Data processing: artificial intelligence – Knowledge processing system – Knowledge representation and reasoning technique
Reexamination Certificate
1999-12-16
2003-02-25
Black, Thomas (Department: 2121)
Data processing: artificial intelligence
Knowledge processing system
Knowledge representation and reasoning technique
C382S115000, C382S116000, C382S124000
Reexamination Certificate
active
06526396
ABSTRACT:
BACKGROUND OF THE INVENTION
1. Field of the Invention
The present invention relates to a personal identification method, a personal identification apparatus, and a computer readable recording medium that stores a program for realizing the functions of the personal identification apparatus, wherein, in personal identification using data indicating physical features, such as finger prints, of the subject, the determination of identification is carried out using as a reference the degree of similarity of these physical features.
2. Description of the Related Art
The personal certification method using a plurality of individual physical features (Japanese Unexamined Patent Application, First Publication, No. Hei 8-16788) is an example of a conventional personal identification method. This method reads a plurality of individual features such as voice, finger prints, and handwriting, partitions the read feature parameters into a plurality of categories that reflect these individual features, unifies each of the partitioned categories by applying a weight depending on the ease of the recognition of these individual features, unifies each feature parameter obtained by this unification by applying a weight depending on the ease of the recognition of these individual features, and carries out recognition of the subject by comparing these unified results with threshold values.
In addition, the fingerprint verification apparatus disclosed in Japanese Unexamined Patent Application, First Publication, No. Hei 10-124669, is a personal recognition apparatus. This fingerprint verification apparatus outputs an image signal by optically extracting a fingerprint, converts this image signal into fingerprint image data, extracts fingerprint feature quantities from this fingerprint image data, and outputs the code label of the closest code vector to one frame of this fingerprint feature quantity. Thereby, one sequence of a quantified code sequence for one sequence of fingerprint feature quantity is found, a statistical probability model is inferred using this quantified code sequence, an acceptance probability of the symbol sequence of the identified fingerprint is found by the statistical probability models of each registered finger, and if the acceptance probability exceeds a preset threshold, the person who is the object of the identification is determined to be the registered person.
Furthermore, the fingerprint sensor disclosed in Japanese Unexamined Patent Application, First Publication, No. Hei 8-138046, is a personal identification apparatus. This fingerprint sensor determines whether or not the fingerprint is that of the person based on the correlation output obtained by calculating the optical correlation between reflected light that contains the ridges of the fingerprint obtained on the reflective surface of a prism with which the part of a finger having the fingerprint is in contact, and a fingerprint image registered in advance as being that of the person. A Fourier spectrum integration value detection device that detects the integral value of the Fourier spectrum of the reflected light and a correlation output central value detection device that detects the peak value or integral value of this correlation output are provided, the ratio of the Fourier spectrum integral value of the reflected light and the peak value or integral value of this correlation output serves as the decision value, and by comparing this decision value with a predetermined threshold value, determines whether or not the person is the person to be identified.
In the conventional personal identification methods and personal identification apparatuses described above, the verification threshold value for identifying whether or not the person is the person to be identified uses a constant value. This is because the setting of the verification threshold value is difficult.
In the conventional personal identification methods and personal identification apparatuses described above, because the verification value for determining whether the person is the person to be identified or not uses a constant value, by setting the verification threshold value high, the danger of identifying the person to be identified as another person is decreased, but, for example, in the case of using the fingerprint as the data showing the physical feature of the individual, if fingerprints of sufficient quality cannot be obtained, the person to be identified will not be identified as that person, even though the person is the person to be identified.
Contrariwise, when the verification threshold value is set low, in the case that the fingerprint quality is not high, the person can be identified, but there is the problem that the next time a different person may be mistaken for that person.
SUMMARY OF THE INVENTION
In consideration of these problems, it is an object of the present invention to provide a personal identification method, a personal identification apparatus, and a recording medium that sets various separate threshold values for each of the feature data extracted from the data representing the physical features of the individual without decreasing the probability of excluding other people and improving the identification rate of the person.
A first aspect of the present invention is a personal identification method that finds the verification result of feature data extracted from input data showing the physical features of the individual who is the verification source and pre-registered feature data of the verification object, and identifies the individual by whether or not the degree of similarity obtained by the verification results exceeds a separate threshold value set to correspond to pre-registered feature data, and wherein this individual threshold value is calculated based on the verification score distribution of the feature data extracted from the data showing the physical features of the individual who is the verification source and the corresponding feature data, and the verification score distribution between the feature data extracted from the data showing the physical features of the individual who is the verification source and the non-corresponding verification data.
According to the first aspect of the invention, because of this personal identification method that finds the verification result of feature data extracted from input data showing the physical features of the individual who is the verification source and pre-registered feature data of the verification object, and identifies the individual by whether or not the degree of similarity obtained by the verification results exceeds a separate threshold value set to correspond to pre-registered feature data, and wherein this individual threshold value is calculated based on the verification score distribution of the feature data extracted from the data showing the physical features of the individual who is the verification source and the corresponding feature data, and the verification score distribution between the feature data extracted from the data showing the physical features of the individual who is the verification source and the non-corresponding verification data., identification of the individual is possible even in the case that the quality of the input data showing the physical characteristics of the individual who is the verification source is low and the verification score of the pre-registered feature data of the verification object is low.
A second aspect of the invention is a personal identification method that finds the verification result of feature data extracted from input data showing the physical features of the individual who is the verification source and pre-registered feature data of the verification object, and identifies the individual by whether or not the degree of similarity obtained by the verification results exceeds a separate threshold value set to correspond to pre-registered feature data, and wherein the verification result of the feature data extracted from the data showing the physical features of the individual who is
Hiratsuka Seiichi
Yano Satoru
Hirl Joseph P.
NEC Corporation
Whitham Curtis & Christofferson, P.C.
LandOfFree
Personal identification method, personal identification... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Personal identification method, personal identification..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Personal identification method, personal identification... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3167910