Image analysis – Histogram processing – For setting a threshold
Patent
1987-08-13
1989-06-27
Boudreau, Leo H.
Image analysis
Histogram processing
For setting a threshold
382 17, 382 48, G06K 936
Patent
active
048436313
DESCRIPTION:
BRIEF SUMMARY
BACKGROUND OF THE INVENTION
The invention relates to a process for analysing a two-dimensional image, wherein the structural identity of stored reference patterns with image contents or portions is determined, irrespective of the position of said image content or portion in the image to be analysed, by the image being subjected to a two-dimensional Fourier transformation operation and the separated-off amplitude distribution or a distribution which can be ascertained therefrom being compared to amplitude distributions or distributions which can be ascertained therefrom, in respect of the reference patterns, in the Fourier range, while determining the respective probability of identity, the twist angle and the magnification factor as between the reference pattern and the image content or portion.
Many areas of use call upon the function of being able to recognise or identify two-dimensional images, for example by means of a television camera, and being able clearly to establish the position thereof with respect to a zero point or an axis system. Mention may be made in this connection solely by way of example of the operation of precisely positioning a gripping arm of an industrial robot relative to a given article to be gripped. In accordance with the present invention, the term two-dimensional images means two-dimensional optical images or pictures but also two-dimensional patterns which are formed by values in two-dimensional association, but not necessarily of optical origin, for example two-dimensional signals in speech analysis.
The use of one-dimensional Fourier analysis for image analysis operations is known from medical diagnostics, in particular for outline classification of organs on X-ray pictures (outline line detection).
Besides the known methods which in the image space are based on outline recognition processes (that is to say corner and edge recognition), optical processes are also known in which two-dimensional Fourier transformation operations are carried out by means of lens assemblies. Such transformation operations provide that an image content or portion can be determined, irrespective of its position in the image to be analysed.
However the known optical processes suffer from major disadvantages which hitherto prevented use thereof in a practical situation for image analysis. Firstly, the optical system is extremely complex and costly while nonetheless being fairly inflexible in regard to the parameters when once set, such as for example the size of the image to be analysed. Identification of an image content or portion irrespective of the twist and size thereof in the image and determining the extent of the twist and increase in size (or reduction in size) relative to a reference pattern is in principle possible in real time, when using such optical arrangements. In practice there is also an interest in the location (or locations) at which the identified image portion occurs in the image. Establishing those locations is not possible by means of optical processes in real time as a photographic plate would be necessary in that respect, for intermediate storage purposes. In addition, at the present time there is still no possibility of rotating and reducing images by deliberate interference.
The object of the present invention is to provide a process of the general kind set forth in the opening part of this specification, with which the position of already identified image contents or portions in the image to be analysed can be determined in real time. The invention further seeks to provide that two-dimensional images (patterns) which are not of optical origin can also be analysed by means of the system.
SUMMARY OF THE INVENTION
In a process of the general kind set forth in the opening part of this specification, that is achieved in that storage and processing of the image and the reference patterns or the Fourier transforms thereof are effected in digital form and that to locate an image content or portion in the original image, which is identical with a reference pattern with the
REFERENCES:
patent: 4227176 (1980-10-01), Moshier
patent: 4282579 (1981-08-01), Speiser et al.
patent: 4446521 (1984-05-01), Inouye
patent: 4454610 (1984-06-01), Sziklai
patent: 4660151 (1987-04-01), Chipman et al.
patent: 4703349 (1987-10-01), Bernstein
patent: 4718094 (1988-01-01), Bahl et al.
Altmann J. et al., "A Fast Correlation Method for Scale and Translation Invariant Pattern Recognition", IEEE Trans or Pattern Ana. & Mach. Int., pp. 46-57, vol. PAMI-6, #1, Jan. 1984.
Tanaka et al., "Imprinted Character Recognition Using Template Matching on Spatial Complex Frequency Plane", Pro. 3d Int. Joint Co. on Patt. Rec., pp. 663-666, Proc. 3, 1976.
Jacobson et al., "Invariant Analogical Image Representation and Pattern Recognition", Pattern Recognition Letters, Netherlands, pp. 289-299, No. 5, Sep. 1984.
Steinpichler Dietmar
sterreicher Gerhard W.
Boudreau Leo H.
Couso Jose L.
LandOfFree
Pattern recognition process does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Pattern recognition process, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Pattern recognition process will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-819237