WO2018212380A1

WO2018212380A1 - Method by which wearable device detects ecg template for personal authentication

Info

Publication number: WO2018212380A1
Application number: PCT/KR2017/005222
Authority: WO
Inventors: 박경원; 전원기; 송병철
Original assignee: 전자부품연구원
Priority date: 2017-05-19
Filing date: 2017-05-19
Publication date: 2018-11-22
Also published as: KR101957019B1; KR20180126949A

Abstract

A method by which a wearable device detects an ECG template for personal authentication is provided. According to an embodiment of the present invention, a biological signal template detection method detects a biological signal template candidate corresponding to a feature by which an individual is identified, calculates a peak value and location of the detected biological signal template candidate, and adds the biological signal template candidate as a biological signal template when the location of the biological signal template candidate and a location of the biological signal template satisfy a specific condition, and the peak value of the biological signal template candidate and a peak value of the biological signal template satisfy a specific condition. Accordingly, a stable ECG template can be extracted.

Description

ECG template detection method of wearable device for personal authentication

The present invention relates to a method for detecting an ECG template, which is a feature that distinguishes individuals from a wearable device that performs personal authentication using an ECG signal, which is a kind of biosignal.

Personal authentication methods using biometric information, such as iris and fingerprint recognition, have been studied so far, and recently, they are being developed as a core technology of Fintech, such as being used for financial transactions centering on smartphones. However, conventional techniques based on biometric information, such as iris and fingerprint recognition, have the disadvantage of easily forgery by others. In particular, fingerprints are vulnerable to theft because they can be obtained in various ways without the consent of the individual, and even in iris recognition, the basic principle is to photograph the iris of the subject.

As a method to compensate for these problems, methods using biosignals such as ECG, EEG, and EMG and skin information of individuals have been actively studied recently. In particular, in the case of ECG, each person has a different shape of the heart, and accordingly, research has confirmed that the ECG signal generated by the heart also varies from person to person, and furthermore, the ECG signal can be used to distinguish individuals with high accuracy. The findings were published.

ECG-based personal authentication can be divided into on-line and off-line methods. The online method performs signal processing while acquiring ECG samples and obtains the final ECG template for authentication. Since the signal processing is performed on a per-sample basis, it has an advantage of easily responding to signal changes, but has a detection threshold. There is a disadvantage in that the convergence of parameters for detection is slow. On the other hand, the offline method has the advantage that it is easy to grasp the characteristics of the entire signal by acquiring the ECG samples in advance and storing them in a memory as much as the length (time) to be processed, so that the determination of the detection parameters can be performed quickly. This has the advantage, but it requires a larger amount of memory than the online approach.

1 is a block diagram illustrating a QRS detector for typical ECG template acquisition.

ECG signals are classified based on the inflection points designated P, Q, R, S, and T. Among them, the R peak expresses a large value compared to other waves so that it can be distinguished by the naked eye. The interval between the R peak and the next R peak determines the heart rate.

As shown in FIG. 1, the ECG template is extracted around the R peak through pre-treatment and post-treatment. As the post-processing method, the Pan-Tompkins algorithm based on the first derivative filter is mainly used.

2 illustrates a process of detecting an R peak. The Pan-Tompkins algorithm can be said that the R peak exists in the region between the point where the threshold crosses upward and the point which crosses downward, as shown in the following figure. Find. After searching for the R peak, an ECG template including P, Q, R, S, and T waves is detected around the R peak.

Since the peak detection method is basically based on the threshold, it is sensitive to the setting of the threshold. In the case of the off-line method, the threshold for determining the threshold is generally the output of the moving average value, the total sample of x _ma [n], and the average value for N.

[Equation 1]

Where α represents a weight.

However, the threshold-based detection method of Equation 1 preferentially detects a large signal. Therefore, when the acquired signal is distorted due to movement or the like during measurement as shown in FIGS. 3 and 4, the normal R peak is not detected, and abnormal peaks are detected, and thus, detection fails. In the registration process, a signal of 20 seconds or more is usually acquired after processing to reduce noise. If a motion occurs during this 20 seconds, the detection may be affected.

Unlike a medical ECG system that adheres and fixes wet electrodes to the skin, wearable devices adopting a method of briefly contacting dry electrodes only with a finger when needed may cause unstable contact, and may cause large noise even in fine movements. have.

For wearable devices equipped with gyro sensors and acceleration sensors, you can consider how these sensors detect motion and do not use the data acquired at the moment the motion is detected. Since the contact of the finger without this sensor is unstable, a large performance improvement cannot be achieved.

The present invention has been made to solve the above problems, an object of the present invention, in the wearable device for performing personal authentication using an ECG signal which is a kind of bio-signal, a signal distorted to extract a stable ECG template The present invention provides a method for detecting and removing a region.

According to an embodiment of the present invention, a biosignal template detection method includes: detecting a biosignal template candidate corresponding to a feature for distinguishing an individual; Calculating peak values and positions of the detected biosignal template candidates; And when the position of the biosignal template candidate and the position of the biosignal template satisfy the first condition and the peak value of the biosignal template candidate and the peak value of the biosignal template satisfy the second condition, It includes; adding as a template.

The first condition is LL _p > T _i , where L is the position of the biosignal template candidate, L _p is the position of the biosignal template, T _i is the critical interval, and the second condition is abs ( RR _p ) <βR _p , where R is the peak value of the biosignal template candidate, R _p is the peak value of the biosignal template, and may be 0 <β <1.

The method may further include removing the biosignal template candidate if the first condition or the second condition is not satisfied.

In addition, the detecting step, the calculating step, and the additional step may be repeated until the number of biosignal templates becomes a predetermined number.

The biosignal template may be a candidate biosignal template having a median peak value among the plurality of candidate biosignal templates.

In addition, the plurality of candidate biosignal templates may be candidate biosignal templates that are less than a predetermined threshold interval apart from an adjacent candidate biosignal template.

In addition, the method for detecting a biosignal template according to an embodiment of the present invention includes: setting a temporary threshold; If the interval between the indexes intersecting the temporary threshold upwards and the indexes intersecting downwards is greater than a predetermined first interval, adding a region between the indexes to the distortion region; and the detecting step includes: The biosignal template candidate may be detected in the excluded region.

If the interval between the index of the first region crossing the temporary threshold downward and the index of the second region crossing the temporary threshold upward is less than the predetermined second interval, the first region and the second region Can be considered an area.

Further, the setting step and the additional step may be repeated if there is no distortion area to be added, if the moving average value is within a specific error with the previously calculated moving average value in the area excluding the distortion area, or if the maximum repetition number is repeated. You can stop.

On the other hand, according to another embodiment of the present invention, the personal authentication device detects a biosignal template candidate corresponding to a feature for distinguishing individuals, calculates a peak value and a position of the detected biosignal template candidate, and calculates a biosignal template If the position of the candidate and the position of the biosignal template satisfy the first condition and the peak value of the biosignal template candidate and the peak value of the biosignal template satisfy the second condition, the biosignal template candidate is added to the biosignal template. Detection unit; And an operation unit configured to perform personal authentication using the biosignal templates detected by the detection unit.

On the other hand, in the biosignal template detection method according to another embodiment of the present invention, the position of the biosignal template candidate and the position of the biosignal template satisfy the first condition, and the peak value of the biosignal template candidate and the biosignal template If the peak value satisfies the second condition, adding the biosignal template candidate to the biosignal template; And if the first condition or the second condition is not satisfied, removing the biosignal template candidate.

On the other hand, in the personal authentication device according to another embodiment of the present invention, the position of the biosignal template candidate and the position of the biosignal template satisfy the first condition, and the peak value of the biosignal template candidate and the peak value of the biosignal template A detector for adding a biosignal template candidate as a biosignal template if the second condition is satisfied and removing the biosignal template candidate if the first condition or the second condition is not satisfied; And an operation unit configured to perform personal authentication using the biosignal templates detected by the detection unit.

As described above, according to embodiments of the present invention, when the wearable device is commercialized, it is possible to provide an authentication means with low risk of theft, and may be applied to an electronic employee card, a door lock, a financial field, and the like.

In addition, according to the embodiments of the present invention, the risk of theft when used in combination with a relatively high theft risk, but provides a superior performance, such as the existing fingerprint and iris has the advantage of lowering the probability of false positives.

In addition, according to embodiments of the present invention, stable performance may be provided even in an embedded system in which a computational capability such as a wearable device is limited.

1 is a typical QRS detector block diagram,

2 is an R peak detection method;

3 is a diagram illustrating instantaneous noise causing a detection failure;

4 is a diagram illustrating a region distortion causing a detection failure;

5 is a flowchart provided to explain the instantaneous noise removing method according to an embodiment of the present invention;

6 is a view provided to explain the distortion area detection method;

7 is a signal after the distortion region is removed by FIG. 6,

8 is a flowchart of a distortion area detection method according to another embodiment of the present invention;

9 is a block diagram of a wearable device for personal authentication based on ECG according to another embodiment of the present invention.

Hereinafter, with reference to the drawings will be described the present invention in more detail.

An embodiment of the present invention provides a method of effectively detecting and removing signal distortion due to unstable contact when an ECG template is acquired and used for personal authentication in a system that acquires an ECG signal by using a dry electrode such as a wearable device. .

As a result of the analysis through the actual ECG measurement, the distortion of the ECG signal due to movement can be classified into the instantaneous noise generated with a very narrow width and the noise with a wide distortion of a specific region as described above.

The embodiment of the present invention proposes a method capable of detecting and removing both noise sources alone and in combination.

1. How to remove instant noise

5 is a flowchart provided to explain an instantaneous noise removing method according to an embodiment of the present invention.

Instantaneous noise affects only a small part of the observation range, so it has a negligible effect on the overall mean value. Therefore, since the threshold value change due to the instantaneous noise is small, the threshold value of Equation 1 can be used.

However, the R peak is detected based on the threshold value, and the ECG template is not immediately acquired through this, and it is designated as a candidate template first, and whether the template is a normal signal or a signal distorted by noise is determined. The determination is made based on the peak value.

In the method according to the embodiment of the present invention, since the ECG template detected immediately before is assumed to be a normal signal, in order to apply the method according to the embodiment of the present invention, the first template must be a normal ECG template. To ensure this, the following initialization process is performed.

1.1 Initialization Steps:

For initialization, as shown in FIG. 5, first, three candidate ECG templates are detected and stored as a, b, and c templates (S105). The R peak interval between two adjacent templates is measured and defined as t _ab , t _bc (S110). If both t _ab and t _bc are greater than T _i (S115-Yes), the process moves to the next step (S125).

If is less man t _ab (S115-No), discards a template to detect the new ECG template and la d (S120), c by measuring the interval and the template (S110), T _i If greater than (S115-Yes), go to the next step (S125). If it is still small (S115-No), the process of detecting a new ECG template again and comparing it with the c template is repeated.

On the other hand, if t _bc is smaller than T _i (S115-No), the c template is discarded and the newly detected ECG template is named c and the interval is compared with the b template (S120, S110, S115). If it is still small (S115-No), the process of detecting a new ECG template again and comparing it with the c template is repeated.

On the other hand, if t t _ab and _bc are both less than T _i (S115-No), all discarded, detecting the three ECG template again and compare the distance back (S120, S110, S115).

Thereafter, the R peaks of the three templates for which the interval verification has been completed are compared. To this end, first, the median value of the three templates is calculated (S125), and the peak larger than β times larger than the median value is removed (S130). Next, the peak value and the peak position of the median template are defined as R _p and L _p (S135).

1.2 ECG template detection steps:

The new template candidate is detected and the peak values and positions are defined as R and L, respectively (S140). Only when LL _p > T _i and abs (RR _p ) <βR _p (S145-Yes), it is acquired with a normal template and updated with Lp = L, R = max (R, R _p ) (S150). If either condition is not met (S145-No), the candidate is removed. This process is repeated until the required number of templates N _T is obtained (S155).

Where T _i is the minimum spacing between two adjacent R peaks, and β is a value between 0 and 1 as the ratio of the allowable difference of R peaks. T _i = 360ms and β = 0.5 can be used as initial default values, and different optimal values can be calculated and applied according to the system and environment.

2. How to detect and remove specific area distortion

When a signal is distorted by noise over a wide area as shown in FIG. 4, it is necessary to remove an area distorted by noise because it affects the determination of the threshold value of Equation 1. To this end, first, a temporary threshold value is set based on the threshold value of Equation 1.

[Equation 2]

Where γ means a positive real number greater than one. As shown in FIG. 6, when D = L _d -L _u is larger than a predetermined D _T when defining the index L _u which intersects this temporary threshold upward and the cross index L _d downward, respectively. Determining that it is an area of interest and adding between L _u and L _d to the removal list.

As shown in the first distortion region of FIG. 6, when the interval at which the lower intersection crosses upward again is smaller than D _i , it is regarded as one region. The optimal D _T and D _i can be determined through experiments, but empirically, D _i ≥800 μs, D _T ≥400 μs can be set.

If O distortion areas are detected, the area is represented by a set of integer values in the start index and the end index.

[Equation 3]

Here, δ is an element that can add a margin to the detected distortion region and is expressed as an integer greater than or equal to zero.

Except for the distortion area, the newly calculated average value is as follows.

[Equation 4]

Where | S | represents the number of elements in the set S. The threshold calculated newly through Equation 4 is as shown in FIG. 7.

The above method may be extended by sequentially canceling the distortion area. 8 is a flowchart illustrating a method for detecting a repeated distortion region.

In order to eliminate the repetition, the threshold calculation of Equation 1 is modified as follows. In the first calculation, the distortion area is initialized to an empty set of S = {} (S210).

[Equation 5]

In FIG. 8, an initial threshold value is set based on an initial value of S = {} (S220 and S230). In this case, since the average value of all the samples is taken, Equation 1 is the same. Next, after setting the temporary threshold of Equation 2 (S240), the distortion area of Equation 3 is detected by detecting the distortion area (S250).

Again, the distortion area is detected in the area except the detected area based on the temporary threshold and added to Equation 3. This process no longer has any additional distortion areas detected (

= 0), if the average value calculated in Equation 4 is within a specific error T _eps with the previously calculated average value (S260-No), the repetition is stopped. If the stop condition is not satisfied (S260-Yes), the above process is repeated.

Accordingly, the distortion region is removed again through Equation 4, and the average value is newly reset, and the temporary threshold value is updated again through Equations 5 and 2 (S220 to S250). If the stop condition is satisfied (S260-No), the algorithm is stopped, and if it does not converge to the stop condition (S260-Yes), only the number of times set as the maximum repetition number (K _max ) is repeated and forcedly terminated (S280).

The above algorithm finally outputs the distortion area set S. Detection thresholds can be set for regions other than this region, and ECG templates can be extracted in the same manner as in the prior art.

3. How to remove instant noise and distortion

The instantaneous noise elimination method deteriorates performance when a large area is distorted, and the instantaneous noise having a narrow generation range cannot be detected in the distortion area elimination method. Therefore, in order to detect and remove both noise sources, two methods according to an embodiment of the present invention must be combined.

Specifically, after removing a wide range of distortion areas through the detection of the distortion areas shown in FIG. 8, the instantaneous noise and the residual distortion areas are removed through the instantaneous noise removal method shown in FIG. 5.

4. Wearable device for personal authentication based on ECG

9 is a block diagram of a wearable device according to still another embodiment of the present invention. The wearable device according to an exemplary embodiment of the present invention includes an input unit 110, a signal processor 120, a detector 130, a calculator 140, a storage 150, and a communicator 160.

The input unit 110 receives the ECG signal and applies it to the signal processor 120. The signal processor 120 performs filtering on the applied ECG signal and removes the distortion region from the ECG signal according to the algorithm shown in FIG. 8. The detector 130 detects ECG templates according to the algorithm shown in FIG. 5.

The calculator 140 registers the ECG templates detected by the detector 130 in the storage 150 or transmits the ECG templates to a server (not shown) through the communicator 160. In addition, the calculator 140 compares the ECG templates detected by the detector 130 with ECG templates stored in the storage 150 or transmits them to the server through the communication unit 160 to perform a personal authentication procedure.

5. Modifications

Up to now, the present invention has been described in detail with reference to a preferred embodiment of a method for robustly detecting an ECG template in a signal in which instantaneous noise and a specific region are distorted.

Specifically, in order to reliably detect an ECG template that distinguishes an individual from a wearable device that performs personal authentication using an ECG signal, a gyro sensor or an instantaneous noise and distortion region that distorts a threshold value and fails to detect the ECG template. A method of improving ECG template detection performance by detecting and removing from an ECG signal itself without using an acceleration sensor is proposed.

The ECG mentioned in the above embodiment is an example of a biosignal. Therefore, the technical idea of the present invention can be applied to replacing ECG with EEG, EMG as well as other types of bio signals.

Furthermore, the wearable device is also only one example for convenience of description. Of course, other types of devices other than wearable devices are included in the scope of the present invention.

On the other hand, the technical idea of the present invention can be applied to a computer-readable recording medium containing a computer program for performing the functions of the apparatus and method according to the present embodiment. In addition, the technical idea according to various embodiments of the present disclosure may be implemented in the form of computer readable codes recorded on a computer readable recording medium. The computer-readable recording medium can be any data storage device that can be read by a computer and can store data. For example, the computer-readable recording medium may be a ROM, a RAM, a CD-ROM, a magnetic tape, a floppy disk, an optical disk, a hard disk drive, or the like. In addition, the computer-readable code or program stored in the computer-readable recording medium may be transmitted through a network connected between the computers.

In addition, although the preferred embodiment of the present invention has been shown and described above, the present invention is not limited to the specific embodiments described above, but the technical field to which the invention belongs without departing from the spirit of the invention claimed in the claims. Of course, various modifications can be made by those skilled in the art, and these modifications should not be individually understood from the technical spirit or the prospect of the present invention.

Claims

Detecting a biosignal template candidate corresponding to a feature for distinguishing an individual;

Calculating peak values and positions of the detected biosignal template candidates; And

If the position of the biosignal template candidate and the position of the biosignal template satisfy the first condition and the peak value of the biosignal template candidate and the peak value of the biosignal template satisfy the second condition, the biosignal template candidate is selected as the biosignal template. Adding to the; biological signal template detection method comprising a.
The method according to claim 1,

The first condition is,

LL p > T i ,

Where L is the position of the biosignal template candidate, L p is the position of the biosignal template, T i is the critical interval,

The second condition is,

abs (RR p ) <βR p ,

Wherein R is a peak value of the biosignal template candidate, R p is a peak value of the biosignal template, and 0 <β <1.
The method according to claim 1,

And removing the biosignal template candidate if the first condition or the second condition is not satisfied.
The method according to claim 1,

The detection step, the calculation step and the additional step are

And repeating until the number of biosignal templates is a predetermined number.
The method according to claim 1,

Biosignal template,

A biosignal template detection method, characterized in that the candidate biosignal template has a median peak value among the plurality of candidate biosignal templates.
The method according to claim 5,

Multiple candidate biosignal templates

12. A method of detecting a biosignal template, wherein the candidate biosignal templates are adjacent candidate biosignal templates and candidate biosignal templates below a predetermined threshold interval.
The method according to claim 1,

Setting a temporary threshold;

If the interval between the indexes intersecting the temporary threshold upwards and the indexes intersecting downwards is greater than the first predetermined interval, adding a region between the indices to the distortion region;

The detection step is

And detecting a biosignal template candidate in a region excluding a distortion region.
The method according to claim 7,

If the interval between the index of the first region that crosses the temporary threshold downward and the index of the second region that crosses the temporary threshold upward is less than the predetermined second interval, the first region and the second region as one region A biosignal template detection method, characterized in that it is considered.
The method according to claim 7,

The setup and additional steps

If no distortion area is added, the biosignal template comprises stopping the repetition when the moving average value is within a specific error from a previously calculated moving average value in a region except the distortion area, or when the maximum repetition number is repeated. Detection method.
Detecting a biosignal template candidate corresponding to a feature for distinguishing an individual, calculating a peak value and a position of the detected biosignal template candidate, and the position of the biosignal template candidate and the position of the biosignal template satisfy the first condition A detector configured to add the biosignal template candidate to the biosignal template when the peak value of the biosignal template candidate and the peak value of the biosignal template satisfy the second condition; And

And a computing unit configured to perform personal authentication using the biosignal templates detected by the detecting unit.
If the position of the biosignal template candidate and the position of the biosignal template satisfy the first condition and the peak value of the biosignal template candidate and the peak value of the biosignal template satisfy the second condition, the biosignal template candidate is selected as the biosignal template. Adding to the; And

If the first condition or the second condition is not satisfied, removing the biosignal template candidate.
If the position of the biosignal template candidate and the position of the biosignal template satisfy the first condition and the peak value of the biosignal template candidate and the peak value of the biosignal template satisfy the second condition, the biosignal template candidate is selected as the biosignal template. A detection unit configured to add to and remove the biosignal template candidate if the first condition or the second condition is not satisfied; And

And a computing unit configured to perform personal authentication using the biosignal templates detected by the detecting unit.