KR20070085857A - Method and apparatus for electro-biometric identity recognition - Google Patents

Abstract

A method and apparatus for electro-biometric identity recognition or verification that produces and stores a first biometric signature that identifies a specific individual by forming the difference between a representation of the heartbeat pattern of the specific individual and a stored representation of common features of the heartbeat patterns of a plurality of individuals; after the producing step, the method and apparatus obtains a representation of the heartbeat pattern of a selected individual and produces a second biometric signature by forming the difference between the heartbeat pattern of the selected individual and the stored representation of common features of the heartbeat patterns of the plurality of individuals; it then compares the second biometric signature with the first biometric signature to determine whether the selected individual is the specific individual. The apparatus and method may be employed as a stand-alone unit or as part of another device pursuant to the many applications described herein.

Description

METHOD AND APPARATUS FOR ELECTRO-BIOMETRIC IDENTITY RECOGNITION}

Cross Reference of Related Application

This application claims priority to PCT Application No. PCT / IB04 / 03899, filed November 8, 2004, which PCT Application No. PCT / US2003 / 23016, filed July 24, 2003, And US Patent Application No. 60 / 398,832, filed Jul. 29, 2002, which is a priority, and are incorporated herein by reference in their entirety.

Identity Recognition includes, but is not limited to, automatic banking services, e-commerce, e-banking, e-investing, e-investing, e-data protection, remote access to resources, electronic It plays an important role in many aspects of life, including e-transaction, work security, anti-theft devices, criminal identification, security access and entry registration in the workplace.

Often, computerized systems use passwords and personal identification numbers (PINs) for user recognition. However, in order to maintain security, the password must be changed regularly, which places a significant burden on the user. Similarly, signature verification methods also have other drawbacks, including forgery and false fraud. See, for example, US Pat. No. 5,892,824 to Beatson et al.

As a result, identity recognition systems using measurements of biological phenomena of individuals-biometrics-have recently grown. When used alone or integrated with other technologies such as smart cards, encryption keys and digital signatures, biometrics is expected to penetrate almost every aspect of the economy and our daily lives.

Several advanced technologies have been developed for biometric identification, including fingerprint recognition, retina and iris recognition, face recognition, and voice recognition. For example, US Pat. No. 5,534,855 to Shockley et al. Generally describes the use of biometric data, such as fingerprints, to grant computer access to an individual. US Pat. No. 6,490,680 to Schheidt et al. Describes identity authentication using biometric data. US Patent No. 6,310,966 to Dulude et al. Describes the use of fingerprints, hand geometry, iris and retinal scans, and voice patterns as part of a biometric authentication certificate. US Pat. No. 6,483,929 to Murakami et al. Generally describes "physiological and histological markers" including infrared radiation for biometric authentication. However, these types of technologies have only penetrated into a limited market due to complex and inconvenient acquisition methods, sensitivity to environmental parameters (such as lighting conditions and background noise), and high prices. In addition, due to the complex acquisition procedure, the above techniques usually require operator intervention.

Fingerprint recognition is the established and most mature technology of this group. However, fingerprint recognition has some disadvantages. That is, the fingerprint recognition system cannot verify the physical presence of the fingerprint owner, and thus is foolish and limits its suitability for online applications, and because optical sensors are expensive and brittle devices, they are generally unsuitable for the consumer market. Has a negative sense related to criminology.

Retinal scan technology features high performance. However, this technology requires high precision optical sensors and is inconvenient to use because they require adjustment of the head posture and operate on a very sensitive organ, the human eye. This optical sensor is also expensive and brittle.

The iris and face recognition system is an easy-to-use technology because the system is far away from recording images and not bothersome. However, the system requires digital photography equipment and is sensitive to lighting conditions, pupil size variations and facial expressions. In addition, iris recognition performance is degraded by the use of dark glasses and contact lenses, and face recognition can be tricked by makeup.

Speech recognition is the most convenient technology in this group, but it requires a low noise environment and is very sensitive to inherently variable speech parameters, including intonation. In addition, existing conventional recording techniques can be used to trick speech-based recognition systems.

Thus, there is a need for identity recognition techniques that can be used independently or integrated with existing security systems, are reliable, stable (online and offline), difficult to cheat, inexpensive and easy to use.

Over the years, electrocardiogram (ECG) measurements have been used for many different purposes. The ECG signal is an electrical signal generated by the heart and can be measured using conventional surface electrodes, usually installed on the subject's chest. The ECG signal consists of several components that represent different functional stages during each heartbeat and are projected according to the electrical orientation of the developing tissue.

Subjects differ in their electro-heart signal due to typical fluctuations in heart tissue structure, heart orientation and electrical tissue orientation, all of which affect the electro-heart signal measured from the limbs. Inherent details are shown. Many types of systems take advantage of these subject-specific variations.

For example, US Pat. No. 6,293,904 to Blazey et al. Describes the use of ECG signals to assess or profile an individual's physiological and cognitive status. In relation to identification , 2001 by Kyoso et al. At the 23rd Annual International IEEE Conference on Engineering in Medicine and Biology Society (host: Istanbul, Turkey) under the " Development of an ECG Identification System ". The conference paper compares the patient's ECG with previously registered ECG feature parameters for identification. US Patent Application No. 2003013509 to Wiederhold proposes to identify subjects using ECG signals obtained directly or remotely, and to "investigate" feature extraction to identify individuals and to perform "preliminary analysis" of such methods. to provide.

However, an ECG signal is composed of ECG components having features that may be common to a group. None of these references describe a system or method for removing common features of ECG components to generate signatures for subject identification. Thus, there is still a need for systems and methods with these properties to identify individuals.

Inclusion of these references in the background of the invention does not constitute an admission that these documents are prior art or similar techniques relating to the invention described herein. However, all references in the Background section of the invention are incorporated as if fully set forth herein by reference to the present invention.

Applicants provide solutions to the above problems of biometric identification in various devices and methods having several aspects.

In a first aspect, Applicant solves each of the above mentioned problems of biometric identification using the following method and variations thereof, which method comprises:

Generating and storing a first biometric signature that identifies the particular individual by forming a difference between the representation of the specific heartbeat pattern of the particular individual and a stored representation of common features of the plurality of individual heartbeat patterns;

After said generating, a second biometric signature is obtained by acquiring a representation of the heartbeat pattern of the selected individual and forming a difference between the stored representation of a common feature of the heartbeat pattern of the selected individual and the plurality of individual heartbeat patterns Generating, and

Comparing the second biometric signature with the first biometric signature to determine whether the selected entity is the particular entity.

The system according to this aspect includes an ECG signal acquisition module, an ECG signal processing module including an ECG signature generator, and an output module.

Thus, according to this first aspect, the systems and methods disclosed herein convert a bio-electric signal into a unique electro-biometric signature. The uniqueness of the electro-heart signature makes the system very difficult to deceive, and the inherent stability of the method makes it ideal for local as well as remote and online applications. In addition, bio-signature-based systems feature high recognition performance and support both open and closed search modes.

In one preferred method according to the first aspect, a stored representation of a common feature of one or more ECG components is obtained by measuring and storing such representations for a plurality of individuals and then averaging all of the stored representations. As another alternative, this common feature may be obtained through techniques such as principal component analysis, fuzzy clustering analysis, wavelet decomposition, and the like.

Since the electro-heart method according to this first aspect is stable, they have other important advantages, that is, they enable an easy and simple acquisition technique that can be implemented as a low cost, easy-to-use acquisition device and also provides a skilled operator It makes it unnecessary.

According to variations of these systems and methods, the common features of one or more of the subject's ECG components may be removed using an analytical model of the common features of the one or more ECG components instead of or in addition to the use of empirical models. Similarly, this common feature first classifies stored representations into subgroups, identifies common features in at least one group, classifies subject signals according to subgroups, and common features of one or more of the ECG components of the subgroups. Can be removed by generating the subject signature by removing the from the subject signal and identifying the subject by calculating the subject signature correlation for the subgroup's signature.

The common feature may be determined by averaging synchronized electrocardiograms from a group of individuals and subtracted from the subject's electrocardiogram to determine the subject's signature. However, this method assumes that common features are constant across a group of individuals. Indeed, some common features exist more or less in any given entity. Therefore, it is better to approximate the common features to best suit the given electrocardiogram before removing the common features to obtain the subject's signature. This technique provides a more accurate determination of the subject's signature.

According to this method, a group of electrocardiograms can be divided (decomposed) into a series of characteristic waveforms. The characteristic waveforms representing the common features of the group are then weighted to best approximate the range of common features present in the subject's ECG. This approximation is then subtracted from the subject's ECG. The remaining includes the subject's ECG signature.

Multiple templates may also be maintained for each subject, such as by storing multiple signatures generated by the subject at different pulse rates. In this embodiment, the subject signature may then be correlated with a suitable template, such as a template for an appropriate pulse rate. Thus, in one variation, the systems and methods disclosed herein may use multiple signature templates to identify individuals over a range of environments and responses. As another alternative, or in addition, according to the first aspect, the subject signal and the registered signal can also be normalized based on the pulse rate.

According to a second aspect described herein, the identification process can set a dynamic threshold. This dynamic threshold may be based on the desired confidence level in the identification, such as the level determined by the confidence score.

According to a third aspect disclosed herein, the systems and methods disclosed herein can use a "Q-factor" to determine whether to reduce signal contamination due to noise. Likewise, a Q-factor or other quality of signal measurement can be used to determine the length of a subject sample needed to identify the subject to the desired confidence level. Registering the sample may also be used so that a sample with the desired confidence level may be suitable for future comparison.

In an alternative embodiment to the "Q-factor" calculation, the systems and methods disclosed herein calculate a standard deviation in a subject signature and / or a registered signature due to noise and from these calculations the signal quality is appropriate for identification. Can be determined.

Likewise, the systems and methods disclosed herein can determine signal quality by measuring the impedance of a contact or probe. Signal quality measurements according to this aspect may also be used to inform the subject to adjust their contact with the sensor or probe or their position relative to the sensor or probe.

According to a fourth aspect, the subject and database signature can be encrypted as a safety precaution against illegal access and use of the signature.

According to the fifth aspect, an ECG signal can be obtained with an electrode placed in contact with any body part that produces a consistent signal. For some body positions, even a slight change in electrode placement can cause a drastic change in the shape of the received signal and even cause other signal components to appear or disappear. Thus, according to this aspect, the methods and systems disclosed herein can use an electrode placement site that produces a stable subject-specific consistent signal despite changes in electrode placement within that site. These sites include arms and legs (including fingers and toes). The robustness of electrode placement within these sites is due to the constant electro-heart signal projection that does not change as long as the electrode is close to limb extremity.

According to this same fifth aspect, any sensing probe known as an ultra-high impedance sensing probe can also be used to acquire signals, including signals from a single body point, such as a fingertip. . As another alternative or in addition, these ultra-high impedance probes can remotely detect electro-heart signals, thereby maintaining signal coherence while eliminating the difficulty of electrode placement.

According to a sixth aspect, the systems and methods disclosed herein may include components and steps that protect against enrollment fraud and reduce the likelihood that a database registrant may misrepresent his identity.

According to a seventh aspect, the systems and methods disclosed herein can identify a subject by comparing the subject's match score with the database registrant's match score.

According to an eighth aspect, the systems and methods disclosed herein can use a weighted correlation technique that assigns different weights to different electro-heart signal components to generate a signature. As another alternative or in addition, the signature may be normalized using various metrics, including root-mean-square calculations or L1 metrics.

Some biometric technologies use challenge-response protocols to ensure that the user data they receive is not outdated. As such, these techniques can reduce the risk of the system being deceived by the reproduction of biometric data. However, to date, the challenge-response mechanism of the biometric system requires active participation of the user. Active user participation complicates and takes a long time for the user verification process. For example, speech recognition systems generally require a user to repeat a randomly selected word or sentence. Thus, according to another aspect, the biometric identification system can reduce the risk of being deceived by beneficially utilizing a biological-challenge-response mechanism that does not require a conscious response from the user.

The system and method in accordance with each of the above aspects preferably performs its task automatically for identity recognition. In addition, these systems and methods may be included in a wide variety of devices and systems. Some non-limiting examples are: Smart card, passport, driver's license device, bio-logon identification device, personal digital assistant (PDA), cellular embedded identification device, anti-theft device, ECG monitoring device, e-banking device, E-transaction devices, pet identification devices, physical access devices, logical access devices, and devices that combine ECG and fingerprint monitoring, blood pressure monitoring, and / or any other form of biometric device.

In addition, the systems and methods disclosed herein may be used to compare the width of a subject's QRS complex, or more generally the width of a subject's QRS-related signature components with the width of a registered group or analytical ECG model, and the like. This may be used to identify a person's age.

In other applications, the systems and methods herein may be used to identify a person undergoing drug treatment, such as by registering and calculating, or analytically deriving a series of drug-related signature templates. This method can also be used to identify or catch a subject attempting to trick the system by using a drug to alter its ECG signal.

Other applications include building and indoor access control, surveillance system access, wireless device access, control and user verification, mobile phone activation, computer access control (including via laptop, PC, mouse, and / or keyboard), data access (Document control, etc.), passenger identification in public transport, elevator access control, firearm locking, vehicle control systems (including startup dialing and door locks), smart card access control and smart card credit check, online Access to materials (including copyrighted works), electronic ticket releases, access and control of nuclear material, robot control, aircraft access and control (passenger identity, flight control, maintenance worker access), vending machine access and control Laundry and dryer access and control, locker access, child locks, television and / or video access control, decryption key access and Usage, cashless vending machines, vending machine maintenance access, game console access (including online trading functions), computer network security (including network access and control), point of sale buyer identification, online transactions (customer identification and accounts) Access), the use of the systems and methods disclosed herein for cash payment service or wireline transfer identification, building maintenance access and control, and implanted medical device programming control. Other applications will be apparent to those skilled in the art and are within the scope of the present invention.

For any application, an apparatus in accordance with any or all of the above aspects may operate continuously or on demand. The device may be configured to obtain a representation of the heartbeat pattern of the selected subject by having one or more electrodes or sensors away from the subject in contact with the subject. If the device is provided with a smart card, it can be activated for a limited period of time after a successful recognition and then deactivated until the next successful recognition is performed. The device may be configured to operate with an encryption key or digital signature.

With respect to the method disclosed herein, the steps of the method described above may be performed sequentially or in any other order. The systems and methods disclosed herein can be used in human or other animal subjects.

Each of these aspects can be used in permutation and combination with each other. Further embodiments, as well as variations, modifications and improvements, are also described herein.

1 is a simplified block diagram of a system for use in aspects disclosed herein, consisting of a signal acquisition module, a signal processing module, and an output module.

2 is a block diagram of one embodiment of a signal acquisition module of the system of FIG.

3 is a block diagram of one embodiment of a signal processing module of the system of FIG.

4 shows the contribution of the first six most potent PCs extracted from the pool of 100 subjects and the first 10 PCs to the representation of data variation.

5 shows its respective signature configured by eliminating an optimal combination of the original ECG signal and the three most potent PCs and their delayed shifted versions.

6 is a diagram showing the grand-average electro-heart signal waveform calculated from a database of 20 subjects.

FIG. 7 shows a group of electro-heart signal waveforms of 10 of the subjects participating in the database and contributing to the average waveform of FIG. 6.

8 illustrates a group of electro-biosignal waveforms or templates, derived from the signal waveforms of FIG.

9 is a scatter plot and distribution histogram of the sign-maintained squared correlation values of 20 subjects who contributed to the total mean waveform of FIG. 6.

FIG. 10 is a table of z-scores based on desired confidence levels for identification cut-off. FIG.

11A shows a correlation distribution.

11B shows the distribution of Z-transformed correlations.

12 shows an identification performance curve (static).

Fig. 13 shows the identification performance curve (dynamic).

14 shows signal quality as a function of NSR.

FIG. 15 is a plot of match score distribution for a 5 second segment as a function of signal quality. FIG.

FIG. 16 is a plot of match score distribution for a 20 second segment as a function of signal quality. FIG.

17 shows the match score (for Q = 0.8) as a function of the recording period.

18 shows the match score (for Q = 0.5) as a function of the recording period.

19 shows the functional elements of a good system.

20 illustrates the functional elements of a good signal processor.

Justice

Unless stated otherwise, the terms “identify”, “identify” and “identify” include the concepts of “verify identity”, “verify identity” and “verification of identity” respectively.

"Closed search" means a search in which a single stored signature is checked to verify the identity of an entity.

"Open search" means a search in which a plurality of stored signatures are searched to identify a subject.

1st aspect:

According to a first aspect, a bio-electrical signal is acquired, processed and analyzed to identify an individual's identity. Preferred embodiments of the system and method according to this first aspect are, for example, shown in FIG. 1. 1 shows a system called an Electro-Biometric IDentification (E-BioID) system. In this preferred embodiment, the stored representation of common features of one or more ECG components of the plurality of individuals is the average of one or more ECG components of those individuals. However, other embodiments may use other types of stored representations of common features, such as may be achieved by principal component analysis, fuzzy clustering analysis, or wavelet decomposition, or provided by an analytical model.

In a preferred embodiment, the basic components of the E-BioID system include a signal acquisition module 12, a signal processing module 14, and an output module 16, which are implemented within a single housing. In another preferred embodiment, the system can provide telemetry of locally acquired electro-biological signals. Each of the components shown in FIG. 1 may be easily implemented by those skilled in the art based on principles and techniques known in the art in connection with the present invention.

2 shows a preferred configuration of the signal acquisition module 12 in the E-BioID system. This data acquisition module preferably includes one or more sensors 22, preamplifiers 24, band-pass filters 26, and analog-to-digital (A / D) converters 28. Each of these components can be readily implemented by those skilled in the art based on principles and techniques known in the art with respect to the present invention.

Sensor 22 may be of any type capable of detecting a heartbeat pattern. For example, the sensor may be a metal plate sensor that is an "add-on" to a standard computer keyboard. According to another aspect, a single sensor may, by itself, obtain a signal from a single point of contact, such as by touching a finger, and as another alternative, the sensor may not need to contact the subject at all.

3 shows a preferred component of the signal processing module 14 in an Ei-BioID system. This signal processing module preferably includes a digital signal processor (DSP) 32, dual port RAM (DPR) 34, electrically erasable programmable read only memory (E ² PROM) 36 and an I / O port ( 38). Each of these components can be readily implemented by those skilled in the art based on principles and techniques known in the art with respect to the present invention. Signal processing module 14 is connected to signal acquisition module 12 and output module 16 via port 38.

In an alternative embodiment, the signal processing module may be implemented with suitable programming on a personal computer, which is a flexible computing platform, thereby facilitating integrating the system into existing computing facilities in a home, office, or institutional / corporate environment. Will be.

The output module 16 preferably consists of a dedicated display unit, such as an LCD or CRT monitor, and may comprise a relay for activation of an external electrical device such as a locking mechanism. As another alternative, the output module may include a communication line for relaying the recognition result to the remote point for further action.

Signal Acquisition, Processing, and Analysis

The bioelectrical signal or heart rate signal is obtained in a simple manner, where the subject is instructed to contact the at least one sensor 22 for a few seconds. One or more sensors, which may be metal plates, conduct the bioelectrical signal to amplifier 24, which amplifies the bioelectrical signal to a desired voltage range. In a preferred embodiment, this voltage range is 0-5 volts.

The amplified signal is passed through filter 26 to remove contributions outside the desired frequency range of 4 Hz-40 Hz. As another alternative, a wider range of 0.1 Hz-100 Hz can be used in conjunction with the notch filter to eliminate major frequency interference (50/60 Hz). Digitization of the signal is preferably performed at a sampling frequency of about 250 Hz, preferably in a 12-bit A / D converter 28.

In module 14, the signal is generally normalized by the 'R' peak magnitude to correspond to signal magnitude variations that are generally related to exogenous electrical properties. The normalized signal is converted into an electro-biosignal that is compared with a pre-stored electro-biosignal template. The result of this comparison is quantified, optionally assigned a confidence value, and sent to output module 16, which provides recognition feedback to the user of the E-BioID system and also provides external devices such as locks or siren. , A virtual device, such as a network login confirmation, or a communication link can be activated.

As another alternative, or in addition, this signal can be normalized to pulse rate. This is useful because the electro-heart signal is affected by changes in the pulse rate, a known electro-cardiologic modifier. Pulse rate changes can cause changes in delay, amplitude and shape of the 'P' and 'T' components relative to the 'QRS' component of the electro-heart signal (these components are shown in FIG. 7). However, pulse rate changes can be compensated for automatically by retrospective, pulse rate-driven adjustment of the signal complex. In addition, the adaptive mode of operation of the system can track and compensate for pulse rate induced changes. This can be done by compressing or extending the time scale of one cycle of the heartbeat waveform. More complex formulas can be used that describe the relationship between waveform characteristics (eg, S-T, P-Q segment duration) and pulse rate. Thus, the method according to this variant may be based on electro-heart signal discrimination, in which the analysis is performed in synchronization with the heartbeat to remove features common to the general population and thus form an electro-bio or biosignature. Enhances inherent features (typically not detectable in primitive electro-heart signals).

In another embodiment, the E-BioID system is implemented as a fully integrated compact device, where many of the functional elements are implemented on an ASIC based system.

In another embodiment, the device may be included in a watch worn on the wrist, where the signal is measured between the wrist of the hand with the watch filled and the other hand of the wearer. The back side of the watch may consist of a conductive medium (eg, a metal plate) in contact with the back of the wrist, and the front of the watch may be provided with another metal contact which needs to be in contact with the fingers of the other hand. Can be. The watch may transmit a signal indicative of the identity of its wearer and / or activating a physically or logically locked device, such as a door, computer, safe, or the like. The watch can also transmit individual information about its wearer. In the same way, the device can be included in any other garment article including a belt or conductive medium. The belt or other garment item may then transmit a signal indicative of the identification of its wearer and / or activate a physically or logically locked device and / or transmit individual information about its wearer.

Principle of operation

Biometrics requires comparing the newly acquired biometric signature with a signature template in a recorded or registered biometric signature template database. This requires two stages of system operation: registration and recognition.

Registration step

In a preferred embodiment, each new subject is instructed to touch the first sensor with the finger of the left hand and at the same time touch another sensor with his right hand. In an alternative embodiment, the subject may touch the sensor, which is generally made of metal, with other parts of the body, preferably with hands or feet. In other embodiments, the subject may touch a single sensor with a single body point. As another alternative, the subject may not need to touch the sensor at all. The system monitors the subject's pulse rate, initiates recording and preferably continues for at least 20 seconds. Shorter intervals may be used depending on the level of accuracy required. Once recording is complete, the system may perform a self-check to verify signature consistency by comparing at least two biometric signatures derived from two portions of the registered segment. These two parts may be two half or two larger overlapping segments. These two parts can be used to derive two biometric signatures. If the self-test results are successful, the subject's registration is complete, and if not successful, the procedure is repeated. This successful record is used for the construction of an electro-heart signal or a series of electro-heart signals, which are added to the electro-heart signal database.

The electro-heart signal is then converted into a series of electro-biological signature templates by removing features common to all or a subset of the subjects participating in the dataset to enhance the subject-specific discrimination features.

In a preferred embodiment, the system produces a grand-average electro-heart template, which is synchronous averaging of normalized electro-heart signals from the entire pool of subjects. Is calculated. The total mean represents the common feature described above, so subtracting the total mean from each signal of the electro-heart signals yields a series of different subject-specific electro-bio template signatures. In alternative embodiments, other means of removing common features such as principal component analysis, fuzzy clustering analysis, or wavelet decomposition may be used.

In a better embodiment, a group of electrocardiograms can be divided (decomposed) into a series of characteristic waveforms. According to this preferred embodiment, noise is removed from the ECG of a group of individuals. The system can use Principal Component Analysis (PCA) to resolve a group's ECG into a series of orthogonal (uncorrelated) components. These uncorrelated components, when summed, represent 100% of the signal's total energy, ie signal variance.

First principal component-the component associated with the largest eigenvalue of the PCA representation. Usually, the first three to five components, and, anyway, the first less than ten components of the group's electrocardiogram, generally represent nearly 90% of the energy or dispersion of the electrocardiogram and include common features. Notably, these first components are present throughout the general human population and exhibit stable common characteristics. As a result, these first principal components can be used to identify the signature of any human subject and need not be recalculated for each subject. The smaller components (typically 10% of the total waveform energy) represent some individual information and noise in the group.

A characteristic waveform representing the common characteristics of the group is then subtracted from the subject's ECG. Remaining includes the subject's ECG signature and any remaining noise.

Characteristic waveforms may be generated in other ways and may depend on the desired "distance" or "overlapping" between each waveform. For example, the correlation function can preferably be used to determine the desired distance between the waveforms, but other methods are possible.

Notably, when an electrocardiogram is obtained from an individual who does not participate in a registered dataset, its ECG signature is usually determined with reference only to the first three to four PCA components and their time-shifted versions of the registered dataset. It is possible to do that.

Determination of signature

The electrocardiogram of all subjects contains more or less of each of the first principal components. According to this preferred embodiment, the electrocardiogram of the subject can be approximated using the principal component from the sample set according to the following equation.

In this equation, C _i is the reconstruction coefficient, p is the model order, and PC is the principal component. The goal is to find coefficients that weight the database principal components for the best approximation of the subject's electrocardiogram. In other words, the goal is to minimize the error between the original subject signature and the approximation of the subject signal constructed by weighting the principal components of the database.

This can be done by various methods. One method is to use the least squares approximation to determine the reconstruction coefficients to minimize the norm of the reconstruction error. This is shown below.

또는 오차 =

Or error =

Once the optimal coefficients are determined, these coefficients can be used to sum the first principal components (top three or four, etc.) of the database according to the following equation.

합산

Summing

This summation is then subtracted from the subject signal. What remains is the subject signature and possibly some noise.

In addition, since noise is uncorrelated by definition, this noise is usually described by the last principal component-the principal component associated with the smallest eigenvalue. As a result, the noise can be selectively removed from the subject signal by weighting these last principal components and then removing them from the subject signal to an optimal approximation to the subject signature. Noise can also be removed by other methods.

Responding to Latency Changes

Part of the variation in the ECG component database is due to latency changes, ie time variance in registered data signatures. As a result, the method can be improved by time shifting the principal component, preferably both to the left and to the right. For example, if three principal components are used to approximate common ECG characteristics, six more components may be added to counter latency variations-two for each component moved left and right Moved to-.

In this example, three principal components and six time shifted components are used to calculate the construction coefficients. Once the best composition factor is determined, common feature components are constructed and subtracted from the original subject ECG signature to yield the individual signature.

Signature =

4 shows the contribution to the representation of data variance of the first six most potent PCs and the first ten PCs extracted from a pool of 100 subjects. 5 shows its respective signature configured by removing the optimal combination of the original ECG signal and the three most potent PCs and their delayed shifted versions.

Although PCA is a stable algorithm that provides a progressively potent component representation with an apparent magnitude distinction between the main signal, secondary variation, and noise, at least two alternative techniques may be used to resolve the group's ECG. In a first alternative embodiment, independent component analysis (ICA) may be used to resolve the synthesized signal into independent components (as opposed to orthogonal components of the PCA). These independent components can then be used for modeling and reconstruction of the ECG in a manner similar to PCA.

In a second alternative embodiment, wavelet decomposition (WD) may be used to decompose the synthesized signal into a series of time-scaled waveforms called wavelets. WD is based on transient wavelet waveforms, unlike Fourier decomposition (based on continuous sine and cosine decomposition). As a result, WD has an advantage over Fourier analysis in that wavelets are more efficient descriptors of transient signal components such as electrocardiograms.

As another alternative or in addition, common features may be removed using analytical models for common features of one or more ECG components, rather than using empirical models calculated from registered data.

In another preferred embodiment, the database is partitioned into several subsets in a manner that improves intra-subset similarity and inter-subset disparity. This embodiment then calculates a separate total mean or other common feature determination for one or more of the subsets. This database partitioning itself may be performed using standard pattern classification schemes such as linear classifiers, Bayesian classifiers, fuzzy classifiers, or neural networks. For large databases, it is useful to simplify and shorten the retrieval process as well as partition the database into subsets to ensure the validity of the total mean as appropriately representing the similarity between the electro-heart signals. The subject signature can then be generated by removing common features found in the appropriate subgroup.

6 shows an example of the total mean constructed from a pool of 20 subjects participating in the database.

FIG. 7 shows ten examples of electro-heart signals, and FIG. 8 shows electro-bio template signatures derived from the electro-heart signals by removing features common to all subjects included in the database. Specifically, each signature of FIG. 8 is obtained by subtracting the waveform of FIG. 6 from the corresponding signal of FIG. It will be observed that while the original electro-heart signal is very similar, the resulting electro-biological signatures are distinctly different. These differences have been found to reflect the inherent inherent electro-heart differences that underlie the recognition capabilities of E-BioID systems.

Recognition stage

In the recognition phase, the subject interacts with the system in a manner similar to the registration phase, but shorter recording times of a few seconds are sufficient.

In a preferred embodiment, the system executes a verification procedure (closed search), which processes the acquired signal and is found in a partitioned subgroup of the database or by an analytical ECG model, which is found in the entire database. Eliminating the common features provided forms an electro-biographical subject signature, adjusts the signature according to the pulse rate, and compares the adjusted electro-biological signature with the subject's registered electro-biological signature template.

In another preferred embodiment, the system executes an identification procedure (open search), and the system repeats the comparison process for the entire database or for a divided subgroup of databases, thereby providing identification of matching identities.

Comparison process

과 전기-생체 서명 템플릿

간의 상관 계수

의 계산에 의해 수행된다.In a preferred embodiment, this comparison is an electro-biological signature as follows:

And electricity-bio sign template

Correlation coefficient between

Is performed by the calculation of.

. 대체 실시예에서, 이 비교는 전기-생체 서명들 간의 RMS 오차 등의 다른 유사성 척도에 기초할 수 있다.This correlation coefficient is squared while maintaining its original sign, i.e.

. In alternative embodiments, this comparison may be based on other similarity measures, such as RMS error between electro-biological signatures.

은 인식 결정을 행하는 데 사용되며, 사전 설정된 임계치보다 큰 값은 긍정적 식별, 즉 일치로서 간주되고, 어느쪽이라고 결정하기 어려운 임계치 근방의 값은 연장된 또는 반복된 기록을 필요로 함을 나타낼 수 있다. 개방형 검색 모드에서, 모든 부호-유지된 상관 계수 제곱 중에서 가장 큰 부호-유지된 상관 계수 제곱은, 가장 높은 계수가 선택된 임계치를 넘는다면, 가장 유망한 피험자 식별을 가져온다.This comparison may yield one or several correlation coefficients depending on the mode of operation, namely closed search or open search. In closed search mode, the signed correlation coefficient squared

Is used to make a recognition decision, a value greater than a preset threshold is considered as a positive identification, i.e. a match, and a value near the threshold that is difficult to determine which may indicate that an extended or repeated recording is required. . In the open search mode, the largest sign-maintained correlation coefficient square of all sign-maintained correlation coefficient squares results in the most promising subject identification if the highest coefficient exceeds the selected threshold.

The preset threshold is derived from the required confidence level, and higher desired confidence levels require higher thresholds. In one embodiment, sign-maintained correlation squares greater than 0.8 indicate a match and values lower than 0.7 indicate a mismatch. Thus, sign-maintained correlation squares higher than 0.8 may be considered as true matches and values lower than 0.7 may be considered as inconsistencies.

The upper diagram of FIG. 9 shows a scatter plot of the sign-sustained correlation squared, with a 0.8 threshold indicated by a dotted line. There is a clear distinction between consensus (circle) and disagreement (star). The histograms in the other two diagrams represent the strong recognition capabilities of the E-BioID system, where inconsistencies are concentrated near zero values (no correlation), while matches are densely distributed near 1.0. (Absolute correlation).

In alternative embodiments, more complex decision methods, such as multi-parameter methods (eg, fuzzy logic methods) using two or more distance measures, may be used, for example analyzing data in which multiple correlation values are segmented. Can be derived from

In a preferred embodiment, the system improves its performance over time by adding an electro-heart signal to the subject's database file when there is a change in the signal. In subsequent recognition, the system processes the newly acquired signal, calculates the pulse rate, forms an electro-bio subject signature, selects a registered electro-bio signature template with the most similar pulse rate, and selects a new electro-bio The signature is compared with the selected registered electro-biological signature template.

In another preferred embodiment, the system uses the signals acquired during long-term system operation to track possible variations in registered subject electro-heart signals, and if consistent changes occur, the registered signals are subject to these changes. It is adjusted automatically to reflect this. This tracking process compensates for gradual changes in electro-heart signals over long periods of time, but does not compensate for rapid and severe changes such as those expected with respect to clinical heart conditions. In other embodiments, such severe changes may be reported to the subject, indicating a need for medical consultation.

2nd sun:

Biometric identification methods benefit from the proper determination of an identification threshold. The identification threshold may be derived from a correlation analysis between the candidate signature and the registered database signature. This threshold can be determined using the distribution of empirical data to achieve optimal identification performance. However, the fixed threshold implicitly assumes a deterministic signature and stationary noise, but in practice the signature is variable and the noise relies largely on unpredictable external influences. Thus, biometric identification methods, including the method according to the first aspect, may be adversely affected by signal and noise variations in databases and test measurements. In general, this will result in a reduced correlation for both matches and disagreements.

Thus, according to the second aspect, biometric identification methods and systems, including the method and system according to the first aspect, can use dynamic thresholds that can compensate for the effects of signal variations and noise interference. This aspect yields a dynamic data-dependent identification threshold. In a preferred embodiment, the dynamic threshold is recalculated in each identification attempt using a kinetic approach to normalize the correlation data, thereby allowing the calculation of a quantifiable and statistically significant identification threshold. This threshold shows that it is immune to variable signal and noise conditions.

The preferred method according to this second aspect is based on the determination of a confidence limit for correlation-based scoring between a test signature and a series of registered signatures. These ECG signatures can be determined empirically, but they can also be synthetic, in which case no background database is needed in the biomatch process. The synthetic ECG signature can be generated by using a random set of reconstruction coefficients in the PCA-based ECG model. As another alternative, the set of reconstruction coefficients may be derived according to a set of rules extracted from the distribution of real-life reconstruction coefficients derived from real subjects.

In any case, the confidence limit describes the upper and lower bounds on the value of the problem, given the statistical confidence. The two-tailed limit describes both the upper and lower bounds, while the one-tailed limit knows that there is no lower or upper limit to the value of the variable. Only the upper cutoff or the lower cutoff is described. If the variable under consideration satisfies some statistical criterion appropriate for each statistical method, the confidence limits can be determined statistically in several different ways.

Most statistical methods rely on the values of normal distribution variables, ie, follow a species-shaped Gaussian distribution. Normal distribution variables are well characterized statistically, and their statistical limits can be determined in a simple manner based on the mean and the variation of the variables.

If the variable is not normally distributed, a normalizing transformation can be used to convert the original variable into a new variable that is normally distributed, and thus can be used to determine the confidence limit. The appropriate mathematical transformation can be determined using statistical considerations or by empirical examination of a sufficiently large dataset. In order to express the confidence limits as original variables, back-transformation is also required.

Signal cross-correlation analysis can be used for a matching procedure. Values range from -1 [absolute negative correlation] to 0 (no correlation) through +1 [absolute positive correlation]. In general, a fairly positive correlation represents almost probable true identification, so the upper confidence limit should be used to describe the dynamic identification threshold.

By definition, the correlation is a bounded variable and therefore not normally distributed. Mathematical transformations are needed to normalize the correlation distribution to enable determination of upper confidence limits. As another alternative, empirical techniques that do not rely on this transformation can be used.

The preferred method described in more detail below is particularly suitable for correlation analysis. It is based on the Fisher Z transformation that transforms the correlation into a normal distribution variable.

Another method may use correlation squares. Since raw correlation is not additive, the mean or other statistical function of the correlation has no statistical significance. Correlation squares are additive, but they are also not normally distributed, thus requiring additional transformations. If the preprocessing of the correlation changes the distribution of its variables, additional transformations may be needed to counter these changes. These additional transformations include, but are not limited to, logarithmic, square, square root, and transcendental functions.

Another method requires some prior empirical testing, which is good if a large number of candidates are correlated to a large database. The likelihood of identification errors can be determined directly by inspection of this database, or an appropriate conversion can be determined empirically. However, since this method is not dynamic and must be performed before the actual test, the effects of the test conditions cannot be easily compensated for and require the development of a mathematical model for the effects of noise.

The preferred method according to this second aspect, the Fischer-transformation method, requires a transformation of the correlation between the candidate signature and the registered signature to obtain a distribution of scores that are more or less normally distributed. As noted above, data that satisfies the assumption of normality may be used to derive the parameter confidence limit.

The Fischer Z transform is designed to normalize the correlation. This transformation can be expressed as

Z _f = arctanh (r)

Where Z _f is a transformed value, arctanh is a hyperbolic arc tangent function, and r is a correlation. arctanh must be expressed in radians.

Once all correlations have been transformed, the one-tailed confidence limit for the transformed score can be determined by obtaining the mean of all transformed correlations except the candidate correlations and the standard deviation of all transformed correlations and calculating the following: have.

Confidence bound = tanh (Z _{f mean} + z * sd _zf )

Here, z is the normal distribution 'z score', z _{f mean} is the mean of transformed correlations with the database, _zf sd is the standard deviation of the transformed correlations with the database.

Lowercase z here refers to the value of a normal distribution z-score derived based on the desired confidence in the cut-off. A table of these scores is provided in FIG. 10.

In the table of FIG. 10, the standard deviation is multiplied by the appropriate z-score, added to the mean, and the total amount is inversely transformed into a correlation by obtaining the hyperbolic tangent.

For example, a 95% confidence limit can be determined using a z score of 1.65. Thus, if the mean of the transformed values is 0.05 and the standard deviation is 0.25, the 95% confidence limit is 0.72. That is, correlation values above 0.72 only happen by chance at less than 5%.

The reverse procedure is used to determine the probability that any particular candidate identification is due to random chance. Solve for z-score,

z = (Z _fc -Z _{f mean} ) / sd _zf

Where z is the normal distribution 'z score' , z _fc is the transformed candidate correlation, z _{f mean} is the _mean of the transformed correlation with the database, and sd _zf is the standard deviation of the transformed correlation with the database.

The resulting z-score can be converted to a one-tailed probability value by interpolation, if necessary, also by referring to a table of cumulative normal distributions. For example, referring to the abbreviated table above, a z-score of 1.80 implies a 3,75% probability that the candidate will be accidentally highly correlated.

As mentioned above, if the noise in the registered signature or in the candidate signature is random, this reduces the overall correlation with the candidate value. If there is a true identification, that identification thus has a lower correlation with the candidate. Note that the variability of the primitive correlation increases as the primitive value decreases because the high primitive correlation is less variable due to the ceiling effect of the maximum correlation of 1 and this is compensated by the transform. Thus, a dynamic threshold with the desired certainty can be recalculated at each identification attempt using the methods described above. Importantly, overall random noise still tends to push all correlations to zero, reducing the overall true variability, thus lowering the confidence limit, but true matching unless the signal-to-noise ratio falls below some limit. (true match) is still meaningful.

The following example of the second aspect is based on the 38-subject data base. All subjects are healthy individuals and volunteer to participate in this study.

Example 1: Normalization of correlation

By correlating all pairs in the database, a series of 703 cross correlations were obtained. The raw z-transformed correlation distribution is provided in FIG. 11. While the primitive correlations are not normally distributed (upper), the transformed correlations appear to represent nearly normal distributions (lower).

Example 2: performance

A biometric identification method was implemented using 38 registered signatures and 38 test signatures. 12 shows the FAR and FRR performance curves as a function of the static threshold, and FIG. 13 shows the performance curves as a function of the dynamic threshold. Clearly, the dynamic threshold gives a fairly good result (e.g., EER _Static = 3%, EER _Dynamic = 0%).

3rd sun:

As noted above, the dynamic identification threshold is a data-driven threshold and is preferably recalculated in each identification session to establish a confidence limit and to specify the statistical significance of the identification process. However, the overall score still decreases with deterioration of signal quality due to background noise, lowering the dynamic threshold and thereby reducing identification confidence. This problem requires evaluation of the signal quality at both the registration and identification stages to facilitate high performance recognition.

The third aspect solves this problem by calculating the Q value (a kind of signal quality index). Signal quality index Q is a quantitative description of the quality of an ECG signature. This is based on the analysis of the random error in two or more ECG groups, derived with reference to its signal average ECG.

The Q value can be used to verify signal quality during the registration and identification phase, which ensures proper system performance. If the Q factor is lower than required by the predefined threshold (which itself is based on the desired level of identification confidence), the measurement can be extended or repeated until the confidence requirement is met.

One preferred method is to derive Q in a series of steps.

(1) The input ECG signal is segmented into ECG groups consisting of conventional wave morphology features (e.g., P, Q-R-S, T elements).

(2) The ECG group is aligned (“time synchronized”) with respect to the R wave peak.

(3) The average ECG is derived from the aligned ECG group. The preferred method is to find the arithmetic mean, but other methods such as harmonic mean, geometric mean, weighted mean, or median may be used. Other alternatives include converting the original signal by other methods such as principal component analysis.

(4) Each original ECG group is processed for the average ECG so that some difference is derived for the average ECG. The preferred method is to perform the subtraction, ie, the original ECG minus the mean ECG, but other methods (eg, dividing the original ECG by the average ECG) can be used. If the mean ECG is a stable true representation of the subject's ECG, the resulting difference is a representation of the intrinsic noise (ECG noise) in each individual ECG group.

(5) The corresponding sample points in time are processed together over each ECG noise complex to derive a measure of variability. The best way is to determine the variance. Other measures that can be used include standard deviations or ranges of variation.

(6) The average of these variability measures is obtained. The best way is to find the arithmetic mean. Alternative methods may include finding an average after the transformation (eg, log) or alternative means (geometric mean, harmonic mean, median). Other summary scores such as maximum values may also be used.

Note that the signal can be normalized prior to analysis, and the mean itself can be used as the Q index, because the mean is directly related to the SNR. Alternatively, various other scaling transformations can be applied to the mean to convert the mean to a branch with the desired minimum, maximum, and linearity characteristics.

Example 1: According to the third aspect Q (signal quality) vs. NSR ( noise  Signal ratio)

으로 표시된다. 시간 n에서의 모든 점에 대해, 오차항

을 계산하고, 그의 분산은

으로 표시 한다. 변동성의 평균을 0 내지 1 범위로 변환하는 양호한 스케일링 변환은 다음과 같이 정의된다.When X represents an ECG data matrix, each row representing one ECG group can be represented by x _i (n), where i is the exponent of the ECG group and n represents discrete time units. The average of all ECG groups is

Is displayed. For all points at time n, error term

, And its variance is

Marked with. A good scaling transform that transforms the average of variability into the range 0-1 is defined as follows.

The simulation shown in FIG. 14 shows the usefulness of using the Q factor to evaluate signal to noise level. This simulation uses real-life ECG recording with increasing levels of Gaussian white noise added to the signal. 14 provides the Q value as a function of noise to signal ratio (NSR). If Q begins to decrease from its plateau, it can be seen that Q decreases monotonically with increasing NSR until the ECG alignment procedure is stopped (NSR -35 dB, Q-0.2).

Example 2: Score as a Function of Signal Quality According to Third Aspect

In theory, a match score close to 1 would indicate a positive match, while a mismatch score should tend to be 0, indicating no correlation at all. In practice, however, the true match score is affected by transient variations in the ECG signature and more heavily from background noise. Therefore, higher signal quality is required for short time high score identification. It will be appreciated that a high quality signal increases the upper boundary with respect to the coincidence score but does not affect the lower boundary, which depends on cardiac signature variability. 15 and 16 show the distribution of scores as a function of signal quality, based on a database of 38 subjects. 15 shows short data segments of 5 seconds each. In contrast, Figure 16 shows a longer segment of 20 seconds each (Figure 16). Obviously, for longer segments, the effect of noise is compensated to some extent and the score distribution is flattened.

Example 3: Signal quality and recording period according to the third aspect

Signal quality can be quantified using Q parameters. For smaller Q values, longer recordings are needed to maintain some level of statistical significance, unless Q falls below some limit at which the ECG alignment process is interrupted. 17 and 18 show an increase in the identification score as a function of the record length for a given Q value.

Thus, according to this third aspect, the methods and systems disclosed herein can calculate signal quality using Q-factors or other measures and allow the system to search for samples with reduced noise or to Q-factor or other Longer samples can be taken based on signal quality measures and desired identification reliability.

4th sun:

According to a fourth aspect, the methods and systems disclosed herein can encrypt a stored signature. This safety feature is designed to prevent misuse of the data in the database, although the various methods and systems herein generally operate on stored signatures rather than raw ECG data. Thus, an added layer of security can be used by encrypting the signature itself. For that purpose, various scrambling techniques can be used, including the public key infrastructure (PKI) technique used for credit card data. This fourth aspect makes it more difficult for an improper use of the data of a registered subject because an unauthorized person must encrypt the signature and still convert the signature back to a raw data signal, which is a common feature. This is impossible without knowing whether it was removed from this raw data. Thus, one advantage of the systems and methods disclosed herein is that they make it extremely difficult for anyone to misuse stored information.

5th sun:

Biometric identification systems are generally vulnerable to enrollment fraud. The system and method according to this fifth aspect solve this problem by using ECG data from genetically related entities registered in a database. Immediate family members often have ECGs that share common characteristics. By comparing the subject's signature with the general population and / or its registrants known to be associated with it, the system can reliably determine whether the subject is that person as known. This technique can be used in addition to identifying the subject's identity through conventional methods such as photo identification and / or fingerprint matching. However, unlike those methods that are non-Euclidian and not suitable for similarity-based clustering, this technique is fraudulent at any stage of the registration process by determining the probability of a genetic relationship based on the registrant's ECG signature. Can be determined.

6th sun:

The systems and methods described herein can also use ultra-high impedance probes to measure ECG. Because reliability and ease of use are important for ECG-based biometric identification systems, it is beneficial to measure ECG at a single point or even without contacting the subject. Electrical potential probes may operate to improve reliability and ease of use for biometric identification in biometric methods and systems, including the methods and systems described herein. Ultra high impedance probes come in a variety of forms. For example, Electric potential probes-new directions in the remote sensing of the human body, Harland et al., Meas. Sci. Technol. 13 (2002), pages 163-169. Ultra-high input impedance probes according to this aspect preferably have an ultra-low noise characteristic and do not require a current conducting path to operate. As a result, this probe works well with the methods and systems described above even when used by non-experts without the assistance of expert system operators. Thus, these probes can be used in airport-based biometric identification systems, such as by acquiring ECG signals when an individual passes through a scanner (similar to a metal detector) in a suit. Similarly, a single probe can be used to collect ECG from the fingertips of the object, such as in an ATM or game machine. The use of a single probe contact gives the subject more freedom of movement and makes it easier for the subject to follow the identification and registration scheme. This is particularly useful when the biometric identification system described herein is used to control a subject's machine operation, especially when the machine (eg, a firearm or a vehicle) requires physical contact to operate. The single probe and remote probe ECG capture system according to this aspect can be complemented by a noise reduction strategy to reduce body noise and EMG.

7th sun:

According to a seventh aspect, a biometric identification method and system is provided by comparing a match score for a subject (generated by comparing a subject's signature with a database registrant's signature) to a plurality of registrant's match scores (compare the registrant's signature with a database registrant's signature). Generated). Thus, rather than analyzing the distribution of the subject's correlated match scores, this identification technique analyzes the distribution of the correlation of the subject's match scores and the registrant's match scores. As in the fifth aspect, methods and systems according to this aspect are useful for identifying related entities. This is because individuals with a group of registrants have a Gaussian distribution of match scores with a median that is substantially higher than the Gaussian distribution of match scores for individuals not associated with the registrants. Thus, by examining the distribution of coincidence scores, the probability of a genetic relationship with the subject's registrants can be ascertained.

8th Sun:

Finally, as another alternative to or in addition to the correlation techniques described above, the methods and systems described herein may use weighted correlation for identification. According to this aspect, this correlation may assign different weights to various signature differences. For example, signature differences due to QRS group features may be weighted more than signature differences due to T or P group features. The system and method can also use the root mean square root of the signature value as part of a weighting function because T is very variable, QRS is stable, and P is somewhere in between. Thus, the signature can be normalized using a root mean square calculation, L1 metric, or other normalization technique.

Good that can be used in all sun Example :

19 shows a functional diagram of a good system. Similarly, FIG. 20 shows a functional diagram of a good signal processor. The term “processor” is used generically herein and the processing may be performed by physically separate components, such as in a coprocessor on an IC chip, or the processor may comprise a physically unitary unit.

Which can be used in all sun General example Registration algorithm

The following is an exemplary algorithm for the registration step that can be used in any of the above aspects.

이 i번째 새로운 피험자의 20-초, 250Hz 디지털화된 샘플을 나타내는 것으로 하며, 여기서 n은 이산 시간 단위이다.i.

This represents the 20-second, 250 Hz digitized sample of the i-th new subject, where n is a discrete time unit.

은 범위 4 Hz - 40 Hz에서 대역-통과 필터링된다.ii.

Is band-pass filtered in the range 4 Hz-40 Hz.

으로 표시된다.iii. This filtered signal

Is displayed.

은 QRS군에 대해 검색되고, 'R' 피크를 앵커 포인트(anchor point)로서 식별한다.iv. Filtered signal

Is searched for the QRS group and identifies the 'R' peak as an anchor point.

가 유지되거나 반전된다.v. Signal filtered to obtain positive 'R' peak

Is maintained or inverted.

로 되어 있는 평균 맥박수를 확정하기 위해 식별된 QRS군이 카운트된다.vi.

The identified QRS group is counted to confirm the average pulse rate.

가 세그먼트화되고, 각각의 'R' 앵커 포인트 이전의 50개 샘플 및 그 이후의 90개 샘플을 취한다.vii. Signal filtered around anchor point

Is segmented, taking 50 samples before each 'R' anchor point and 90 samples thereafter.

viii. Each data segment is normalized by the amplitude of the 'R' anchor point.

으로 표시된 피험자 전기-심장 신호를 생성하기 위해 세그먼트들이 앵커 포인트를 중심으로 정렬되고 평균된다.ix.

Segments are aligned and averaged around the anchor points to produce a subject electro-heart signal indicated by.

가 평균 맥박수

에 따라 조정된다. 조정된 전기-심장 신호는

으로 표시된다.x. Subject electro-heart signal by normalizing 'P' and 'T' delays according to pulse rate

Average pulse rate

Is adjusted accordingly. The tuned electro-heart signal is

Is displayed.

가 데이터베이스에 추가되고, 총평균

에 도입된다.xi. Electro-heart Signal of Pulse Rate-Adjusted Subject

Is added to the database, the total average

Is introduced.

을 시스템 데이터베이스에 저장된 맥박수 조정된 전기-심장 신호 각각으로부터 차감함으로써 일련의 전기-생체 서명

이 구성된다.xii. Total average

A series of electro-biological signatures by subtracting from each of the pulse rate adjusted electro-heart signals stored in the system database

This is made up.

Example: Recognition Algorithm

The following is an exemplary algorithm for the recognition step.

가 테스트된 피험자의 10-초, 250Hz 디지털화된 샘플을 나타내는 것으로 한다.i.

Denotes a 10-second, 250 Hz digitized sample of the tested subject.

는 범위 4 Hz - 40 Hz에서 대역-통과 필터링된다.ii.

Is band-pass filtered in the range 4 Hz-40 Hz.

로 표시된다.iii. This filtered signal

Is displayed.

가 검색된다.iv. Filtered signal for positions in the QRS group, using P peak as anchor point

Is searched.

가 유지되거나 반전된다.v. Signal filtered to obtain positive 'R' peak

Is maintained or inverted.

로 되어 있는 평균 맥박수를 확정하기 위해 식별된 QRS군이 카운트된다.vi.

The identified QRS group is counted to confirm the average pulse rate.

가 세그먼트화되고, 각각의 'R' 앵커 포인트 이전의 50개 샘플 및 그 이후의 90개 샘플을 취한다.vii. Signal filtered around anchor point

Is segmented, taking 50 samples before each 'R' anchor point and 90 samples thereafter.

으로 표시된 피험자 전기-심장 신호를 생성하기 위해 세그먼트들이 앵커 포인트를 중심으로 정렬되고 평균된다.viii.

Segments are aligned and averaged around the anchor points to produce a subject electro-heart signal indicated by.

가 평균 맥박수

에 따라 정규화된다. 맥박수 조정된 피험자 전기-심장 신호는

으로 표시된다.ix. Subject Electro-Heart Signal

Average pulse rate

Normalized according to Pulse rate adjusted subject electro-heart signal is

Is displayed.

을 맥박수 조정된 전기-심장 신호

로부터 차감함으로써 전기-생체 서명

이 구성된다.x. Total average

Pulse rate adjusted electro-heart signal

Electricity-biosignal by subtracting from

This is made up.

과 모든 등록된 전기-생체 서명

간의 상관 계수가 계산되고 제곱되며, 그의 원래의 산술 부호를 유지한다.xi. Electricity-bio sign

And all registered electricity-biological signatures

The correlation coefficient between is calculated and squared and retains its original arithmetic sign.

xii. The largest sign-maintained correlation squared is selected and compared with a preset threshold.

xiii. If the largest sign-maintained correlation square chosen is greater than a preset threshold, a positive match is indicated and the subject is identified.

Thus, the method and apparatus for obtaining, processing and analyzing an electro-heart signal for electro-biological identification may comprise any subset of the following registration and recognition steps.

Enrollment

Acquisition, Digitization, and Storage of Electro-Heart Signals from Subjects:

a. Formation of Electro-heart Signal Database

b. Partitioning into several subsets of template database based on electro-heart signal similarity

c. The composition of one or more total means

d. Derivation of Subject-unique Biological-Biological Signature

recognition

Verification

The newly captured electro-biosignature is compared with the subject-specific registered electro-biosignature template.

a. Correlation and confidence analysis of newly captured subject bio-biosignatures with their associated stored bio-biosignal templates

b. Display and registration of recognition results and / or activation of physical or virtual local / remote mechanisms

discrimination

The newly captured electro-biosignal is compared with all of the electro-biosigning templates participating in the database.

a. Correlation and confidence analysis of newly captured subject bio-biosignatures with all stored bio-biosignal templates

b. Display and registration of recognition results and / or activation of physical or virtual local / remote mechanisms

In a preferred embodiment, the E-BioID system measures the electrical bio-signal from the human body through the conductive sensor plate. These same plates can be used for bidirectional interactions with the subject's nervous system, for example by inducing a sympathetic skin response in the user with a small amount of electrical stimulation provided through the plate. This bidirectional interaction constitutes a biological challenge-response mechanism that ensures the provision of up-to-date biosignals without requiring active participation of the user in the challenge-response procedure.

Others can easily be modified and / or refined to the embodiments disclosed herein for various applications without departing from the general concept without undue experimentation. Such improvements and modifications should be interpreted within the meaning of the disclosed embodiments and their equivalents. It is to be understood that the phraseology and terminology used herein is for the purpose of description and not of limitation. Means, materials, and steps for carrying out the various disclosed functions may have various alternative forms and still fall within the literal or equivalent scope of the claims.

Accordingly, the expressions "means for ..." and "means for ...", or any method step expression and subsequent functional description, as may be found in the foregoing description and / or in the claims below, Any structural, physical, chemical, or electrical element or structure that may exist now or in the future, performing any of the described functions, whether or not exactly equivalent to the embodiment or embodiments disclosed in the detailed description, or any It is intended to define and encompass method steps, that is, other means or steps may be used that perform the same function, and such expressions should be interpreted broadly.

Claims (123)

As a method for identifying an entity, Generating and storing a first biometric signature identifying the particular individual by forming a difference between the representation of the particular heart's heartbeat pattern and the stored representation of the common features of the plurality of individual's heartbeat patterns; After the generating and storing step, by obtaining a representation of the heartbeat pattern of the selected individual, and forming a difference between the representation of the heartbeat pattern of the selected individual and the stored representation of common features of the heartbeat pattern of the plurality of individuals, Generating a biometric signature; And Comparing the second biometric signature with the first biometric signature to determine if the selected entity is the particular entity Object identification method comprising a. The method of claim 1, The generating and storing step forms a difference between the representation of the heartbeat pattern of each individual entity and the stored representation of the common features of the heartbeat pattern, thereby generating and storing a plurality of first biometric signatures that each identify the individual entity. Including steps And the comparing step is performed for each of the first biometric signatures. The method of claim 2, Obtaining a representation of a plurality of individuals 'heartbeat patterns and deriving and storing a representation of a common feature of the plurality of individuals' heartbeat patterns from at least a selected number of representations. The method of claim 3, Deriving and storing representations of common features of the heartbeat patterns of the plurality of individuals includes deriving and storing representations of common features of the heartbeat patterns, each of the plurality of representations being different from each other of the plurality of entities. Individual identification method derived from group. The method of claim 3, Deriving and storing a representation of a common feature of the plurality of individual heartbeat patterns comprises generating an average of the plurality of individual heartbeat patterns. The method of claim 3, Deriving and storing a representation of a common feature of the heartbeat pattern of the plurality of individuals comprises performing one of principal component analysis or wavelet decomposition. The method of claim 2, The comparing includes correlating the second biometric signature with each of the first biometric signatures, and identifying one of the first biometric signatures most correlated to the second biometric signature. Identification method. The method of claim 7, wherein The correlating step includes obtaining a correlation coefficient associated with each first biometric signature, and the comparing step further includes comparing the correlation coefficient associated with the identified first biometric signature with a correlation coefficient threshold. How to identify the containing object. The method of claim 1, The comparing may include correlating the second biometric signature with the first biometric signature to obtain a correlation coefficient; And comparing the correlation coefficient associated with the identified first biometric signature with a correlation coefficient threshold. The method of claim 1, Generating and storing the first biometric signature includes storing the signature in a local database. The method of claim 1, Generating and storing the first biometric signature comprises storing the signature in a remote database. The method of claim 1, Acquiring a representation of the heartbeat pattern of the selected subject comprises compensating for a deviation in the selected subject's pulse rate from the selected pulse rate. The method of claim 1, Acquiring a representation of the heartbeat pattern of the selected subject comprises acquiring several representations of the heartbeat pattern. The method of claim 1, Generating and storing the first biometric signature of the particular individual includes obtaining a plurality of representations of the heartbeat pattern of the specific individual over a period of time and generating a continuous first biometric signature, wherein the continuous Wherein each of the biometric signatures is derived from an individual representation of a plurality of representations of the heartbeat pattern of the particular subject. Device for identifying an entity, Means for generating and storing a first biometric signature that identifies the particular individual by forming a difference between the representation of the heartbeat pattern of the particular individual and a stored representation of common features of the heartbeat pattern of the plurality of individuals; After the first biometric signature is generated and stored, a representation of the heartbeat pattern of the selected individual is obtained, and a difference between the representation of the heartbeat pattern of the selected individual and the stored representation of the common feature mean of the heartbeat pattern of the plurality of individuals is obtained. Means for generating a second biometric signature by forming; And Means for comparing the second biometric signature with the first biometric signature to determine if the selected entity is the particular entity Object identification device comprising a. The method of claim 15, The means for generating and storing creates and stores a difference between the representation of the heartbeat pattern of each individual entity and the stored representation of the common features of the heartbeat pattern, thereby generating and storing a plurality of first biometric signatures each identifying an individual entity. Means for doing so; And said means for comparing is performed for each of said first biometric signatures. The method of claim 16, Wherein the means for generating and storing comprises means for obtaining a representation of a plurality of individual heartbeat patterns and means for deriving a representation of a common feature of a plurality of individual heartbeat patterns from at least a selected number of the representations. The method of claim 17, And the means for deriving comprises means for deriving stored representations of the plurality of common features, each representation being derived from a different group. The method of claim 16, The means for comparing includes means for correlating the second biometric signature with each of the first biometric signatures, and means for identifying one of the first biometric signatures most correlated with the second biometric signature. Device. The method of claim 19, The means for correlating comprises means for obtaining a correlation coefficient associated with each first biometric signature, and the means for comparing further comprises means for comparing the correlation coefficient associated with the identified first biometric signature with a correlation coefficient threshold. Object identification device that contains. The method of claim 15, The means for comparing includes means for correlating the second biometric signature with the first biometric signature to obtain a correlation coefficient; And means for comparing the correlation coefficient associated with the identified first biometric signature with a correlation coefficient threshold. The method of claim 15, The device may include a smart card, passport, driver's license device, bio-logon identification device, palm pilot, cellular embedded identification device, anti-theft device, electrocardiogram monitoring device, electronic banking ( e-banking devices, e-transaction devices, pet identification devices, physical access devices, logical access devices, devices that combine ECG and fingerprint monitoring, and ECG signature comparisons and any other form of biometric analysis An object identification device that is one of the devices. The method of claim 15, And the device is a biometric logon identification device for remote logon to secure security for a resource. The method of claim 15, And the device is continuously operating. The method of claim 15, And the means for obtaining is configured to contact the hand or foot of the selected object. The method of claim 15, The device is provided for a smart card that is operable for a limited time after successful recognition and then inoperable until the next successful recognition is performed. The method of claim 15, And the device is configured to operate with an encryption key or digital signature. The method of claim 15, The device is included in a watch worn on the wrist, wherein the signal is measured between the wrist filled with the watch and the other hand of the wearer. Biometric Identification System, a) an ECG signal acquisition module; b) an ECG signal processor, the signal processor including an ECG signature template generator; And c) output module Biometric identification system comprising a. The method of claim 29, The ECG signature template generator has an analytical ECG model input, the generator using the input to remove common features from one or more ECG components of the ECG signals provided by the ECG signal acquisition module. The method of claim 29, Further comprising a registered signature database partitioned into subsets, wherein the ECG signature template generator, using at least one database subset, is common from one or more ECG components of the ECG signals provided by the ECG signal acquisition module; Biometric identification system to remove features. Biometric Identification System, a) an ECG signal acquisition module; b) a registered signature database; c) a signal processor further comprising an ECG signature generator, and a signature comparator comparing the ECG signature with at least one registered ECG signature; And d) output module Biometric identification system comprising a. 33. The method of claim 32, And said comparator is a closed search comparator. 33. The method of claim 32, And said signature comparator is a signature correlation analyzer. Biometric Identification System, a) a signal acquisition module; b) a registered signature database; c) an ECG signature generator, a signature comparator that compares one or more ECG signatures generated by the ECG signature generator with a plurality of registered ECG signatures from the registered signature database, a series of match scores based on the output of the signature comparator a signal processor including a match score generator for outputting a match score and a match score correlator for correlating the match score for the one or more ECG signatures with a match score for at least one registered signature; And d) output module Biometric identification system comprising a. Biometric Identification System, a) a signal acquisition module; b) a signal processor comprising an ECG signature generator and a signature comparator comprising a fuzzy logic analyzer to compare the at least one ECG signature with at least one registered ECG signature; And c) output module Biometric identification system comprising a. Biometric Identification System, a) a signal acquisition module; b) i) an ECG signature generator,   ii) a signature comparator, and   iii) further comprising a dynamic threshold generator Signal processor; And c) output module Biometric identification system comprising a. The method of claim 37, And the signature comparator is a signature correlator. Biometric Identification System, a) a signal acquisition module; b) i) an ECG signature generator,   ii) a signature correlator, and  iii) a dynamic threshold generator, the generator comprising a correlation transformer; Signal processor; And c) output module Biometric identification system comprising a. The method of claim 39, The correlation transducer is a Z-score generator. The method of claim 39, The correlation transformer is a squared correlation transformer. Biometric Identification System, a) a signal acquisition module; b) i) an ECG signature generator,   ii) a signature correlator, and   iii) further comprising a signal quality calculator Signal processor; And c) output module Biometric identification system comprising a. The method of claim 42, wherein And said signal quality calculator is a Q-value generator. The method of claim 43, And the signal quality calculator is coupled to the signal acquisition module such that low quality signal calculation causes the acquisition module to use a longer acquisition period. The method of claim 43, And the signal quality calculator is coupled to the output module in a manner that causes low quality signal calculation to indicate that the output module requires a new signal acquisition with reduced noise. Biometric Identification System, a) an ECG signal acquisition module; b) a registered signature database; c) a signal processor comprising an ECG signature generator and a signature comparator for comparing one or more ECG signatures generated by the ECG signature generator with a plurality of registered ECG signatures from the registered signature database; d) signature encryption module; And e) output module Biometric identification system comprising a. 47. The method of claim 46 wherein The signature encryption module includes a scrambler using a public key infrastructure technology. Biometric Identification System, a) a signal acquisition module, the module including an ultra-high input impedance probe; b) an ECG signal processor further comprising an ECG signature generator and a signature comparator; And c) output module Biometric identification system comprising a. The method of claim 48, And the ultra-high input resistance probe has an ultra-low noise characteristic. Biometric Identification System, a) a signal acquisition module; b) a signal processor further comprising an ECG signature generator, an ECG signature correlator, and a signature correlation weighting mechanism; And c) output module Biometric identification system comprising a. As a lock, a) a signal acquisition module; b) an ECG signal processor further comprising an ECG signature generator and an ECG signature comparator; And c) locking mechanism Locking device comprising a. As indoor access control device, a) a signal acquisition module; b) an ECG signal processor further comprising an ECG signature generator and an ECG signature comparator; And c) indoor access control; Indoor access control device comprising a. Biometric Identification System, a) an ECG signal acquisition module; b) a signal processor further comprising a pulse rate normalization module; And c) identification output module Biometric identification system comprising a. The method of claim 53, And the ECG signal processor is a digital signal processor. The method of claim 53, At least one said ECG signal processor is integrated in another device. The method of claim 53, The signal acquisition module, the signal processor, and the signal output module are part of an integrated device. As a way of identifying an entity, Generating and storing a first biometric signature identifying the particular individual by forming a difference between the representation of the particular heart's heartbeat pattern and the analytical representation of a common feature of the heartbeat patterns; After the generating and storing, generating a second biometric signature by obtaining a representation of the heartbeat pattern of the selected subject and forming a difference between the heartbeat pattern of the selected subject and the analytical representation of a common feature of the heartbeat patterns ; And Comparing the second biometric signature with the first biometric signature to determine whether the selected entity is the specific entity Object identification method comprising a. As a biometric identification method, a) obtaining a first ECG signal; b) processing the first ECG signal to generate an ECG signature template; c) obtaining a second ECG signal; d) processing the second ECG signal to generate an ECG signature; e) comparing the ECG signature with the ECG signature template; And f) outputting the result of said comparison Biometric identification method comprising a. The method of claim 58, Generating the ECG signature template removes common features of one or more ECG components from the ECG signal by subtracting common features of one or more ECG components provided by an analytical ECG model. The method of claim 58, g) creating a database of the ECG signature template; h) partitioning the ECG signature template into subsets; And i) removing common features of one or more ECG components from the ECG signal using at least one database subset Biometric identification method further comprising. As a biometric identification method, a) obtaining a first ECG signal; b) processing the first ECG signal to generate an ECG signature template; c) storing the ECG signature template in a registered signature database; d) repeating steps a) to c); e) acquiring a second ECG signal; f) processing the second ECG signal to generate an ECG signature; g) comparing the second ECG signature with at least one registered ECG signature; And h) outputting the result of said comparison Biometric identification method comprising a. 62. The method of claim 61, And the comparing step merely compares the ECG signature with a single registered ECG signature. 62. The method of claim 61, And the comparing step correlates the ECG signature with a plurality of registered signatures. As a biometric identification method, a) obtaining an ECG signal; b) processing the ECG signal to generate a registered signature database; c) placing the resulting ECG signature in a database; d) repeating steps a) to c); e) comparing the one or more ECG signatures with the plurality of registered ECG signatures; f) generating a series of match scores based on the results of said comparing step; g) correlating the match score for the one or more ECG signatures with a match score for at least one registered signature; And h) outputting the correlation result Biometric identification method comprising a. As a biometric identification method, a) obtaining an ECG signal; b) generating an ECG signature from the ECG signal; c) comparing the ECG signature with at least one registered ECG signature using fuzzy logic; And d) outputting the result of said comparison Biometric identification method comprising a. As a biometric identification method, a) obtaining an ECG signal; b) processing the ECG signal to generate an ECG signature; c) comparing the ECG signature with a plurality of registered ECG signatures; d) generating a dynamic threshold for the comparison; And e) outputting the identification result Biometric identification method comprising a. The method of claim 66, And comparing the signatures correlates the signatures. As a biometric identification method, a) obtaining an ECG signal; b) processing the ECG signal to generate an ECG signature; c) correlating the ECG signature with a plurality of registered ECG signatures, d) transforming one or more said correlations; e) generating a dynamic threshold for the transformation; And f) outputting the identification result Biometric identification method comprising a. The method of claim 68, And wherein transforming the one or more correlations is used to generate a Z-score. The method of claim 68, And transforming the one or more correlations squares the one or more correlations. As a biometric identification method, a) obtaining an ECG signal; b) calculating the quality of the signal; c) processing the ECG signal to generate an ECG signature; d) correlating the ECG signature with one or more registered ECG signatures; e) comparing the result of the correlation step with a threshold; And f) outputting the result of said comparison Biometric identification method comprising a. The method of claim 71, wherein And calculating the signal quality comprises calculating a Q-value. The method of claim 71, wherein And adjusting the acquisition time based on the quality of the signal. The method of claim 71, wherein And obtaining a new signal in response to the signal quality calculation. As a biometric identification method, a) obtaining a first ECG signal; b) processing the ECG signal to generate an ECG signature; c) encrypting the signature; d) adding the encrypted signature to a registered signature database; e) acquiring a second ECG signal; f) processing the ECG signal to generate a second signature; And g) comparing the second signature with one or more of the registered signatures in the registered signature database. Biometric identification method comprising a. 76. The method of claim 75, The signature encryption step scrambles the signature using a public key infrastructure technology. As a biometric identification method, a) acquiring an ECG signal using an ultra high input impedance probe; b) processing the ECG signal to generate an ECG signature; c) comparing the signature with at least one registered signature in a registered signature database; And d) outputting the result of said comparison Biometric identification method comprising a. 78. The method of claim 77 wherein The ultra-high input resistance probe has an extremely low noise characteristic. As a biometric identification method, a) obtaining an ECG signal; b) processing the signal to generate an ECG signature; c) correlating the ECG signature with at least one ECG signature template in a registered signature database; d) weighting the result of the signature correlation; e) comparing the result of the weighted correlation with a threshold; And f) outputting the result of said comparison Biometric identification method comprising a. As a way to lock a security device, a) obtaining an ECG signal; b) processing the ECG signal to generate an ECG signature; c) comparing the ECG signature with one or more ECG signature templates in a registered signature database; d) comparing the result of said comparison with an identification threshold; And e) affecting the locking mechanism based on the comparison Security device locking method comprising a. As a way to control indoor entry, a) obtaining an ECG signal; b) processing the ECG signal to generate an ECG signature; c) comparing the ECG signature with one or more ECG signature templates in a registered signature database; d) comparing the result of said comparison with an identification threshold; And e) allowing or denying indoor entry based on the comparison Indoor access control method comprising a. As a biometric identification method, a) obtaining an ECG signal; b) processing the signal by normalizing the signal to a pulse rate; c) generating an ECG signature; d) correlating the ECG signature with at least one ECG signature template from a signal obtained with the normalized pulse rate or a signal normalized to a pulse rate; e) comparing the result of said correlation with a threshold; And f) outputting the result of said comparison Biometric identification method comprising a. 83. The method of claim 82, And said processing step digitally processes said signal. 83. The method of claim 82, Biometric identification method further comprising obtaining a non-ECG biometric reading. 85. The method of claim 84, And identifying the individual by evaluating the non-ECG biometric value and the output comparison result. The method of claim 1, The method of claim 1 further comprising obtaining a non-ECG biometric. The method of claim 15, And a credit card operable for a limited period of time after positive identification and then inoperable until the next successful positive identification is performed. The method of claim 15, An object identification device further comprising a non-ECG biometric acquisition module. The method of claim 29, A biometric identification system further comprising a non-ECG biometric acquisition module. As an age analyzer, a) an ECG acquisition module; b) an ECG signal processor; c) processed ECG signal comparator; And d) age analysis output module Age analyzer comprising a. 91. The method of claim 90, The ECG signal processor includes a signature generator and the processed signal comparator is a signature comparator. 91. The method of claim 90, The processed ECG comparator compares the width of the subject's QRS group (QRS complex) with the width of the QRS group signal template. 91. The method of claim 90, And the output module outputs its output through the internet. As the age detection method, a) obtaining an ECG signal; b) processing the ECG signal; c) comparing the processed ECG signal with one or more reference signals; And d) controlling access to an Internet web site based on the results of said comparing step; Age detection method comprising a. 95. The method of claim 94, And the comparing step compares the ECG signature with one or more ECG signature templates. 95. The method of claim 94, And the comparing step compares the width of the QRS signal group with the width of the one or more reference signal QRS groups. Biometric Identification System, a) an electrocardiogram (ECG) signal acquisition module; b) a registered signature database; c) an ECG signal processor including an ECG signature generator for removing characteristic waveforms representing common ECG characteristics of a group of entities from the ECG signal obtained by the ECG signal acquisition module; d) an ECG signature comparator that compares the ECG signature with at least one registered ECG signature; And e) output module Biometric identification system comprising a. 97. The method of claim 97, And the ECG signature generator removes characteristic waveforms representing first principal components derived from Principal Component Analysis (PCA) of the group's ECG signals. 99. The method of claim 98, The characteristic waveforms are weighted to approximate a range of characteristic waveforms present in the ECG signal obtained by the ECG signal acquisition module. 105. The method of claim 99, And the ECG signature generator removes the approximation from the ECG signal obtained by the ECG signal acquisition module. 99. The method of claim 98, And the ECG signal acquisition module obtains an ECG signal from an entity that is not a member of the group of entities. The method of any one of claims 97-99, wherein The characteristic waveforms are derived from synthetic ECGs. 97. The method of claim 97, And the ECG signature generator removes characteristic waveforms derived from ICA of the group's ECG signals. 97. The method of claim 97, And the ECG signature generator removes characteristic waveforms derived from wavelet decomposition (WD) of the group's ECG signals. 105. The method of claim 99, The system uses a reconstruction coefficient to weight the characteristic waveforms. The method of any one of claims 97-99, wherein The ECG acquisition module includes a bi-directional interface that provides a biological challenge-response mechanism that does not require a conscious response from the user. 107. The method of claim 106, And said bidirectional interface comprises a conductive medium. 107. The method of claim 107 wherein And the conductive medium is contained within a garment article. As a way of identifying an entity, a) obtaining an ECG of the subject; b) decomposing the ECGs from the group of entities to determine a series of characteristic waveforms representing common features of the group; c) processing the subject's ECG by removing the characteristic waveforms; And d) identifying the subject using the subject's processed ECG Object identification method comprising a. 109. The method of claim 109, And the subject is not a member of the group of individuals. 109. The method of claim 109, Wherein said decomposed ECG is synthetic. 109. The method of claim 109, Wherein said steps are performed in the order listed. 109. The method of claim 109, Weighting the characteristic waveforms to approximate a range of common features present in the subject's ECG. 113. The method of claim 113, Removing the characteristic waveforms is achieved by removing the approximation. 113. The method of claim 113, The decomposing step is performed by applying PCA to ECGs from the group of individuals, wherein the weighting step is a reconstruction coefficient for the principal components representing common features to approximate the range of the characteristic waveforms present in the subject's ECG. An object identification method performed by determining reconstruction coefficients. 116. The method of claim 115, And the approximation is removed from the subject's ECG. 116. The method of claim 115, The principal components representing common features are time-shifted to determine the reconstruction coefficients. 109. The method of claim 109, Wherein said dissolving step is independent component analysis (ICA). 109. The method of claim 109, And said dissolving step is wavelet decomposition (WD). 112. The method of any of claims 109-112, And subjecting the subject to a test in a manner that does not require a conscious response of the subject. 121. The method of claim 120, The requesting examination includes the step of providing an electrical stimulus through a conductive medium that is also used to obtain the subject's ECG. As a biometric identification method, a) obtaining an ECG signal; b) processing the ECG signal to generate an ECG signature; c) correlating the ECG signature with a plurality of synthetic ECG signatures; d) transforming one or more of the correlations; e) generating a dynamic threshold for the correlation; And f) outputting the identification result Biometric identification method comprising a. 123. The method of claim 122 wherein Wherein said steps are performed in the order listed.

Images