Image analysis – Applications – Personnel identification
Reexamination Certificate
1998-04-27
2001-03-06
Mehta, Bhavesh (Department: 2621)
Image analysis
Applications
Personnel identification
C382S126000
Reexamination Certificate
active
06198836
ABSTRACT:
BACKGROUND
The invention relates to a method and an apparatus for the identification of not-enrolled fingerprints. Numerous methods and apparatuses are known, which allow the identification of a fingerprint. Generally the finger is pressed against a suitable optical device for producing contrast and e.g. by means of prisms and the inhibited total reflection principle, an image of the skin lines is produced. For example a digital grey value image is recorded, which is converted into a binary image and compressed by means of suitable algorithms to a biometric pattern with a relatively small storage requirement, which can then be compared with stored finger patterns of an associated data bank for identification purposes. The identification of fingerprints by enrolled finger line images is well established and proven in the police field. However, an increasing number of non-police uses require not-enrolled fingerprints. Thus, in most cases it is not possible or desirable to prescribe the enrolling of the fingerprint. In fact in many cases it is legally forbidden. In the case of not-enrolled fingerprints, which are required for normal applications, a very serious problem arises. Thus, if the orientation of the finger with respect to its longitudinal axis, i.e. a twisting or turning of the finger relative to the image recording surface occurs, then completely different partial image quantities of a complete enrolled fingerprint are identified, as a function of the twisting or rotation angle of the finger applied. In an extreme case this can mean that the identified partial image quantity of the surface of the finger on rotating the finger about its axis by e.g. +20° and −20° leads to a common image section quantity of these positions, which makes it impossible in practice to identify with the mathematical method used as a basis the person in question with an adequate accuracy using the differently applied finger. Thus, the information still contained in the common image section quantity can be too small to allow a reliable identification.
The problem of the invention is to give a method and an apparatus enabling this serious difficulty to be overcome and which allow a considerably increased identification reliability and accuracy and this is to be performable with a relatively small storage requirement for a significant pattern.
SUMMARY OF THE INVENTION
According to the invention this problem is solved by a method according to claim
1
and by an apparatus according to claim
17
. Images of skin lines of two juxtaposed fingers of a hand are recorded, the orientations of the longitudinal axis L
1
, L
2
being determined or are to be determined and in which the rotation of the finger about its longitudinal axis is largely avoided by a guidance means, so that the fingers are held virtually in the normal position and consequently clearly defined, almost identical partial image quantities of the entire enrolled fingerprint images are detected and transferred by suitable evaluation algorithms into a reduced data quantity. As a result of the unrotated position and the determination of the finger longitudinal axes, the evaluation algorithms become simpler, because generally more complicated algorithms are required in the case of an undefined finger longitudinal axis and which are rotation-independent, i.e. of the Lorenz type. The method according to the invention on the one hand gives simpler algorithms and on the other makes it possible for the same memory size to determine more significant patterns. Simultaneously through the detection of the image informations of two fingers the false acceptance rate FAR and false reject rate FRR are significantly improved. If e.g. the FAR in a method is 1:1000 per finger, the reliability of identification through the use of two fingers is 1:1 million. Or, conversely, it is possible to obtain the same identification reliability with simpler features of both fingers.
The dependent claims relate to advantageous further developments of the invention with further improvements concerning system expenditure, storage requirements, identification reliability and also with regards to the making and evaluation of picture records.
REFERENCES:
patent: 3668633 (1972-06-01), Sadowsky
patent: 4573193 (1986-02-01), Shuto et al.
patent: 4876725 (1989-10-01), Tomko
patent: 5073950 (1991-12-01), Colbert et al.
patent: 5335288 (1994-08-01), Faulkner
K. Asai, Y. Hoshino, N. Yamashita, S. Hiratsuka “Fingerprint Identification System,” Second USA-Japan Computer Conference, pp. 30-35, 1975.
H. Yoshikawa, S. Ikebata, “A Microcomputer-Based Personal Identification System,” 1984 International Conference on Industrial Electronics, Control and Instrumentation, pp. 105-109, 1984.
M. Takeda, S. Uchida, K. Hiramatsu, T. Matsunami, “Finger Image Identification Method for Personal Verification,” 10th International Conference on Pattern Recognition, pp. 761-766, 1990.
A-Tronic MGM AG
Mehta Bhavesh
Oppedahl & Larson LLP
LandOfFree
Method and device for recognizing non-unrolled fingerprints 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 device for recognizing non-unrolled fingerprints, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Method and device for recognizing non-unrolled fingerprints will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2542372