Biometric identification apparatus and method using bio signals and artificial neural network

• US 7,630,521 B2
• Filed: 12/21/2005
• Issued: 12/08/2009
• Est. Priority Date: 03/25/2005
• Status: Expired due to Fees

• Current Assignee: Samsung Electronics Co. Ltd.
• Original Assignee: Samsung Electronics Co. Ltd.
• Inventors: Kim, Kyung-ho, Yoon, Tae-ho, Kim, Kyeong-seop
• Primary Examiner(s): Desire; Gregory M
• Application Number:
  US11/312,442
• Publication Number:
  US 20060215883A1
• Time in Patent Office: 1,448 Days
• Field of Search: 382/115, 382/128, 382/156, 382/159, 600/300, 600/509, 600/516
• US Class Current: 382/115
• CPC Class Codes: G06F 18/00 Pattern recognition
• CPC Class Codes: G06F 2218/00 Aspects of pattern recognition specially adapted for signal processing
• CPC Class Codes: G06V 40/15 Biometric patterns based on physiological signals, e.g. heartbeat, blood flow

Abstract

A biometric identification apparatus and method using bio signals and an artificial neural network, are provided. The biometric identification apparatus includes: a periodic signal extraction unit which extracts one or more periodic signals from an input bio signal; a template calculation unit which calculates a template value using the extracted periodic signals; a template storage unit which stores a plurality of template values corresponding to a plurality of living bodies; and a reading unit which reads the template value that is most approximate to the template value calculated by the template calculation unit from the template storage unit. Accordingly, it is possible to identify a living body by taking into consideration all of the characteristics of bio signals detected from the living body.

Claims

1. A biometric identification apparatus comprising:

a periodic signal extraction unit which extracts one or more periodic signals from an input bio signal;

a template calculation unit which calculates a template value using the periodic signals which are extracted, the template value comprising a connection strength between neurons of an artificial neural network corresponding to the bio signal;

a template storage unit which stores a plurality of template values which correspond to a plurality of living bodies; and

a reading unit which reads the template value that is most approximate to the template value which is calculated by the template calculation unit from the template storage unit.

2. The biometric identification apparatus of claim 1 further comprising an identification unit which identifies a living body corresponding to the template value which is read out from the template storage unit as a living body which corresponds to the input bio signal.

3. The biometric identification apparatus of claim 1 further comprising:

an examination unit which determines whether a living body corresponding to the template value which is read out from the template storage unit is identical to a living body to be authenticated; and

an authentication unit which authenticates the input bio signal as a bio signal which is detected from the living body to be authenticated in response to the determination results which are provided by the examination unit.

4. The biometric identification apparatus of claim 1 further comprising an analog-to-digital conversion unit which converts the input bio signal into a digital signal through a sampling operation,

wherein the periodic signal extraction unit extracts one or more periodic signals from the digital signal into which the input bio signal is converted.

5. The biometric identification apparatus of claim 1, further comprising:

an examination unit which determines whether a living body corresponding to the template value which is read out from the template storage unit is identical to a living body to be authenticated; and

an updating unit which updates the stored template value in response to the determination results which are generated by the examination unit.

6. The biometric identification apparatus of claim 1, wherein the periodic signal extraction unit generates magnitude-normalized signals by normalizing the magnitude of the periodic signals which are extracted, and the template calculation unit calculates the template value using the magnitude-normalized signals.

7. The biometric identification apparatus of claim 6 further comprising:

an examination unit which determines whether a living body corresponding to the template value read out from the template storage unit is identical to a living body to be authenticated; and

an authentication unit which authenticates the input bio signal as a bio signal which is detected from the living body to be authenticated in response to the determination results which are generated by the examination unit.

8. A biometric identification apparatus comprising:

a periodic signal extraction unit which extracts one or more periodic signals from an input bio signal;

a template calculation unit which calculates a template value using the periodic signals which are extracted;

a template storage unit which stores a plurality of template values which correspond to a plurality of living bodies; and

a reading unit which reads the template value that is most approximate to the template value which is calculated by the template calculation unit from the template storage unit,

wherein the periodic signal extraction unit generates temporally normalized signals by computing a Fourier series of each of the periodic signals which are extracted and then computing an inverse Fourier series of the Fourier series of each of the periodic signals which are extracted.

9. The biometric identification apparatus of claim 8 further comprising an analog-to-digital conversion unit which converts the temporally normalized signals which are generated into digital signals through a sampling operation,

wherein the template calculation unit calculates the template value using the digital signals, and the digital signals each comprises an identical number of samples.

10. The biometric identification apparatus of claim 9, wherein the template values which are stored in the template storage unit are calculated using the following equations:

W_i=_i,j^T(t)·

_i,j(t)

_i,j(t)=[X_i,j(1),X_i,j(2),X_i,j(3), . . . ,X_i,j(N₂)]

X_i,j(t)=IFS(F(f_i,j(t)))

where W_i is a template value for an i-th living body which is stored in the template storage unit, i is an integer between 1 and p (where p is a positive integer greater than

     1), j indicates the extracted period signal and is an integer between 1 and m (where m is a positive integer greater than

     1), F indicates a Fourier series function, IFS indicates an inverse Fourier series function, X is a temporally normalized signal, N₂ indicates a total number of samples included in the digital signal into which the temporally normalized signal X is converted, is a template vector, and T is a transpose.

11. The biometric identification apparatus of claim 10, wherein the template calculation unit calculates the template value for a k-th living body corresponding to the input bio signal using the following equation:

W_k,j=_k,j1^T(t)·

_k,j2(t)

where j1 and j2 are values to which j can be set and are determined in advance.

12. The biometric identification apparatus of claim 11, wherein the reading unit reads the template value corresponding to k from the template storage unit that minimizes a magnitude of a vector (e_k) calculated using the following equation:

e_k=_k,j1(t)·

(W_i−

W_k,j2).

13. The biometric identification apparatus of claim 8, wherein the template calculation unit calculates the template value using the temporally normalized values which are generated, the temporally normalized signals which are generated each having an identical temporal width.

14. A biometric identification apparatus comprising:

a periodic signal extraction unit which extracts one or more periodic signals from an input bio signal;

a template calculation unit which calculates a template value using the periodic signals which are extracted;

a template storage unit which stores a plurality of template values which correspond to a plurality of living bodies;

a reading unit which reads the template value that is most approximate to the template value which is calculated by the template calculation unit from the template storage unit;

an examination unit which determines whether a living body corresponding to the template value which is read out from the template storage unit is identical to a living body to be authenticated; and

an updating unit which updates the stored template value in response to the determination results which are generated by the examination unit,

wherein the updating unit updates the stored template value using the following equations;

W_i^new=a ·

W_i^old+b·

W_k,j

_i,j^new(t)=c·

_i,j^old(t)+d·

_k,j(t)

where W^new and ^new are an updated template value and an updated template vector, respectively, W^old and ^old are a template value yet to be updated and a template vector yet to be updated, respectively, a+b =1, and c+d=1.

15. The biometric identification apparatus of claim 14, wherein the reading unit reads the template value corresponding to k from the template storage unit that minimizes a magnitude of a vector (e_k) is calculated using the following equation:

e_k=_k,j1(t)·

(W_i^new−

W_k,j2).

16. A biometric identification apparatus comprising:

a periodic signal extraction unit which extracts one or more periodic signals from an input bio signal;

a template calculation unit which calculates a template value using the periodic signals which are extracted;

a template storage unit which stores a plurality of template values which correspond to a plurality of living bodies; and

a reading unit which reads the template value that is most approximate to the template value which s calculated by the template calculation unit from the template storage unit;

an examination unit which determines whether a living body corresponding to the template value read out from the template storage unit is identical to a living body to be authenticated; and

an authentication unit which authenticates the input bio signal which is detected from the living body to be authenticated in response to the determination results which are generated by the examination unit,

wherein the periodic signal extraction unit generates magnitude-normalized signals by normalizing the magnitude of the periodic signals which are extracted, and the template calculation unit calculates the template value using the magnitude-normalized signals, and

wherein the template values stored in the template storage unit are calculated using the following equations;

W_i=(α

_i·

_i,j^T(t))·

(α

_i·

_i,j(t))

_i,j(t)=[X_i,j(1),X_i,j(2),X_i,j(3), . . . ,X_i,j(N₂)]

X_i,j(t)=IFS(F(f_i,j(t)))

where

${\mathrm{\alpha <br><br>}}_i=\frac{A}{{\mathrm{\aleph <br><br>}}_{i,j<br><br>}^{\max }},$

 

_i,j(t) is a template vector, _i,j^max is a greatest element among a plurality of elements of the template vector, W_i is a template value of an i-th living body stored in the template storage unit, i is an integer between 1 and p (where p is a positive integer greater than

     1), j indicates the extracted period signal and is an integer between 1 and m (where m is a positive integer value greater than

     1), F indicates a Fourier series function, IFS indicates an inverse Fourier series function, X is a temporally normalized signal, N₂ indicates a total number of samples included in the digital signal into which the temporally normalized signal X is converted, and T is a transpose.

17. The biometric identification apparatus of claim 16, wherein the template calculation unit calculates the template value for a k-th living body using the following equation:

W_k,j=(α

_k·

_k,j1^T(t))·

(α

<sub>k·

k,j2</sub>(t))

where

${\mathrm{\alpha <br><br>}}_k=\frac{A}{B}$

 

(where A is a predetermined constant and B is a maximum magnitude of the input bio signal), and j1 and j2 are values to which j can be set and are determined in advance.

18. The biometric identification apparatus of claim 17, wherein the reading unit reads the template value corresponding to k from the template storage unit that minimizes a magnitude of a vector (e_k) calculated using the following equation:

e_k=(α

_k·

_k,j1(t))·

(W_i−

W_k,j2).

19. A biometric identification method comprising:

by a sensor, sensing a bio signal of a living body;

using a processor (or computer) to perform the steps of;

extracting one or more periodic signals from the bio signal;

calculating a template value using the extracted periodic signals, the template comprising a connection strength between neurons of an artificial neural network corresponding to the bio signal; and

reading a template value that is most approximate to the template value which is calculated from among a plurality of prepared template values which correspond to a plurality of living bodies and identifying a living body which corresponds to the template value which is read as a living body which corresponds to the bio signal.

20. The biometric identification method of claim 19 further comprising:

determining whether the living body which corresponds to the template value which is read is identical to a living body to be authenticated; and

authenticating the given bio signal as a bio signal which is detected from the living body to be authenticated if the living body which corresponds to the read template value is determined to be identical to the living body to be authenticated.

21. The biometric identification method of claim 19 further comprising:

determining whether the living body which corresponds to the template value which is read is identical to a living body to be authenticated; and

updating the prepared template values which correspond to the living bodies if the living body which corresponds to the template value which is read is determined to be identical to the living body to be authenticated.

22. The biometric identification method of claim 19, wherein the extracting of the periodic signals comprises generating magnitude-normalized signals by normalizing the magnitudes of the extracted periodic signals, and the calculating of the template value comprises calculating the template value using the magnitude-normalized signals.

23. A biometric identification method comprising:

by a sensor, sensing a bio signal of a living body;

using a processor (or computer) to perform the steps of;

extracting one or more periodic signals from a the bio signal;

calculating a template value using the extracted periodic signals; and

reading a template value that is most approximate to the template value which is calculated from among a plurality of prepared template values which correspond to a plurality of living bodies and identifying a living body which corresponds to the template value which is read as a living body which corresponds to the bio signal,

wherein the extracting of the periodic signals comprises generating temporally normalized signals by computing a Fourier series of each of the periodic signals and then computing an inverse Fourier series of the Fourier series of each of the periodic signals.

24. The biometric identification method of claim 23, wherein the extracting of the periodic signals comprises converting the temporally normalized signals into digital signals through a sampling operation,

wherein the calculating of the template value comprises calculating the template value using the digital signals, and the digital signals comprise an identical number of samples.

25. The biometric identification method of claim 23, wherein the calculating of the template value comprises calculating the template value using the temporally normalized signals, said temporally normalized signals having an identical temporal width.

26. A computer-readable recording medium storing a computer program for executing a biometric identification method, the biometric identification method comprising:

extracting one or more periodic signals from a given bio signal;

calculating a template value using the extracted periodic signals, the template value comprising a connection strength between neurons of an artificial neural network corresponding to the bio signal; and

reading a template value that is most approximate to the template value which is calculated from among a plurality of prepared template values which corresponds to a plurality of living bodies and identifying a living body which corresponds to the template value which is read as a living body which corresponds to the given bio signal.