Browse Prior Art Database

Connectivity Based Thresholding

IP.com Disclosure Number: IPCOM000117251D
Original Publication Date: 1996-Jan-01
Included in the Prior Art Database: 2005-Mar-31
Document File: 4 page(s) / 62K

Publishing Venue

IBM

Related People

Prakash, R: AUTHOR

Abstract

A method for combining two binarized images is disclosed. The two images, one richer in information with higher noise level and the other sparse in information with negligible noise, are merged using image connectivity. The result is a clean, information rich, image.

This text was extracted from an ASCII text file.
This is the abbreviated version, containing approximately 92% of the total text.

Connectivity Based Thresholding

      A method for combining two binarized images is disclosed.  The
two images, one richer in information with higher noise level and the
other sparse in information with negligible noise, are merged using
image connectivity.  The result is a clean, information rich, image.

      Extraction of a binary image from a grey scale image has seen
vast improvements over the last several years.  Most processes have
used some form of document knowledge to vary the threshold setting,
resulting in a clean binarized image.  Disclosed is a technique of
combining results of two binarized images to lead to a high quality
image.

      Figs. 1 and 2 show the results of a high performance
thresholding algorithm.

      In Fig. 1, the threshold was biased toward the high side, and
thus show a sparse image, with broken characters.  Note that the
image is very clean of background, etc.  Fig. 2 shows the same image,
with bias toward the low side for threshold setting, and thus results
in an information rich image, however with a lot of background noise.

The results of the proposed technique are shown in Fig. 3.

      Proposed methodology - The idea exploits pixel connectivity.
Examining the image in Fig. 1, all the connected pixels are
identified and placed in strings.  Each string contains connected
pixel members.  The same is done to the low biased image (Fig. 2).

      Next only those strings from Fig. 2 are extracted which...