Image analysis – Pattern recognition – Feature extraction
Reexamination Certificate
1999-11-30
2004-07-20
Mehta, Bhavesh M. (Department: 2625)
Image analysis
Pattern recognition
Feature extraction
C358S453000, C358S538000, C380S051000, C382S165000, C382S209000, C382S282000, C382S288000
Reexamination Certificate
active
06766056
ABSTRACT:
FIELD OF THE INVENTION
The present invention relates to the field of image processing and, in particular, to where image processing techniques are used to detect a predefined pattern in the input image.
BACKGROUND ART
The accurate detection of marks or other indicia in documents is important in the prevention of fraud and disreputable handling of such documents. Specifically, with the introduction of high quality colour photocopying, the simple copying of banknotes has presented an immediate need for arrangements to prevent such reproductions. Whilst watermarking and holography have been used in some currencies with relative success, not all banknotes or other valuable instruments (e.g. stock certificates, bonds etc) are similarly protected.
European Patent Publication No. EP 0 744 705 A2 discloses an image detection system that stores templates of image portions and compares those templates against regions of the image a various rotations.
SUMMARY OF THE INVENTION
In accordance with a first aspect of the present invention there is provided a method of detecting a predetermined mark forming part of an image, the method comprising the steps of:
(a) providing a digitized form of the image comprising a plurality of picture elements;
(b) determining within a current region of said image a set of features comprising at least:
(i) a dominant colour feature within the current region; and
(ii) at least one moment feature based on picture elements pertaining to the predetermined mark within the current region; and
(c) comparing each determined feature with a corresponding predetermined feature of said predetermined mark;
(d) calculating, using the comparison, a confidence level for a detection of said predetermined mark within said current region; and
(e) determining, based on the confidence level, whether or not said predetermined mark is detected in the current region.
In accordance with another aspect of the present invention there is provided a method of providing multiple encoded representations of an image, said image comprising a plurality of pixels, each pixel being represented in multiple colour channels, said method comprising step of:
providing a plurality of pixels of said image data in a predetermined order;
determining a dynamic threshold value based on the provides pixel values;
thresholding each pixel of against said dynamically determined threshold to determine whether each said pixel is a foreground or a background pixel and outputting an encoded representation comprising either:
a) a background code symbol if the pixel is a background pixel; or
b) a colour code symbol representing a colour value of the pixel if the pixel is a foreground pixel;
producing a binary bitmap representation at a first desired resolution from said encoded representation the image; and
subsampling said encoded representation in accordance with said background and colour code symbols of the encoded representation to provide a second desired resolution of the encoded representation of the image.
In accordance with another aspect of the present invention there is provided a method for detecting a centre location of a predetermined pattern in an image, said method comprising the steps of:
(a) examining a plurality of pixel values at predetermined pixel locations within said image, said examined locations being determined by a plurality of groups comprising multiple locations in each group and said groups being configured into a predetermined geometric arrangement having a corresponding geometric centre location;
(b) determining an evidence value for each said group of examined pixel values;
(c) determining a deviation measure value from the evidence values of substantially all said groups of said geometric arrangement; and
(d) thresholding the deviation measure value against a predetermined threshold value and outputting either:
(i) a centre value, indicating a pixel location corresponding to the centre location, if the deviation value for each group is below the predetermined threshold value; or
(ii) an invalid value otherwise.
In accordance with another aspect of the present invention there is provided a method of determining a moment feature of a portion of an image, said image portion comprising a plurality of pixel values, said method comprising the steps of:
storing a predetermined subset of coefficient values of a moment matrix, said moment matrix having a predetermined symmetry and said subset being those coefficient values which exploit at least part of said symmetry;
combining pixel values at selected pixel locations in said image portion to provide a combined result, wherein said selected locations are associated with a respective coefficient value of said subset;
multiplying each combined result with a corresponding coefficient value; and
accumulating each multiplied combined result to determine a moment feature for said image portion.
Other aspects of the present invention including apparatus and computer program products for performing the methods are also disclosed.
REFERENCES:
patent: 5825892 (1998-10-01), Braudaway et al.
patent: 5848155 (1998-12-01), Cox
patent: 5875040 (1999-02-01), Matraszek et al.
patent: 5901245 (1999-05-01), Warnick et al.
patent: 6411392 (2002-06-01), Bender et al.
patent: 0744705 (1996-11-01), None
patent: 10-180989 (1996-12-01), None
patent: 9-329527 (1997-12-01), None
Bell Peter Leslie
Chen Yu-Ling
Foo Yoong-Chert
Huang Kai
Webb Michael John
Canon Kabushiki Kaisha
Desire Gregory
Fitzpatrick ,Cella, Harper & Scinto
Mehta Bhavesh M.
LandOfFree
Image pattern detection method and apparatus does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Image pattern detection method and apparatus, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Image pattern detection method and apparatus will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3197046