Image analysis – Image enhancement or restoration – Image filter
Patent
1997-01-09
2000-06-27
Rogers, Scott
Image analysis
Image enhancement or restoration
Image filter
382275, 382176, 358463, 358464, G06K 940, H04N 1409
Patent
active
060816260
DESCRIPTION:
BRIEF SUMMARY
BACKGROUND OF THE INVENTION
1. Field of the Invention
The invention relates to the separation of foreground information from background information in a document.
2. Description of the Art
In order to be able to carry out an image or character recognition procedure (optical character recognition) in an electronically-scanned and digitalized document, it is frequently necessary, in an initial preparatory stage (image preprocessing) for foreground information to be separated from the background information. In this situation, the foreground information is often the actual image information which comes into consideration for the character recognition, while the background information in many cases has an interfering effect, and may even lead to misinterpretation of the foreground information. The background information of the picture also frequently contributes to a reduction in the identification performance of the foreground information. The background information should therefore, for preference, be removed before the picture or character recognition. Foreground information, which is intended to represent a script, is, in many applications, required to be converted by the subsequent character recognition, for preference into ASCII data.
The background data in a picture is in most cases composed of the following groups: interference in digitalization (such as noise), and components of the digitalized document which cannot be filtered out (such as dirt, script fragments, stamps, etc.). In order to achieve the best possible identification rate in a subsequent processing stage, the image data should, for preference, after an image processing procedure, only contain that data which it is intended should be processed later by the character recognition process. The better the image processing prior to recognition, the more favorable the recognition results can be expected to be.
Many image processing units expect, as input, an image reduced in the grey scales; e.g. a binarised black/white image. For the production of an electronically-scanned image of document, reader devices (scanners) are frequently used, which supply images at a predetermined scanning frequency (e.g. 200 to 300 pixels per square inch), with a large number of grey scales (e.g. 16 or 256 grey scales).
Due to a series of potential faults when a document is being read in, such as, for example, scanner tolerances or an inadequate match or balance between the individual laser elements of a scanner, it is possible that a series of influencing factors may arise as early as at this stage, which could exert a disadvantageous effect on image recognition. Likewise, the type and quality of the document itself may decisively influence the image recognition. For example, color contrasts which may be present might be lost during the conversion of a color image into a grey image. In the case of handwritten entries on a document, but with typewritten entries too, the readability and recognition capacity also depend in particular on the degree of contrast involved, caused for example by a more or less strong pressure exerted when writing, etc.
For a series of applications, there is a requirement for what is known as an archive picture to be created in addition to the character recognition, in order, for example, to reduce the amount of paper used. In other words, two images have to be created, which should, as far as possible, be done without loss of throughput. Of the applications available at the moment, black/white pictures are used for this purpose, both for the recognition unit as well as for archiving purposes. Grey scale images must be intensively compressed and decompressed by computer processing, which accordingly means they have a higher memory requirement than black/white pictures.
By electronic scanning of a document, an electronic image of the document is obtained, in which an image value is allocated to every pixel, for example a grey value, a black or white value, or a color value. By means of the use of filters, these allocations can be influenc
REFERENCES:
patent: 5142589 (1992-08-01), Lougheed et al.
patent: 5335292 (1994-08-01), Lovelady et al.
patent: 5710830 (1998-01-01), Holeva
Bruegmann Winfried
Schroeder Horst
Hesse Karl O.
International Business Machines - Corporation
Rogers Scott
LandOfFree
Method and system for background removal in electronically scann does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Method and system for background removal in electronically scann, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Method and system for background removal in electronically scann will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-1790915