text.

Page 1 of 4

ADAPTIVE BINARISER Sagdish C. Tandon Jack T. Latone
James C. Stoffel

Proposed Classification
U.S. CI. 358/280 Int. CI. H04n 1/40

In converting the video signal from a line copy type document in-a scanning system into digital form, the video signal is usually compared to a fixed threshold, and a binary output is obtained depending upon whether the video signal is greater than or less than the threshold. This method, however, does not handle variations in background and text densities well, resulting in inaccurate reproduction of line widths or missing of the lines completely. Furthermore, a similar problem is encountered with small signal amplitudes due to system Modulation Transfer Function (MTF) for narrow lines.

To provide accurate binarisation over a wider range of background, text density, and MTF conditions as compared to a fixed threshold system, two different algorithms are proposed.

Volume 6 Number 5 September/October 1981 283

[This page contains 1 picture or other non-text object]

Page 2 of 4

ADAPTIVE BINARISER (Cont'd)

FIG: 2

VIDEO SIGNAL f5 510 h

).

51Kn

DELAY LINE

DATA BINARY - --7-

Algorithm 1:

The threshold is computed by the relation:

      N vth,(t) =

IKa

and kn are weighting factors so that kn = 0.5 (3)

The threshold is, however, limited in a range between V1 and V2, i.e.,

1' 'thl"2 (4)

Algorithm 2:

The threshold in this case is computed by the relation:

XEROX 284 DISCLOSURE JOURNAL

Volume 6 Number 5 September/October 1981

- + lkn = where, n = 1, 2, 3, ... N (v(t nAt) + v(t nbt)

+V b

[This page contains 1 picture or other non-text object]

Page 3 of 4

ADAPTIVE BINARISER (Cont'd)

where K is a constant less than 1, VF is a fixed level and vthl is obtained from equations (11, (21, and (3).

Specific Examples and Implementation:

In the ensuing discussion, the simplest case of N and implementation. However, both algorithms general form.

Algorithm 1 (N = 1)

In this case

v (t) = 0.5 (V (t + At) + v (t -At) th 1

and

1' 'thl' '2

= 1 is used both for example can be implemented in their

This means that the threshold is an average of two adjacent values of the signal separated bybt from the time under consideration (At can be chosen for a specific application), but is limited to stay in the range between V1 and V2.

One possible implementati...