Image analysis – Applications – Personnel identification
Patent
1997-03-06
1999-07-06
Boudreau, Leo H.
Image analysis
Applications
Personnel identification
382115, 382118, 382119, 382124, 382126, 382128, 382173, 382177, 382179, G06K 900
Patent
active
059206411
DESCRIPTION:
BRIEF SUMMARY
BACKGROUND OF THE INVENTION
The invention relates to a method for reconstructing linear structures present in raster form. Such linear structures can be used to identify persons. In order to identify persons, use is usually made of their fingerprints, but also of other linear structures such as those of the retina, the vascular plexus of the retina of the eye. Use can also be made of the linear structure in the iris of the human eye for the purpose of a unique identification of a person.
In order to be able to use the said linear structures to identify persons, a multiplicity of such linear structures must be combined in a database. A linear structure of a person to be identified is then compared with the content of this database. In the meantime, ever more voluminous collections of linear structures are exceeding the possibilities for carrying out visual comparisons in an acceptable time. However, because of the rapid developments in the fields of storage media and the techniques of digital image processing, tools are available which permit the design of powerful recognition systems.
Barry Blain, Introduction to Fingerprint Automation, Home Office Police Department, Police Systems Research & Development Group, Publication No. 1/93, 1993 discloses a method by means of which the line directions can be determined. In order to determine the line direction, the image is parqueted into slightly overlapping subregions with a size of 20.times.20 pixels, four to six pixels overlapping at the edge of the subregions. After determination of the direction, which is performed by regional Fourier transformations, a directional smoothing with subsequent binarization is undertaken. The threshold values for the binarization are determined by histograms in the subregions. The binary image thereby produced still contains prominent defects, which are later corrected by means of diverse consistency criteria. It is not always possible to expect an error-free correction of the defects, for example the merging of regions which do not belong together, or the separation of lines.
In order to achieve a high processing rate in the recognition of linear structures, the storage of essential information such as, for example, the precise position and shape of lines as well as their thickness, is frequently dispensed with during the storage of linear-structure data. Moreover, the linear structures to be coded are frequently present only in a reduced image quality. In subregions, the image is for example blurred or has insufficient contrast. In the case of coding in accordance with the prior art, the linear structure in such regions is estimated. However, it is then no longer possible to recognise from the stored binary image of the linear structure the degree of reliability with which the individual image sections were generated from the basic image pattern.
SUMMARY OF THE INVENTION
It is the object of the present invention to specify a method for reconstructing linear structures present in raster form, in which a reliable linear-structure coding which retains all the essential information of a linear structure is achieved in conjunction with as little computational outlay as possible.
By contrast with known automatic recognition systems for linear structures, in the method according to the invention essential information such as, for example, the precise position and shape of lines as well as their thickness is not dispensed with. By applying the method steps, the real linear structure is transformed into an ideal model with a locally parallel, one-dimensional linear structure, the ideal model being applied separately for each pixel. Conclusions can be drawn on the reliability of the ideal model at each pixel with the aid of the vector coefficients, which are determined as a response characteristic. The vector coefficients permit compact storability of the linear structure.
In accordance with a development and refinement of the method according to the invention, smoothing of the vector coefficients determined is carried out. A smooth
REFERENCES:
patent: 4817183 (1989-03-01), Sparrow
patent: 4881177 (1989-11-01), McClean et al.
patent: 4947442 (1990-08-01), Tanaka et al.
patent: 4953018 (1990-08-01), Martens
patent: 5465303 (1995-11-01), Levinson et al.
9.sup.th International Conference on Pattern Recognition, IEEE (1988), vol. 1, Danielsson et al, Rotation-Invariant Operators Applied to Enhancement of Fingerprints, pp. 329-333.
Mathematical Imaging: Wavelet Applications in Signal and Image processing, SPIE vol. 2034, (1993), Ronald R. Coifman, Adapted Waveform Analysis, Wavelet-Packets and Local Cosine Libraries as a Tool for Image Processing, pp. 2-12.
Mustererkennung (1990), Joachim Dengler et al, The Pseudo-Logarithmic Transformation for Bobust Displacement Estimation, pp. 275-281.
IEEE Technical Report (1990), William T. Freeman et al, Steerable Filters for Early Vision, Image Analysis, and Wavelet Decomposition, pp. 406-415.
Computer Vision, Graphics, and Image Processing, M. Hashimoto et al, Multiple-Order Derivatives for Detecting Local Image Characteristics, pp. 28-55.
Computer Vision, Graphics, and Image Processing, Michael Kass et al, Analyzing Oriented Patterns, pp. 362-385.
Transactions on Pattern Analysis and Machine Intelligence, vol. 14, No. 6, Jun. 1992, Jan J. Koenderink et al, Generic Neighborhood Operators, pp. 597-605.
Home Office Police Department, Police Systems Research and Development Group, Publication No. 1, (1993), Barry Blain, An Introduction to Fingerprint Automation, pp. 1-24.
Dengler Joachim
Ueberreiter Birgit
Boudreau Leo H.
Patel Kanji
Siemens Nixdorf Informationssysteme Aktiengesellschaft
LandOfFree
Method for reconstructing linear structures present in raster fo 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 for reconstructing linear structures present in raster fo, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Method for reconstructing linear structures present in raster fo will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-905342