JPS63106798A - Personal validation apparatus - Google Patents

Abstract

(57) [Abstract] This bulletin contains application data before electronic filing, so abstract data is not recorded.

Description

DETAILED DESCRIPTION OF THE INVENTION [Object of the Invention (Field of Industrial Application) The present invention relates to a personal authentication device that performs personal authentication using, for example, voice.

(Prior Art) A conventional voice personal authentication device is shown in FIG. Reference numeral 1 denotes a voice input unit such as a microphone that converts input voice into a corresponding electrical signal, and the user speaks appropriate voices, such as names and numbers, into the voice input unit 1. The electrical signal converted by the audio input unit 1 is sent to the feature extraction unit 2.
and is converted into a time-series pattern of personal characteristic parameters representing the characteristics of the individual. This is for example multiple (
Frequency analysis using a bandpass filter bank (usually several channels to about 20 channels), FFT analysis, cepstral analysis, linear prediction parameters, etc. are used. At the time of dictionary registration, the time-series patterns of feature parameters resulting from these analyzes are input into the dictionary section 3 and stored.

At the time of authentication, the time-series pattern of the feature parameters is input to the temporary storage section 4 and temporarily stored. Thereafter, the authentication determining section 5 performs a pass/fail determination by performing a matching operation between the time series pattern of the feature parameters in the dictionary section 3 and the time series pattern of the feature parameters in the temporary storage section 4. This is done by calculating the distance, for example, and then making a pass/fail judgment based on a predetermined appropriate equivalence value.

However, in such conventional personal authentication devices, it is known that the input information, the input voice pattern in the above column, changes over time, and after a certain amount of time, even if the speaker is the same The disadvantage is that it will not be accepted. Another drawback is that even if you temporarily catch a cold and develop a nasal voice, you will not be able to accept it in the same way.

(Problems to be Solved by the Invention) The present invention has been made in order to eliminate the drawback of not being able to follow changes over time in input information (for example, voice patterns) as described above. ,
The purpose is to provide a personal authentication device that can reliably determine the identity of the person.

[Structure of the Invention] (Means for Solving Problems) The personal authentication device of the present invention is provided with a second dictionary means that is updated hourly in addition to the first dictionary means used mainly as a dictionary for determination. It is characterized by

(Operation) Since the second dictionary means is updated every hour, if the pass/fail judgment cannot be determined by the first dictionary means, the second dictionary means is used to make the pass/fail judgment again. For example, it is possible to follow changes in voice patterns over time, so that the correct individual will not be rejected, and the identity of the individual can be determined reliably.

(Example) Hereinafter, an example of the present invention will be described with reference to the drawings.

In FIG. 1, reference numeral 11 denotes a voice input section as information input means, which converts input voice into a corresponding electrical signal.

Reference numeral 12 denotes a feature extraction section, into which the electric signal converted by the voice input section 11 is input, and converts it into a time-series pattern of personal characteristic parameters representing the characteristics of the individual by performing appropriate analysis.

Reference numeral 13 denotes a mode changeover switch for switching between registration mode and authentication mode. Reference numeral 14 denotes a first dictionary section that permanently stores and stores the time-series pattern of personal characteristic parameters output from the feature extraction section 12 in the registration mode. Reference numeral 15 denotes a temporary storage section that temporarily stores and stores the time-series pattern of personal characteristic parameters output from the feature extraction section 12 in the authentication mode.

Reference numeral 16 denotes a second dictionary section which stores and stores the time-series pattern of personal characteristic parameters in the temporary storage section 15 when an authentication determination section 1B, which will be described later, determines that the dictionary is successful.

Reference numeral 17 denotes a switch means, which is turned on when an authentication determination section 18 (to be described later) determines that the authentication is successful, and inputs the time-series pattern of personal characteristic parameters in the temporary storage section 15 to the second dictionary section 16. Reference numeral 18 denotes an authentication determination unit that performs a matching operation between the time-series pattern of personal characteristic parameters in the first dictionary unit 14 or the second dictionary unit 16 and the time-series pattern of personal characteristic parameters in the temporary storage unit 15, and makes a pass/fail determination. Do the following. A selection switch 19 is controlled by the authentication determination section 18 and selects the first dictionary section 14 or the second dictionary section 16. 2
0 is a main processor unit, which controls the operation of the corresponding device according to the determination result of the authentication determination unit 18.

Next, the operation in the above configuration will be explained. First, the user registers his/her own voice.

That is, the mode selector switch 13 is set to the registration side,
Enter an appropriate voice into the voice input section 11, such as a name,
Say a number etc. The input voice is sent to the voice input section 1.
After being converted into an electrical signal in step 1, the signal is input to the feature extraction section 12. The feature extraction unit 12 performs appropriate audio analysis processing, such as frequency analysis using a bandpass filter bank, FF
It is converted into parameters such as frequency analysis by T, cepstral analysis, and linear prediction coefficients. Also, if necessary,
Furthermore, it may be converted into a parameter representing the characteristics of an individual, such as a specific formant frequency. In short, here the voice is converted into a parameter group that includes the individuality of the voice. Then, these feature parameters are sampled at appropriate time intervals to obtain a feature pattern. This characteristic pattern may be further subjected to processing such as normalization of the time axis for convenience in later processing. Through such processing, a time-series pattern of personal characteristic parameters representing the characteristics of the individual is obtained. The time-series pattern of personal characteristic parameters obtained in this manner is input to the first dictionary section 14 and stored (registered). On this occasion,
For example, even if the user utters the words multiple times and registers them all, or even registers the multiple patterns after performing statistical processing such as the composite similarity method, essentially the It's the same. These dictionaries are recorded, for example, in a storage device such as a fixed disk device or a floppy disk device within the device, or in a personal information storage medium such as an IC card or an optical memory card.

Next, at the time of authentication, the mode changeover switch 13 is first set to the authentication side. If there is a dictionary in a storage device such as a fixed disk device, load the corresponding dictionary by an appropriate method (for example, inputting a 10 number, etc.), or if it is in a personal information storage medium such as an IC card. , load the dictionary by, for example, inserting a card. Thereafter, the user utters the same voice as the registered voice toward the voice input unit 11. Alternatively, the device may emit a voice specified by the device. The input voice is converted into an electrical signal by the voice input section 11, converted into a time series pattern of personal characteristic parameters by the feature extraction section 12, and then temporarily stored in the temporary storage section 15. Thereafter, the authentication determination section 18 performs a pass/fail determination. This pass/fail determination is performed according to the flowchart shown in FIG. That is, first, the selection switch 19 is set to the first dictionary section 1.
4 side, and calculates the degree of similarity between the time-series pattern of personal characteristic parameters in the first dictionary section 14 and the time-series pattern of personal characteristic parameters in the temporary storage section 15, or, equivalently, the distance.

For example, weighted distance, Markrid distance, cepstral distance, cross-correlation, composite similarity, etc. are used for this. If this result is expressed as the degree of similarity Si between the input pattern P of the dictionary IDi of the user i, it becomes [5i-1 (P, Di), where I is a function], and the thresholds t1 and t2 determined in advance through preliminary experiments etc. (tl < t2) is used to determine pass/fail as shown in the following equation.

3i>t2・・・・・・・・・・・・・・・Total t2≧3
1>tl...Uncertain t1≧S1...No, in other words, if it is t2 or more, it is recognized as the person, and if it is less than t1, it is rejected as not the person. However, it is determined that there is uncertainty between tl and t2. The threshold values t1 and t2 may be determined by, for example, conducting a preliminary experiment, and setting the minimum value at which it can be determined that the person is the person is t2, and the maximum value at which it is possible to determine that the person is not the person is tl.

As a result of the pass/fail judgment, if the result cannot be confirmed (if it is judged to be uncertain), the authentication judgment unit 18 switches the selection switch 1
9 on the second dictionary section 16 side, and now the second dictionary section 16
The matching calculation between the time series pattern of the personal characteristic parameters in the temporary storage section 15 and the time series pattern of the personal characteristic parameters in the temporary storage section 15 is similarly performed. The second dictionary section 16 stores and stores patterns that were passed in the previous pass/fail determination, as will be described later. In the pass/fail judgment using the second section 16, one threshold value is used to make the pass/fail judgment. The results determined as described above are output to the main processor section 20. In accordance with the pass/fail determination result of the authentication determining section 18, the main processor section 20 unlocks the door in the case of a negative determination, requests input again in the negative determination, or notifies a staff member. If a negative determination is made, the authentication determination section 18 turns on the switch means 17 to input and load the time-series pattern of the personal characteristic parameters in the temporary storage section 15 to the second dictionary section 16. . At this time, the second dictionary section 16 can also store a plurality of patterns, so that pass/fail judgment can be made from the latest few patterns, or statistical processing can be performed again from the plurality of patterns using a composite similarity method, etc. You may also create two dictionaries. Furthermore, the pattern matching calculation methods for the first dictionary section 14 and the second dictionary section 16 may be different.

The personal authentication device described above includes a second dictionary section 16 that is updated hourly in addition to the first dictionary section 14 that is mainly used as a dictionary for judgment, so that when the first dictionary section 14 makes a pass/fail judgment. If the pass/fail determination cannot be determined, the second dictionary section 16 is used to make the pass/fail determination again. According to such a personal authentication device, the second dictionary section 16 is updated every hour and keeps the latest accepted voice pattern as a dictionary, so that it is possible to follow changes over time in input voice patterns. Therefore, the correct individual will not be rejected, and the identity of the individual can be reliably determined.

In addition, although the above-mentioned embodiment explained the case where an individual is authenticated using voice, the present invention is not limited to this, and can also be applied to the case where an individual is authenticated using a fingerprint or a signature, for example. can. In short, the present invention can be applied to any personal authentication device that authenticates an individual using information representing the individual's biological characteristics.

[Effects of the Invention] As described in detail above, according to the present invention, it is possible to provide a personal authentication device that can follow changes in input information over time and can reliably determine the identity of the person.

[Brief explanation of the drawing]

FIG. 1 is a block diagram showing an embodiment of the present invention, FIG. 2 is a flowchart explaining the pass/fail judgment operation of the authentication judgment section in the same embodiment, and FIG. 3 is a block diagram showing a conventional personal authentication device. . 11... Voice input section (information input means), 12.
...Feature extraction unit (feature extraction means), 14...
... first dictionary section (first dictionary means), 15 ... temporary storage section (temporary storage means), 16 ... second dictionary section (second dictionary means), 18. ...Authentication judgment unit (judgment means). Applicant's representative Patent attorney Takehiko Suzue 2nd figure 2. Amended document for funeral settlement procedures Showa ♀2・-19th Director General of the Patent Office Kuro 1) Akio Tono 1, case indication patent application No. 1988-253206 2, Name of the invention Personal authentication device 3, relationship with the amended person case Patent applicant (307) Toshiba Corporation 4, agent UBE Building 7, 3-7-2 Kasumigaseki, Chiyoda-ku, Tokyo;
From line 5 to line 6 of page 10 of the statement of contents of the amendment, ``
Correct "Marklid distance" to "Euclidean distance".

Claims (5)

[Claims] (1) Information input means for converting the inputted information representing the biological characteristics of the individual into an electrical signal; The electrical signal converted by the information input means is input, and personal characteristic information representing the characteristics of the individual is extracted. a first dictionary means for storing and storing the personal characteristic information extracted by the feature extracting means when registering the personal dictionary; temporary storage means for storing the personal characteristic information in the temporary storage means; second dictionary means for inputting personal characteristic information in the temporary storage means and temporarily storing it; personal characteristic information in the temporary storage means and the first dictionary; a determining means for evaluating the personal characteristic information in the means or the second dictionary means and determining whether the inputted information representing the biological characteristics of the individual matches the registered personal characteristic information; and means for storing the personal characteristic information in the temporary storage means in the second dictionary means when it is determined that the personal characteristic information is valid. (2) In the normal judgment, the judgment means makes a pass/fail judgment using the personal characteristic information in the first dictionary means, and outputs the result when the pass/fail can be determined, and when the pass/fail cannot be determined. Claim 1 characterized in that the pass/fail determination is made using the personal characteristic information in the second dictionary means.
Personal authentication device described in section. (3) The personal authentication device according to claim 1, wherein the information representing the biological characteristics of the individual is voice. (4) The personal authentication device according to claim 1, wherein the information representing the biological characteristics of the individual is a fingerprint. (5) The personal authentication device according to claim 1, wherein the information representing the biological characteristics of the individual is a signature.