Biometric Authentication Based on Heart Signals
Non-contact cardiac signal processing via radar detection and machine learning algorithms addresses the limitations of existing heart signal authentication methods, providing accurate and resource-efficient identification under diverse conditions.
Patent Information
- Application Number
- JP2024522147
- Authority / Receiving Office
- JP · JP
- Patent Type
- Applications
- Current Assignee / Owner
- Priority Date
- 2021-10-21
- Filing Date
- 2022-10-19
- Publication Date
- 2024-11-13
AI Technical Summary
Existing biometric authentication methods using heart signals require specialized medical equipment, physical contact, and are prone to malfunction and inaccuracy under low light or obstructed conditions, limiting their usability and accuracy.
A method and system for biometric authentication using non-contact cardiac signals, such as ballistocardiography, processed through radar detection and machine learning algorithms to generate unique frequency encodings for subject identification, enabling accurate classification without direct physical contact or line of sight.
Enables real-time, high-accuracy identification of individuals using minimal resources, functioning under various conditions, including low light and through obstructions, with iterative model refinement for improved predictive power.
Smart Images

Figure 2024541835000001_ABST
Abstract
Description
[Technical field]
[0001] The present invention relates generally to the fields of biometric authentication and classification, temporal signal processing, and machine learning analysis. [Background technology]
[0002] Biometric identifiers are distinct and measurable biological attributes that can be used to describe, identify, and categorize individuals. Biometric identifiers can be related to unique physiological characteristics, such as fingerprints, eye features (e.g., iris recognition), facial features, or hand / palm features. One highly reliable form of physiological identification is DNA profiling, which involves analyzing unique variations in DNA sequences, primarily known as short tandem repeats, for matching DNA samples obtained from bodily fluids. DNA profiling is commonly used today in a variety of applications ranging from criminal investigations to paternity testing. Biometric identifiers can also be based on behavioral characteristics, examples of which include signature recognition, voice recognition, and gait analysis.
[0003] Recently, biometric authentication based on electrocardiograms, which measure the electrical activity of the heart, has been developed. However, this requires dedicated medical equipment, including an electrocardiograph or electrocardiogram (ECG) machine with multiple electrodes, which must be placed directly on the patient's body (usually on the chest or limbs) to obtain the ECG signal. A qualified doctor or clinician is generally required to implement the proper positioning of the electrodes and to operate the ECG machine. Information about the cardiac cycle can also be derived from photoplethysmography (PPG) measurements, which use optical measurement techniques to monitor volume fluctuations in blood circulation. PPG measurements are obtained by illuminating the skin with a light source and then measuring changes in reflection or absorption of light with a photodetector, such as by means of a pulse oximeter. However, such devices usually require components to be in physical contact with the subject's body parts, such as attached to a finger. There are also wearable devices, such as smart watches or chest straps, that incorporate smaller sensors configured to obtain heart rate or cardiac cycle information. However, these wearable cardiac monitoring devices are often cumbersome and prone to malfunction and inaccuracy. Furthermore, these devices must operate under sufficient lighting conditions and have a direct line of sight with clear visibility to the skin area being measured. Thus, these devices cannot function under low light or poor visibility conditions, or through obstructions or obstacles such as clothing worn by the subject.
[0004] Publications describing biometric authentication using heart rate signals include: Paiva, JS,Dias, D.,&Cunha, J. (2017).Beat-ID:Towards a computationally low-cost single heartbeat biometric identity check system based on electrocardiogram wave morphology.PloS one, 12(7), e0180942, Odinaka, I., Lai, P., Kaplan, AD, O'Sullivan, J., Sirevaag, E., & Rohrbaugh, J. (2012).ECG Biometric Recognition:A Comparative Analysis.IEEE Transactions on Information Forensics and Security, 7, 1812-1824, Calleja, A., Peris-Lopez, P., Tapiador, JE Electrical Heart Signals can be Monitored from the Moon:Security Implications for IPI-Based Protocols.In Information Security Theory and Practice, Springer International Publishing:Cham, Switzerland, 2015, pp. 36-51, and Wang, W., Stuijk, S., De Haan, G. Unsupervised subject detection via remote PPG.IEEE Trans.Biomed.Eng.2015, 62, 2629-2637. Summary of the Invention
[0005] According to one aspect of the invention, a method for biometric authentication is provided. The method includes acquiring a cardiac signal from at least one reference subject and segmenting the acquired cardiac signal into a plurality of cardiac segments over a selected duration. The method further includes applying at least one linear mapping to each of the cardiac segments to produce a respective cardiac frequency encoding, and assigning each cardiac frequency encoding an identification label for the reference subject. The method further includes applying at least one machine learning process to the set of cardiac frequency encodings during a modeling phase to generate a model for subject classification. The method further includes applying the model to the input cardiac frequency encodings during an identification phase to classify the input cardiac frequency encoding as belonging to the reference subject if a matching classification is obtained, or to determine that the input cardiac frequency encoding belongs to a non-reference subject if a matching classification is not obtained. The cardiac signal may be a ballistocardiogram (BCG) signal. The cardiac signal may be acquired using non-contact means. Segmenting the acquired cardiac signal may be performed by applying at least one process of peak detection, zero crossing detection, RR interval detection, and / or interbeat interval (IBI) detection. Applying the linear mapping may include filtering with a plurality of band pass filters. The BCG signal may be acquired using a remote non-invasive radar detector, the remote non-invasive radar detector including at least one radar transmitter (Tx) configured to transmit a THz signal to a predetermined body tissue of the subject, at least one radar receiver (Rx) configured to receive a reflected THz signal reflected from the body tissue of the subject, and a radar detector processor communicatively coupled to the radar transmitter and the radar receiver and configured to process the received reflected THz signal to produce an encoding representative of the heart-related information of the subject. The subject classification may include at least one characteristic of age, sex, race, physiological state, mental state, health state, and any combination thereof.The model may be developed using a machine learning process among a neural network algorithm, an artificial neural network algorithm, a convolutional neural network algorithm, a recurrent neural network algorithm, a deep learning algorithm, a linear regression model, a logistic regression model, a data clustering model, a linear classifier model, a support vector machine (SVM) model, a random forest process, and any combination thereof. The method may include simultaneously monitoring and / or identifying multiple subjects at a location.
[0006] According to another aspect of the invention, a system for biometric authentication is provided. The system includes a cardiac signal detector configured to acquire a cardiac signal of at least one reference subject. The system further includes a cardiac signal processor configured to segment the acquired cardiac signal into a plurality of cardiac segments over a selected duration, apply at least one linear mapping to each of the cardiac segments to generate a respective cardiac frequency encoding, and assign each cardiac frequency encoding an identification label for the reference subject. The system further includes a machine learning processor configured to apply at least one machine learning process to the set of cardiac frequency encodings during a modeling phase to create a model for subject classification. The machine learning processor is further configured to apply the model to the input cardiac frequency encodings during an identification phase to classify the input cardiac frequency encodings as belonging to the reference subject if a matching classification is obtained, or to determine that the input cardiac frequency encodings belong to a non-reference subject if a matching classification is not obtained. The cardiac signal may be a ballistocardiogram (BCG) signal. The cardiac signal may be acquired using non-contact means. Segmenting the acquired cardiac signal may be performed by applying at least one of the following processes: peak detection, zero-crossing detection, RR interval detection, and / or interbeat interval (IBI) detection. Applying the linear mapping may include filtering with a plurality of bandpass filters. The BCG signal may be acquired using a remote non-invasive radar detector, the remote non-invasive radar detector including at least one radar transmitter (Tx) configured to transmit a THz signal to a predetermined body tissue of the subject, at least one radar receiver (Rx) configured to receive a reflected THz signal reflected from the body tissue of the subject, and a radar detector processor communicatively coupled to the radar transmitter and the radar receiver and configured to process the received reflected THz signal to produce an encoding representative of the heart-related information of the subject.The subject classification may include at least one of the following characteristics: age, sex, race, physiological condition, mental condition, health condition, and any combination thereof. The model may be developed using a machine learning process: a neural network algorithm, an artificial neural network algorithm, a convolutional neural network algorithm, a recurrent neural network algorithm, a deep learning algorithm, a linear regression model, a logistic regression model, a data clustering model, a linear classifier model, a support vector machine (SVM) model, a random forest process, and any combination thereof. The system may include simultaneously monitoring and / or identifying multiple subjects at a location. [Brief description of the drawings]
[0007] The invention will be more fully understood and appreciated from the following detailed description taken in conjunction with the drawings in which:
[0008] [Figure 1] 1 is a schematic diagram of a system for biometric authentication constructed and operative in accordance with an embodiment of the present invention; [Diagram 2] 2 is an illustration of an exemplary cardiac signal segmentation operating in accordance with an embodiment of the present invention; [Diagram 3] FIG. 13 illustrates an example frequency encoding of a single cardiac segment subjected to six different linear filters operating in accordance with an embodiment of the present invention. [Figure 4] FIG. 2 is a schematic diagram of a flow diagram of a biometric authentication method operating in accordance with an embodiment of the present invention. [Diagram 5] FIG. 4 is a block diagram of a method for biometric authentication operating in accordance with another embodiment of the present invention. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS
[0009] The present invention overcomes the shortcomings of the prior art by providing a method and system for uniquely identifying a biometric using a distinct physiological signature extracted from cardiac information. A subject is uniquely identified using a single heartbeat acquired from a cardiac signal, such as a BCG signal or a segment of such a signal, which may be acquired in a contactless manner. The disclosed system and method involves segmenting and filtering the acquired cardiac signal to generate a unique frequency encoding representative of the subject's heartbeat. The encoding is assigned a corresponding identification label and fed into a supervised learning process to create classifications and models of subjects based on different features and profiles, which are then utilized to facilitate identification. The disclosed system and method enables real-time identification of people with high accuracy and minimal resources. Subject identification may be applied to a variety of applications, such as healthcare monitoring. Furthermore, since cardiac signals can be extracted directly from biological skin tissue, the disclosed method may be used to distinguish human skin from other surfaces in video image content.
[0010] The terms "user" and "operator" are used interchangeably herein to refer to any individual or group of individuals who use or operate the methods or systems of the present invention, such as the person implementing the identification process for the subject to be identified.
[0011] The terms "subject" and "living body" are used interchangeably herein to refer to an individual, such as a person to be identified, on whom a method or system of the present invention is operated. A subject can be any living organism, such as a person, human or animal, characterized by a functioning heart beat associated with the cardiac cycle of its heart.
[0012] The term "repeated" as used herein should be interpreted broadly to include any one or more of "continuously," "periodic repeated," and "non-periodic repeated," where periodic repeated is characterized by fixed length intervals between repeats and non-periodic repeated is characterized by variable length intervals between repeats.
[0013] Reference is now made to Figure 1, which is a schematic diagram of a system for biometric authentication, generally referenced as 110, constructed and operative in accordance with an embodiment of the present invention. System 110 includes a cardiac signal detector 112, a cardiac signal processor 114, a machine learning processor 116, and a database 118. Cardiac signal processor 114 is communicatively coupled to cardiac signal detector 112 and database 118. Machine learning processor 116 is coupled to database 118.
[0014] The cardiac signal detector (CSD) 112 is configured to acquire cardiac signals, referenced as 122, related to the cardiac cycle of the subject, referenced as 120. For example, the CSD 112 may include one or more ballistocardiogram (BCG) sensors operative to acquire BCG signals related to vibration or impulse forces in the body resulting from cardiac activity. Alternatively, the CSD 112 may be an electrocardiogram (ECG) machine operative to acquire ECG signals representative of electrical activity associated with the cardiac cycle. Still alternatively, the CSD 112 may be a photoplethysmography (PPG) device, such as a pulse oximeter, operative to acquire PPG signals representative of volumetric fluctuations in blood circulation associated with the cardiac cycle. Still alternatively, the CSD 112 may be a contact-based sensor, such as a pressure sensor (e.g., a cuff pressure gauge) attached or worn on the body, configured to acquire measurements of body motion associated with blood circulation linked to the cardiac cycle. Thus, the cardiac signal 122 may represent any applicable signal related to the cardiac cycle of the subject 120, including, but not limited to, a BCG signal, an ECG signal, a PPG signal, a body motion signal associated with blood circulation, etc. The CSD 112 may be embodied by a radar detector that transmits a THz radar signal to a body part (e.g., chest or back) of the subject 120 and receives and processes the reflected THz radar signal to produce a cardiac signal (e.g., a BCG signal), for example, as described in Neteera Technologies' PCT Publication No. WO2018 / 167777A1 entitled "Method and device for non-contact sensing of vital signs and diagnostic signals by electromagnetic waves in the sub terahertz band" and Neteera Technologies' PCT Publication No. WO2020 / 012455A1 entitled "A sub-THz and THz system for physiological parameters detection and method thereof."The term "Terahertz (THz)" as used herein refers to terahertz and sub-terahertz radiation, which corresponds to sub-millimeter and millimeter wave radiation, such as electromagnetic waves in the frequency band of 0.03 THz to 3 THz, which corresponds to radiation wavelengths of 10 mm to 0.1 mm. It should be noted that the CSD 112 preferably operates in a non-contact manner, i.e., without requiring direct physical contact with the subject, such as, for example, via the aforementioned radar detector, which acquires the BCG signal remotely and does not require device components to be in direct physical contact with or worn or attached to the subject 120. Alternatively, the CSD 112 may acquire the cardiac signal 122 via contact measurements, such as ECG electrodes or a pulse oximeter positioned on or attached to a body part of the subject 120. It is further noted that a radar detector type CSD 112 may enable acquisition of cardiac signals from any direction of the subject 120, such as from in front of or behind the subject 120, or from non-orthogonal angles relative to the subject 120 (i.e., relative to where the measurement radar signals are transmitted and received). Also, a radar detector type CSD 112 may effectively acquire cardiac signals without necessarily having a direct line of sight to the measured body part of the subject 120, such as in low light or poor visibility conditions and through certain types of obstructions or material barriers (e.g., various types of clothing that may be worn by the subject 120).
[0015] The cardiac signal processor (CSP) 114 and the machine learning processor (MLP) 116 receive information or instructions from other components of the system 110 and perform the necessary data processing. For example, the CSP 114 receives and processes the cardiac signals 122 acquired by the CSD 112 to extract a unique identifier therefrom, as described in further detail below. Similarly, the MLP 116 analyzes the processed cardiac signal information and / or associated identifiers acquired from the CSP 114 to produce identification and classification information, as described in further detail below. The database 118 stores relevant information retrieved and processed by the CSP 114 and / or the MLP 116, such as the processed cardiac signal data and associated identification and classification information. The database 118 may be represented by one or more local servers or by remote and / or distributed servers, such as in a cloud storage platform.
[0016] Information may be communicated between components of system 110 via any suitable data communication channel or network, using any type of channel or network model and any data transmission protocol (e.g., wired, wireless, radio, WiFi, Bluetooth, etc.). For example, system 110 may store, manage, and / or process data using a cloud computing model, and components of system 110 may communicate with each other and be remotely monitored or controlled via the Internet, such as via an Internet of Things (IoT) network. Components and devices of system 110 may be based on hardware, software, or a combination thereof. It is understood that functionality associated with each of the devices or components of system 110 may be distributed among multiple devices or components that may be in a single location or in multiple locations. For example, functionality associated with CSP 114 and / or MLP 116 may be distributed among a single processing unit or multiple processing units. CSP 114 and / or MLP 116 may be part of a server or remote computer system accessible via a communication medium or network, such as a cloud computing platform. The CSP 114 and / or the MLP 116 may also be integrated with other components of the system 110, such as incorporating the CSD 112.
[0017] 1 to enable implementation of the disclosed subject matter. For example, system 110 may include a user interface (not shown) that allows a user to control various parameters or settings associated with the components of system 110, a display device (not shown) for visually displaying information regarding the operation of system 110, and / or a camera or imaging device (not shown) for capturing images of the operation of system 110.
[0018] The operation of the system 110 will now be described in general terms, followed by a specific example. The CSD 112 acquires a cardiac signal 122 of a reference subject 120, such as a BCG signal. The cardiac signal 122 may be acquired using a non-contact detection technique, such as radar detection. The CSP 114 receives and processes the acquired cardiac signal 122. Specifically, the continuous cardiac signal is segmented into discrete time portions, referred to herein as "cardiac segments", that represent individual cardiac beats. The segmentation is performed using any suitable segmentation procedure or algorithm known in the art. For example, the segmentation may be based on peak detection (i.e., the duration between successive peaks of the QRS signal), zero crossing detection, RR interval detection (i.e., the duration between successive R waves of the QRS signal), interbeat interval (IBI) detection, etc. The processing also includes the application of a linear mapping to generate an encoding in the frequency domain. The linear mapping may be performed before or after segmentation, and thus the linear mapping may be applied to each heartbeat segment or to the original heart signal. Thus, the signal or the heartbeat segments are linearly mapped, such as by passing through several linear time-invariant filters, e.g. band-pass filters, which may be obtained from a predefined filter bank, to generate a set of filter responses. Alternatively, the linear mapping may involve a Fourier transform, or some type of convolution operator, rather than linear filtering. The linear mapping and segmentation may be combined with an optional non-linear mapping of the values of the segments. For example, the non-linear mapping may include rectification (i.e. taking the absolute value of the (complex) signal), followed by exponentiation (e.g. squaring), followed by gain control (i.e. division by the average energy of the signal over a short period of time).
[0019] The processing results in a number of filter responses (or linear mapping responses) corresponding to each heartbeat segment and representing a unique encoding of the subject's physiology. The filter response for a given heartbeat segment can be represented as a matrix of size "hb_length" x "n_filters", where "hb_length" is the length or duration of the heartbeat segment (i.e., the number of samples) and "n_filters" is the number of filters used to create the filter response (e.g., the number of band-pass / time-invariant filters in the filter bank). The set of filter responses for at least one heartbeat segment is collectively referred to as a "heartbeat frequency encoding", and while the filter response for a single heartbeat segment represents a minimum encoding of the subject, the filter responses for several consecutive heartbeat segments of the subject may provide an enhanced subject encoding. Each heartbeat frequency encoding is assigned an identification (ID) label associated with the particular reference subject to which it belongs. For example, the identification label may include a name, an identification number, and / or other personal information about the reference subject, such as age, sex, location, physical attributes, etc. The heart rate frequency encoding and the associated ID label are then stored in database 118.
[0020] The above process is repeated for multiple reference subjects to obtain a set of heartbeat frequency encodings (all of which are stored in database 118). The collected heartbeat frequency encodings are then analyzed using a machine learning process by MLP 116 to create a model to (implicitly) identify different patterns to facilitate identification and classification of different subjects according to various criteria. The machine learning process may apply machine learning techniques to analyze the training data (i.e., the collected heartbeat frequency encodings) to generate a mapping function that can be used to classify additional instances of new heartbeat frequency encodings according to the associated classification criteria. Data analysis may utilize any suitable machine learning or supervised learning process or algorithm, including, but not limited to, artificial neural network (ANN) processes such as convolutional neural networks, recurrent neural networks (RNNs), or deep learning algorithms, classification or regression analysis such as linear regression models, logistic regression models or support vector machine (SVM) models, decision tree learning approaches such as random forest classifiers, and / or any combination thereof. Data analysis may utilize any suitable tool or platform, such as publicly available open source machine learning or supervised learning tools.
[0021] The MLP 116 establishes a classification or profile of the reference subjects whose heartbeat frequency encodings were collected. The classification process may divide the reference subjects into different groups or categories based on common features. For example, the classification process may provide a model for identifying subjects in various categories, such as based on age, sex, location, physical attributes, etc. Each category may be associated with a relative weighting metric that corresponds to a confidence level for the respective category feature. Different models may then be applied to facilitate the identification of new subjects belonging to the relevant category or classification. The created model may be iteratively updated and improved based on new information, such as taking into account subsequent successful or unsuccessful subject identifications and additional heartbeat encodings collected from new reference subjects. Simulation of multiple heartbeat frequency encodings and classifications may also be applied to increase the reliability and accuracy of the model. The updated model may provide optimal formulas and weighting metrics for different variables or classification features. As more information and statistics are accumulated, the model can be further refined to improve predictive capabilities for subject identification.
[0022] Reference is now made to Figure 2, which is an illustration of an exemplary cardiac signal segmentation operative in accordance with an embodiment of the present invention. Plot 130 shows a BCG signal (represented as voltage levels as a function of time) divided into discrete time segments, such as segment 132, with each segment representing the duration of an individual cardiac beat. Segmentation may be performed, for example, using a peak detection algorithm to detect J-peaks in the BCG signal, whereby the duration between adjacent maximum J-peaks is designated as an individual cardiac beat segment.
[0023] Please refer to FIG. 3, which is an illustration of an exemplary frequency encoding of a single heartbeat segment subjected to six different linear filters operating in accordance with an embodiment of the present invention. FIG. 3 shows six different spectral functions, which are plotted in graph 140 as a function of time (x-axis) and frequency (y-axis). The spectral functions are the result of linear filters (and nonlinear mapping) applied to the segmented heartbeat signal, where six different filters are applied to generate six different spectral functions with common start and end times, collectively forming the heartbeat frequency encoding. Alternatively, a different number of linear filters may be applied (to generate a different number of spectral responses). The heartbeat frequency encoding represents an exemplary sample used to train a machine learning process (e.g., a neural network) during a training or modeling phase.
[0024] It will be appreciated that the systems and methods of the present invention may provide subject identification from a single beat segment of a cardiac signal without requiring a device or component to make direct physical contact with the subject. Also, the subject need not be in direct line of sight to a cardiac signal detector, such as a radar detector, which may operate under poor visibility or light saturation conditions, under obstructions or interference, and from different angles (e.g., from the front or from behind) relative to the subject. Source separation techniques may be utilized to enable simultaneous measurement and identification of multiple subjects. The disclosed systems do not require expensive equipment, have relatively few components, and may be relatively simple to operate and maintain. Machine learning analysis may also provide reliable and accurate predictive models, which may be iteratively refined to enable identification and classification of new subjects based on cardiac frequency encoding.
[0025] Please refer to Figure 4, which is a schematic diagram of a flow diagram of a biometric authentication method operating in accordance with one embodiment of the present invention.
[0026] Reference is now made to Figure 5, which is a block diagram of a method for biometric authentication operating in accordance with an embodiment of the present invention. In step 162, a cardiac signal is acquired. Referring to Figure 1, a cardiac signal detector 112 acquires a cardiac signal 122 related to the cardiac cycle of a reference subject 120, such as a BCG signal. The cardiac signal 122 may be acquired in a non-contact manner and under poor visibility conditions and without a direct line of sight to the measured body area, such as via a designated radar detection unit.
[0027] In procedure 164, the cardiac signal is segmented into cardiac segments over a selected duration. With reference to Figure 1, cardiac signal processor 114 receives and processes cardiac signal 122 acquired by cardiac signal detector 112 by performing a segmentation process to divide cardiac signal 122 into individual cardiac beats, such as using peak detection, zero-crossing detection, RR interval detection, or beat interval detection.
[0028] In step 166, a linear mapping is applied to each cardiac segment to generate a respective cardiac frequency encoding. With reference to Figures 1 and 2, the cardiac signal processor 114 performs the linear mapping operation, such as by applying a series of bandpass filters or other linear time-invariant filters to the cardiac signal 122 or its cardiac segments. The CSP 114 may apply alternative linear mappings, such as a Fourier transform or a convolution operator. The CSP 114 may also perform optional non-linear mappings, such as rectification (taking absolute values), power, and gain control operations. The resulting segmentation and linear mapping process results in a series of frequency responses for each cardiac segment that represent the unique cardiac frequency encoding of the reference subject 120.
[0029] Each heartbeat frequency encoding is assigned a corresponding identification label associated with the respective subject, in step 168. Referring to Figures 1 and 4, the cardiac signal processor 114 assigns the generated heartbeat frequency encodings a reference subject 120 ID label, including a name, ID number and / or additional personal information about the reference subject 120, and stores that information in the database 118.
[0030] In step 170, a machine learning process is applied to the collection of generated heartbeat frequency encodings to obtain a model for classification of subjects. With reference to Figures 1 and 4, the machine learning processor 116 performs a machine learning process, such as one or more supervised learning techniques known in the art, on the collected heartbeat frequency encodings and associated ID labels to generate a model. When applied to an input of the same representation, such as the trained input (i.e., the filtered heartbeats), the model can generate a classification of the subject as belonging to one of the reference groups of subjects introduced in the training phase, or determine that the sample belongs to a stranger, i.e., that a person does not belong to one of the reference subjects and is therefore unknown to the model.
[0031] In step 172, at least one subject is identified according to the heart rate frequency response encoding generated and the assigned identification label. Referring to Figures 1 and 4, the machine learning processor 116 performs a machine learning process to identify new subjects according to the profiles and classifications generated during the modeling phase. Source separation techniques known in the art may be applied to allow for the simultaneous monitoring and identification of multiple subjects at a location.
[0032] The method of FIG. 5 is typically implemented in an iterative manner such that at least some of the steps are performed repeatedly to provide dynamic biometric authentication of multiple subjects in real time.
[0033] The disclosed biometric authentication methods can be used for a variety of applications. For example, the disclosed identification methods can be used to associate collected BCG (or other cardiac signal) data with a general health information file of an identified subject that may be present in cloud storage. The disclosed identification methods can also be used to monitor vulnerable or incapacitated individuals, such as to identify whether a person in an elderly care home or assisted living facility has unintentionally entered a room or used a personal item belonging to someone else (e.g., slept in someone else's bed) and notify staff members accordingly.
[0034] While specific embodiments of the disclosed subject matter have been described to enable those skilled in the art to practice the invention, the foregoing description is intended to be illustrative only and should not be used to limit the scope of the disclosed subject matter, which should be determined by reference to the claims that follow.
Claims
1. 1. A method for biometric authentication, comprising: acquiring a cardiac signal from at least one reference subject; segmenting the acquired cardiac signal into a plurality of cardiac segments over a selected time duration; applying at least one linear mapping to each of the cardiac segments to generate a respective cardiac frequency encoding; assigning to each heartbeat frequency encoding an identification label with respect to said reference subject; applying at least one machine learning process to the set of heart rate frequency encodings during the modeling stage to produce a model for subject classification; applying said model to an input cardiac frequency encoding during a discrimination phase in order to classify said input cardiac frequency encoding as belonging to a reference subject if a matching classification is obtained, or to determine that said input cardiac frequency encoding belongs to a non-reference subject if a matching classification is not obtained; A method for providing the above.
2. The method of claim 1 , wherein the cardiac signal is a ballistocardiogram (BCG) signal.
3. The method of claim 1 , wherein the cardiac signals are acquired using non-contact means.
4. The step of segmenting the acquired cardiac signal comprises: Peak detection, Zero cross detection, RR interval detection, and Interbeat interval (IBI) detection 2. The method of claim 1, wherein the method is performed by applying at least one process selected from the group consisting of:
5. The method of claim 1 , wherein the step of applying a linear mapping comprises filtering with a plurality of bandpass filters.
6. The BCG signal is at least one radar transmitter configured to transmit a THz signal to a predetermined body tissue of the subject; at least one radar receiver configured to receive reflected THz signals reflected from the body tissue of the subject; a radar detector processor communicatively coupled to the radar transmitter and the radar receiver and configured to process the received reflected THz signals to produce an encoding representative of cardiac-related information of the subject; 13. The method of claim 1, wherein the signal is acquired using a remote non-invasive radar detector comprising:
7. The subject classification includes: Age, sex, race, physiological condition, mental condition, health condition, and any combination thereof The method of claim 1 , comprising at least one feature selected from the group consisting of:
8. The model is Neural network algorithms, artificial neural network algorithms, convolutional neural network algorithms, recurrent neural network algorithms, deep learning algorithms, linear regression models, logistic regression models, data clustering models, linear classifier models, support vector machine (SVM) models, random forest processes, and any combination thereof. The method of claim 1 , wherein the method is produced using a machine learning process selected from the group consisting of:
9. The method of claim 1 , comprising simultaneously monitoring or identifying multiple subjects at a location.
10. 1. A system for biometric authentication, comprising: a cardiac signal detector configured to acquire a cardiac signal of at least one reference subject; a cardiac signal processor configured to segment the acquired cardiac signal into a plurality of cardiac segments over a selected duration, to apply at least one linear mapping to each of the cardiac segments to generate a respective cardiac frequency encoding, and to assign to each cardiac frequency encoding an identification label with respect to the reference subject; a machine learning processor configured to apply at least one machine learning process to the set of heartbeat frequency encodings during a modeling phase to create a model for subject classification, and to apply said model to the input heartbeat frequency encodings during a discrimination phase to classify the input heartbeat frequency encodings as belonging to a reference subject if a matching classification is obtained, or to determine that the input heartbeat frequency encodings belong to a non-reference subject if a matching classification is not obtained; A system comprising:
11. The system of claim 10 , wherein the cardiac signal is a ballistocardiogram (BCG) signal.
12. The system of claim 10 , wherein the cardiac signals are acquired using non-contact means.
13. The cardiac signal processor includes: Peak detection, Zero cross detection, RR interval detection, and Interbeat interval (IBI) detection The system of claim 10 , configured to segment the acquired cardiac signal by applying at least one process selected from the group consisting of:
14. The system of claim 10 , wherein the linear mapping comprises filtering with a plurality of bandpass filters.
15. The BCG signal is at least one radar transmitter configured to transmit a THz signal to a predetermined body tissue of the subject; at least one radar receiver configured to receive reflected THz signals reflected from the body tissue of the subject; a radar detector processor communicatively coupled to the radar transmitter and the radar receiver and configured to process the received reflected THz signals to produce an encoding representative of cardiac-related information of the subject; 11. The system of claim 10, wherein the signal is acquired using a remote non-invasive radar detector comprising:
16. The subject classification includes: Age, sex, race, physiological condition, mental condition, health condition, and any combination thereof The system of claim 10 comprising at least one feature selected from the group consisting of:
17. The model is Neural network algorithms, artificial neural network algorithms, convolutional neural network algorithms, recurrent neural network algorithms, deep learning algorithms, linear regression models, logistic regression models, data clustering models, linear classifier models, support vector machine (SVM) models, random forest processes, and any combination thereof. The system of claim 10 , wherein the system is produced using a machine learning process selected from the group consisting of:
18. The system of claim 10 , further comprising simultaneously monitoring or identifying multiple subjects at a location.
Citation Information
Patent Citations
Electrocardiogram authentication method and apparatus, and training method
JP2017123170A
A sub-THZ and THZ system for physiological parameters detection and method thereof
WO2020012455A1
