Image analysis – Image transformation or preprocessing – Measuring image properties
Reexamination Certificate
1998-08-04
2001-05-01
Mancuso, Joseph (Department: 2621)
Image analysis
Image transformation or preprocessing
Measuring image properties
C356S138000, C382S287000
Reexamination Certificate
active
06226417
ABSTRACT:
FIELD OF THE INVENTION
The current invention is generally related to an image pattern recognition process and system, and more particularly related to a method and a system for recognizing a rotated predetermined image pattern with reduced time and memory requirements.
BACKGROUND OF THE INVENTION
To recognize a rotated image pattern, a prior art system generally requires large memory space and or long processing time. This is because the recognition is usually accomplished in real time and the image pattern may be reduced, enlarged or rotated. Any combination of these factors further complicates the processing time and memory space requirements and tends to lower accuracy in correctly recognizing an image pattern.
Prior art attempts include the use of the Zernike Moment in recognizing an image pattern. For example, Japanese Laid Patent Publication Hei 9-147109 (Application Serial Number Hei 7-301250) discloses the reduced memory requirement in recognizing a predetermined image pattern based upon the use of a radial polynomial table containing intermediate values for determining Zernike Moment values. The Zernike Moment values are defined as a product of a radial polynomial value and a pixel value. Although a size difference of the predetermined image pattern is accommodated by adjusting the radial polynomial table values, a rotated image pattern as shown in
FIG. 1
is not recognized according to the disclosure. This is because the Zernike Moment values for a predetermined image pattern are constant over the rotational angle of the image.
In order to recognize a predetermined image pattern which is rotated at an arbitrary angle, prior art attempts such as Japanese Laid Patent Publication Hei 8-279021 (Application Serial Number Hei 7-301250) disclose that a rotational angle of a predetermined image pattern is determined based upon a characteristic value such as a number of “on” pixels at equidistant locations from a common point. The measure characteristic value is compared to a set of standard dictionary values each for a known angle, and a rotational angle is selected according to minimal distance to the standard characteristic value. However, this prior art attempt requires additional processing for higher degree characteristic values as well as the number of comparisons.
SUMMARY OF THE INVENTION
In order to solve the above and other problems, according to a first aspect of the current invention, a method of recognizing an image pattern having an outer boundary, including inputting an input image pattern and a standard dictionary containing standard characteristic values for predetermined image patterns; determining whether an outer boundary exists for the input image pattern; determining a rotational angle of the input image pattern based upon the outer boundary; determining a characteristic value at each of a set of predetermined relative locations within the input image pattern; adjusting the characteristic values according to the rotational angle; and determining whether the input image pattern matches one of the predetermined image patterns based upon a similarity distance between the standard characteristic values and the adjusted characteristic values.
According to a second aspect of the current invention, a method of determining a rotational angle of an image pattern having an outer boundary, including: inputting an input image pattern and a radial polynomial table containing sets of Zernike Moment (ZM) intermediate values for a predetermined specific size of the input image pattern, each set containing the ZM intermediate values each at a predetermined rotational angle for a predetermined periodicity; determining a plurality of Zernike Moment (ZM) values by multiplying a predetermined set of the ZM intermediate values by pixel values at predetermined equidistant location from a center of the input image pattern; assigning an evaluation value for each of the ZM values by multiplying the ZM value and a corresponding periodicity; and determining a rotational angle of the input image pattern based upon a largest one of the evaluation values.
According to a third aspect of the current invention, a system for recognizing an image pattern having an outer boundary, including: an input unit for inputting an input image pattern and a standard dictionary containing standard characteristic values for predetermined image patterns; an outer boundary determination unit connected to the input unit for determining whether an outer boundary exists for the input image pattern; a rotational angle detection unit connected to the outer boundary determination unit for determining a rotational angle of the input image pattern based upon the outer boundary; and a pattern matching unit connected to the rotational angle detection unit for determining a characteristic value at each of a set of predetermined relative locations within the input image pattern and for adjusting the characteristic values according to the rotational angle, the pattern matching unit determining whether the input image pattern matches one of the predetermined image patterns based upon a similarity distance between the standard characteristic values and the adjusted characteristic values.
According to a fourth aspect of the current invention, A system for determining a rotational angle of an image pattern having an outer boundary, comprising: an input unit for inputting an input image pattern and a radial polynomial table containing sets of Zernike Moment (ZM) intermediate values for a predetermined specific size of the input image pattern, each set containing the ZM intermediate values each at a predetermined rotational angle for a predetermined periodicity; a Zernike Moment generation unit connected to the input unit for determining a plurality of Zernike Moment (ZM) values by multiplying a predetermined set of the ZM intermediate values by pixel values at predetermined equidistant location from a center of the input image pattern; an evaluation unit connected to the Zernike Moment generation unit for assigning an evaluation value for each of the ZM values by multiplying the ZM value and a corresponding periodicity; and a rotational angle determination unit connected to the evaluation unit for determining a rotational angle of the input image pattern based upon a largest one of the evaluation values.
These and various other advantages and features of novelty which characterize the invention are pointed out with particularity in the claims annexed hereto and forming a part hereof. However, for a better understanding of the invention, its advantages, and the objects obtained by its use, reference should be made to the drawings which form a further part hereof, and to the accompanying descriptive matter, in which there is illustrated and described a preferred embodiment of the invention.
REFERENCES:
patent: 4748676 (1988-05-01), Miyagawa et al.
patent: 5642106 (1997-06-01), Hancock et al.
patent: 5854854 (1998-12-01), Cullen et al.
patent: 5987162 (1999-11-01), Nakata
patent: 6018990 (2000-02-01), Ueke
patent: 8-279021 (1996-10-01), None
patent: 9-147109 (1997-06-01), None
Lin, et al. “Chinese Signature Verification with Moment Invariants”, IEEE, Jun., 1996.*
Sangassapaciriya, et al. “Similarity Measures for Compressed Image Databases”, IEEE, 1997.*
Kim, et al. “A Practical Pattern Recognition System for Translation, Scale and Rotation Invaviance”, IEEE, Sep., 1994.
Hirano Akihiko
Kuranaga Tatsuya
Tachikawa Michiyoshi
Yamagata Hideaki
Kassa Yosef
Knoble & Yoshida LLC
Mancuso Joseph
Ricoh & Company, Ltd.
LandOfFree
Method and system for recognizing a rotated image pattern... 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 recognizing a rotated image pattern..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Method and system for recognizing a rotated image pattern... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2555769