An identity recognition device and method based on breathing-induced electroencephalogram signals

Abstract

The present invention provides an identity recognition device and method based on breathing-induced electroencephalogram signals, which belongs to the field of identity recognition. The identity recognition device based on breathing-induced electroencephalogram signals includes: a housing, which is fixedly wrapped around the user's head during operation; electrodes, which are arranged on the housing and used to collect the initial electroencephalogram signals induced by the user's breathing; a data processing unit, which is arranged on the housing and connected to the electrodes, and is used to filter and amplify the initial electroencephalogram signals to obtain target electroencephalogram signals; a host computer, which is connected to the data processing unit, and is used to identify the user's identity based on the target electroencephalogram signals. Since electroencephalogram signals are captured through consciousness, the problem of privacy leakage caused by traditional biometric identification methods is avoided, and the security of identity recognition is improved. Moreover, by inducing electroencephalogram signals through breathing, no additional display equipment is required, the device is simple, and the complexity of identity recognition is reduced.

Description

Technical Field

[0001] The present invention relates to the field of identity recognition, and in particular to an identity recognition device and method based on breathing-induced electroencephalogram signals. Background Art

[0002] Biometric systems have become the mainstream key unlocking method, but the existing biometric systems include fingerprint recognition, facial recognition, voice recognition and other biometric systems, which have the risk of personal privacy being stolen, copied, synthesized or forged, causing privacy leakage and insecurity.

[0003] EEG signals are non-stationary, nonlinear random signals generated by the cerebral cortex. EEG (Electro-EncephaloGram) signals are generated by the consciousness of individual humans and must be captured by human consciousness. They are unique and have strong anti-counterfeiting and anti-theft capabilities. At present, when using EEG signals induced by vision or smell for identity recognition, it depends on external conditions such as external display devices or specific foods, and the operation is cumbersome and complicated, and is limited to specific environments.

[0004] Based on the above problems, a new identity recognition method is urgently needed to improve security and reduce complexity. Summary of the invention

[0005] The purpose of the present invention is to provide an identity recognition device and method based on breathing induced electroencephalogram signals, which can improve the security of identity recognition and reduce the complexity of recognition.

[0006] To achieve the above object, the present invention provides the following solutions:

[0007] An identity recognition device based on breathing induced electroencephalogram signals, comprising:

[0008] The outer shell is fixedly wrapped around the user's head during operation;

[0009] Electrodes, arranged on the housing, for collecting initial EEG signals induced by the user's breathing;

[0010] A data processing unit, arranged on the housing and connected to the electrodes, for filtering and amplifying the initial EEG signal to obtain a target EEG signal;

[0011] A host computer is connected to the data processing unit and is used to identify the identity of the user according to the target EEG signal.

[0012] Optionally, the shell is in a strip shape.

[0013] Optionally, the data processing unit includes:

[0014] an electrostatic protection module, connected to the electrode, for performing electrostatic protection on the initial EEG signal to obtain a first EEG signal;

[0015] A filtering module, connected to the electrostatic protection module, configured to filter and reduce noise on the first EEG signal to obtain a second EEG signal;

[0016] A preamplifier module, connected to the filtering module, for amplifying the second EEG signal to obtain a target EEG signal;

[0017] A main control module is connected to the preamplifier module and the host computer, and is used to send the target EEG signal to the host computer.

[0018] Optionally, the data processing unit further includes: a voice prompt module;

[0019] The main control module is also used to generate breathing voice instructions;

[0020] The voice prompt module is connected to the main control module, and is used to send fast breathing and slow breathing voice signals to the user according to the breathing voice instructions.

[0021] To achieve the above object, the present invention also provides the following solution:

[0022] An identity recognition method based on respiration-induced electroencephalogram signals, comprising:

[0023] Collecting initial EEG signals induced by the user's breathing;

[0024] Performing filtering and amplification processing on the initial EEG signal to obtain a target EEG signal;

[0025] The identity of the user is identified according to the target EEG signal.

[0026] Optionally, identifying the identity of the user according to the target EEG signal specifically includes:

[0027] Extracting spatial features of the target EEG signal to obtain a target spatial feature matrix;

[0028] Performing smoothing filtering on the target space feature matrix to obtain filtered EEG data features;

[0029] Calculating a stable value of the filtered EEG data feature according to the proportion of data in the filtered EEG data feature;

[0030] Determining whether the stable value is less than a set threshold;

[0031] If the stable value is less than the set threshold, the EEG signal is collected again;

[0032] If the stable value is greater than or equal to the set threshold, then according to the filtered EEG data features, based on the identity recognition model, the matching degree between the filtered EEG data features and each identity identifier stored in the database is determined; the identity recognition model is obtained by pre-training the support vector machine using a training sample set; the training sample set includes multiple groups of sample filtered EEG data features and unique identity identifiers corresponding to the sample filtered EEG data features;

[0033] If the matching degree between the filtered EEG data feature and each identity identifier stored in the database is less than a first matching degree threshold, then the identity recognition fails;

[0034] If the maximum value of the matching degree between the filtered EEG data feature and each identity stored in the database is greater than the second matching degree threshold, the identity recognition is successful; the second matching degree threshold>the first matching degree threshold.

[0035] Optionally, the extracting spatial features of the target EEG signal to obtain a target spatial feature matrix specifically includes:

[0036] Extracting the EEG signal induced by rapid breathing and the EEG signal induced by slow breathing from the target EEG signal to obtain a rapid breathing EEG signal and a slow breathing EEG signal;

[0037] Determine a first covariance matrix according to the rapid breathing EEG signal;

[0038] Determine a second covariance matrix according to the slow breathing EEG signal;

[0039] Performing eigenvalue decomposition on the first covariance matrix and the second covariance matrix respectively to determine a first diagonal matrix, a first eigenvector matrix, a second diagonal matrix and a second eigenvector matrix; the first diagonal matrix is ​​a diagonal matrix formed by the eigenvalues ​​of the first covariance matrix; the first eigenvector matrix is ​​an eigenvector matrix of the first diagonal matrix; the second diagonal matrix is ​​a diagonal matrix formed by the eigenvalues ​​of the second covariance matrix; the second eigenvector matrix is ​​an eigenvector matrix of the second diagonal matrix;

[0040] Determine a first whitening matrix according to the first diagonal matrix and the first eigenvector matrix;

[0041] Determine a second whitening matrix according to the second diagonal matrix and the second eigenvector matrix;

[0042] Determine a first transformation matrix according to the first whitening matrix and the first covariance matrix;

[0043] Determine a second transformation matrix according to the second whitening matrix and the second covariance matrix;

[0044] Performing eigenvalue decomposition on the first transformation matrix to determine a third diagonal matrix and a third eigenvector matrix;

[0045] Performing eigenvalue decomposition on the second transformation matrix to determine a fourth diagonal matrix and a fourth eigenvector matrix;

[0046] Determine a first spatial filter matrix according to the third eigenvector matrix and the first whitening matrix;

[0047] Determine a second spatial filter matrix according to the fourth eigenvector matrix and the second whitening matrix;

[0048] Determining a first spatial feature according to the first spatial filter matrix and the rapid breathing EEG signal;

[0049] Determining a second spatial feature according to the second spatial filter matrix and the slow breathing EEG signal;

[0050] A target spatial feature matrix is ​​determined according to the first spatial feature and the second spatial feature.

[0051] Optionally, the fast breathing EEG signal and the slow breathing EEG signal are both matrices;

[0052] The first covariance matrix is ​​determined using the following formula:

[0053] ;

[0054] in, R 1 is the first covariance matrix, X 1 is the rapid breathing EEG signal, Rapid breathing EEG signal X The transpose of 1, represents the sum of the elements on the diagonal of a matrix.

[0055] Optionally, calculating the stable value of the filtered EEG data feature according to the proportion of data in the filtered EEG data feature specifically includes:

[0056] Dividing the filtered EEG data features into two parts to obtain a first feature set and a second feature set;

[0057] According to the maximum value and the minimum value in the filtered EEG data features, the first feature set and the second feature set are divided into a plurality of intervals respectively;

[0058] Respectively calculating the proportion of data in each interval in the first feature set and the proportion of data in each interval in the second feature set;

[0059] The stable value of the filtered EEG data feature is determined according to the proportion of each interval data in the first feature set and the proportion of each interval data in the second feature set.

[0060] Optionally, the following formula is used to determine the stable value of the filtered EEG data feature:

[0061] ;

[0062] in, PSI is the stable value of the filtered EEG data feature, A i The first feature set i The proportion of interval data, E i The second feature set i The proportion of interval data, n is the number of intervals of the first feature set; the number of intervals of the first feature set is equal to the number of intervals of the second feature set.

[0063] According to the specific embodiments provided by the present invention, the present invention discloses the following technical effects:

[0064] Identity recognition is performed by collecting the EEG signals induced by the user's breathing. Since the EEG signals are captured through consciousness, the privacy leakage problem caused by traditional biometric methods is avoided, and the security of identity recognition is improved. In addition, by inducing EEG signals through breathing, there is no need for additional display equipment, the device is simple, and the complexity of identity recognition is reduced. BRIEF DESCRIPTION OF THE DRAWINGS

[0065] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.

[0066] Figure 1 This is a schematic diagram of the appearance of the identity recognition device based on the breathing induced EEG signal of the present invention;

[0067] Figure 2 This is a schematic diagram of the module structure of the identity recognition device based on the breathing induced EEG signal of the present invention;

[0068] Figure 3 A schematic diagram of the use process of the identity recognition device based on the breathing induced EEG signal of the present invention;

[0069] Figure 4This is a flow chart of the identity recognition method based on breathing induced EEG signals of the present invention;

[0070] Figure 5 A flowchart for identifying the identity of a user based on a target EEG signal;

[0071] Figure 6 Schematic diagram of the hyperplane of support vector machine.

[0072] Explanation of symbols:

[0073] Shell-1, data processing unit-2, electrostatic protection module-21, filter module-22, preamplifier module-23, main control module-24, voice prompt module-25, transmission module-26, power module-27, electrode-3, host computer-4. DETAILED DESCRIPTION

[0074] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0075] The purpose of the present invention is to provide an identity recognition device and method based on breathing induced EEG signals, which collects the user's breathing induced EEG signals for identity recognition, thereby avoiding the problem of privacy leakage caused by traditional biometric methods. The device is simple and reduces the complexity of identity recognition.

[0076] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the present invention is further described in detail below with reference to the accompanying drawings and specific embodiments.

[0077] like Figure 1 and Figure 2 As shown, the identity recognition device based on the breathing induced EEG signal of the present invention includes: a housing 1, electrodes 3, a data processing unit 2 and a host computer 4.

[0078] The housing 1 is fixedly wrapped around the user's head during operation. In this embodiment, the housing 1 is in the shape of a strip, that is, the identity recognition device based on the breathing-induced EEG signal of the present invention is in the shape of a long headband.

[0079] The electrode 3 is arranged on the housing 1, and is used to collect the initial EEG signal induced by the user's breathing.

[0080] The data processing unit 2 is arranged on the housing 1 and connected to the electrode 3. The data processing unit 2 is used for filtering and amplifying the initial EEG signal to obtain a target EEG signal.

[0081] Specifically, Figure 2 As shown, the data processing unit 2 includes: an electrostatic protection module 21 , a filtering module 22 , a preamplifier module 23 and a main control module 24 .

[0082] The electrostatic protection module 21 is connected to the electrode 3 and is used to perform electrostatic protection on the initial EEG signal to obtain a first EEG signal. The electrostatic protection module 21 prevents electrostatic breakdown from damaging the device.

[0083] The filter module 22 is connected to the electrostatic protection module 21, and is used to filter and reduce noise on the first EEG signal to obtain a second EEG signal. In this embodiment, the filter module 22 is a high-frequency low-pass filter to achieve filtering and noise reduction of the EEG signal in the hardware circuit system.

[0084] The preamplifier module 23 is connected to the filter module 22, and is used to amplify the second EEG signal to obtain a target EEG signal.

[0085] The main control module 24 is connected to the preamplifier module 23 and the host computer 4, and is used to send the target EEG signal to the host computer 4. In this embodiment, the main control module 24 sends the target EEG signal to the host computer 4 via WiFi.

[0086] Furthermore, the data processing unit 2 also includes a voice prompt module 25. The main control module 24 is also used to generate breathing voice instructions. The voice prompt module 25 is connected to the main control module 24, and the voice prompt module 25 is used to send fast breathing and slow breathing voice signals to the user according to the breathing voice instructions.

[0087] Optionally, the data processing unit 2 further includes a transmission module 26 and a power module 27. The power module 27 is connected to the main control module 24, and the power module 27 is used to supply power to the main control module 24. The transmission module 26 is connected to the main control module 24 and a WiFi router.

[0088] The host computer 4 is connected to the data processing unit 2, and is used to identify the identity of the user according to the target EEG signal.

[0089] Specifically, the host computer 4 includes: a spatial feature extraction unit, a smoothing filter unit, a stable value calculation unit, a judgment unit, a re-collection unit and an identity matching unit.

[0090] The spatial feature extraction unit is connected to the data processing unit 2, and is used to extract spatial features from the target EEG signal to obtain a target spatial feature matrix.

[0091] The smoothing filter unit is connected to the spatial feature extraction unit, and is used to perform smoothing filtering on the target spatial feature matrix to obtain filtered EEG data features.

[0092] The stable value calculation unit is connected to the smoothing filter unit, and is used to calculate the stable value of the filtered EEG data feature according to the proportion of data in the filtered EEG data feature.

[0093] The judging unit is connected to the stable value calculating unit, and is used to judge whether the stable value is less than a set threshold.

[0094] The re-collection unit is connected to the judgment unit, and when the stable value is less than the set threshold value, the re-collection unit re-collects the electroencephalogram signal.

[0095] The identity matching unit is connected to the judgment unit, and the identity matching unit is used to determine the degree of matching between the filtered EEG data features and each identity stored in the database based on the identity recognition model according to the filtered EEG data features when the stable value is greater than or equal to the set threshold. The identity recognition model is obtained by pre-training the support vector machine using a training sample set. The training sample set includes multiple groups of sample filtered EEG data features and unique identity identifiers corresponding to the sample filtered EEG data features. If the degree of matching between the filtered EEG data features and each identity stored in the database is less than the first matching degree threshold, identity recognition fails. If the maximum value of the degree of matching between the filtered EEG data features and each identity stored in the database is greater than the second matching degree threshold, identity recognition is successful. The second matching degree threshold>the first matching degree threshold.

[0096] Furthermore, the spatial feature extraction unit includes: a signal extraction module, a first covariance matrix determination module, a second covariance matrix determination module, a first eigenvalue decomposition module, a first whitening matrix determination module, a second whitening matrix determination module, a first transformation matrix determination module, a second transformation matrix determination module, a second eigenvalue decomposition module, a third eigenvalue decomposition module, a first filter matrix determination module, a second filter matrix determination module, a first spatial feature determination module, a second spatial feature determination module and a target matrix determination module.

[0097] The signal extraction module is connected to the data processing unit 2, and is used to extract the EEG signal induced by rapid breathing and the EEG signal induced by slow breathing from the target EEG signal to obtain the rapid breathing EEG signal and the slow breathing EEG signal.

[0098] The first covariance matrix determination module is connected to the signal extraction module, and the first covariance matrix determination module is used to determine a first covariance matrix according to the rapid breathing EEG signal.

[0099] The second covariance matrix determination module is connected to the signal extraction module, and the second covariance matrix determination module is used to determine a second covariance matrix according to the slow breathing EEG signal.

[0100] The first eigenvalue decomposition module is connected to the first covariance matrix determination module and the second covariance matrix determination module. The first eigenvalue decomposition module is used to perform eigenvalue decomposition on the first covariance matrix and the second covariance matrix respectively to determine the first diagonal matrix, the first eigenvector matrix, the second diagonal matrix and the second eigenvector matrix; the first diagonal matrix is ​​a diagonal matrix composed of eigenvalues ​​of the first covariance matrix; the first eigenvector matrix is ​​an eigenvector matrix of the first diagonal matrix; the second diagonal matrix is ​​a diagonal matrix composed of eigenvalues ​​of the second covariance matrix; the second eigenvector matrix is ​​an eigenvector matrix of the second diagonal matrix.

[0101] The first whitening matrix determination module is connected to the first eigenvalue decomposition module, and is used to determine a first whitening matrix according to the first diagonal matrix and the first eigenvector matrix.

[0102] The second whitening matrix determination module is connected to the first eigenvalue decomposition module, and is used to determine a second whitening matrix according to the second diagonal matrix and the second eigenvector matrix.

[0103] The first transformation matrix determination module is connected to the first whitening matrix determination module and the first covariance matrix determination module, and the first transformation matrix determination module is used to determine the first transformation matrix according to the first whitening matrix and the first covariance matrix.

[0104] The second transformation matrix determination module is connected to the second whitening matrix determination module and the second covariance matrix determination module, and the second transformation matrix determination module is used to determine the second transformation matrix according to the second whitening matrix and the second covariance matrix.

[0105] The second eigenvalue decomposition module is connected to the first transformation matrix determination module, and is used to perform eigenvalue decomposition on the first transformation matrix to determine a third diagonal matrix and a third eigenvector matrix.

[0106] The third eigenvalue decomposition module is connected to the second transformation matrix determination module, and is used to perform eigenvalue decomposition on the second transformation matrix to determine a fourth diagonal matrix and a fourth eigenvector matrix.

[0107] The first filter matrix determination module is connected to the second eigenvalue decomposition module and the first whitening matrix determination module, and the first filter matrix determination module is used to determine the first spatial filter matrix according to the third eigenvector matrix and the first whitening matrix.

[0108] The second filter matrix determination module is connected to the third eigenvalue decomposition module and the second whitening matrix determination module, and the second filter matrix determination module is used to determine the second spatial filter matrix according to the fourth eigenvector matrix and the second whitening matrix.

[0109] The first spatial feature determination module is connected to the first filter matrix determination module and the signal extraction module, and the first spatial feature determination module is used to determine the first spatial feature according to the first spatial filter matrix and the rapid breathing EEG signal.

[0110] The second spatial feature determination module is connected to the second filter matrix determination module and the signal extraction module, and the second spatial feature determination module is used to determine the second spatial feature according to the second spatial filter matrix and the slow breathing EEG signal.

[0111] The target matrix determination module is connected to the first spatial feature determination module and the second spatial feature determination module, and the target matrix determination module is used to determine the target spatial feature matrix according to the first spatial feature and the second spatial feature.

[0112] like Figure 3 As shown, the working process of the identity recognition device based on breathing induced EEG signals of the present invention is: the user puts on the identity recognition device, turns on the power switch, and performs fast breathing and slow breathing respectively according to voice instructions, collects the EEG signals induced by the user's breathing, and detects the quality of the EEG signals. After the quality is qualified, a database comparison analysis is performed. If the identity recognition fails, you can choose whether to re-identify.

[0113] Since EEG signals are captured through consciousness, the problem of privacy leakage caused by traditional biometrics is solved and it has good security. Moreover, EEG signals are induced by breathing and do not require additional screens or other display devices, making it easy to operate.

[0114] like Figure 4 As shown, the identity recognition method based on the breathing induced EEG signal of the present invention includes:

[0115] S1: Collect the initial EEG signal induced by the user's breathing.

[0116] S2: Filter and amplify the initial EEG signal to obtain a target EEG signal.

[0117] S3: Identify the identity of the user according to the target EEG signal.

[0118] Specifically, S3 includes:

[0119] S31: Extracting spatial features of the target EEG signal to obtain a target spatial feature matrix.

[0120] Furthermore, if Figure 5 As shown, S31 specifically includes:

[0121] (1) Extracting the EEG signal induced by rapid breathing and the EEG signal induced by slow breathing from the target EEG signal to obtain the rapid breathing EEG signal and the slow breathing EEG signal. In this embodiment, the rapid breathing EEG signal and the slow breathing EEG signal are both represented in matrix form.

[0122] (2) Determine a first covariance matrix according to the rapid breathing EEG signal:

[0123] ;

[0124] in, R 1 is the first covariance matrix, X 1 is the rapid breathing EEG signal, Rapid breathing EEG signal X The transpose of 1, represents the sum of the elements on the diagonal of a matrix.

[0125] (3) Determine a second covariance matrix according to the slow breathing EEG signal:

[0126] ;

[0127] in, R 2 is the second covariance matrix, X 2 is the slow breathing EEG signal, Rapid breathing EEG signal X The transpose of 2.

[0128] (4) Performing eigenvalue decomposition on the first covariance matrix and the second covariance matrix respectively to determine a first diagonal matrix, a first eigenvector matrix, a second diagonal matrix, and a second eigenvector matrix. The first diagonal matrix is ​​a diagonal matrix consisting of eigenvalues ​​of the first covariance matrix. The first eigenvector matrix is ​​an eigenvector matrix of the first diagonal matrix. The second diagonal matrix is ​​a diagonal matrix consisting of eigenvalues ​​of the second covariance matrix. The second eigenvector matrix is ​​an eigenvector matrix of the second diagonal matrix.

[0129] Specifically, the formula , determine the first diagonal matrix and the first eigenvector matrix U 1, Represents the first eigenvector matrix U The transpose of 1.

[0130] Using formula , determine the second diagonal matrix and the second eigenvector matrix U 1, Represents the second eigenvector matrix U The transpose of 2.

[0131] As a specific implementation method, firstly adopt the formula R = R 1+ R 2. Determine the mixed space covariance matrix R Then, the mixed space covariance matrix is ​​subjected to eigenvalue decomposition to obtain a mixed diagonal matrix consisting of eigenvalues ​​of the mixed space covariance matrix and a mixed eigenvector matrix of the mixed diagonal matrix. The mixed diagonal matrix includes a first diagonal matrix and a second diagonal matrix; the mixed eigenvector matrix includes a first eigenvector matrix and a second eigenvector matrix.

[0132] (5) Determine a first whitening matrix according to the first diagonal matrix and the first eigenvector matrix P 1: ,in, is the first diagonal matrix The inverse matrix of .

[0133] (6) Determine a second whitening matrix according to the second diagonal matrix and the second eigenvector matrix: P 2: ,in, is the second diagonal matrix The inverse matrix of .

[0134] (7) Determine a first transformation matrix according to the first whitening matrix and the first covariance matrix S 1: ,in, Represents the first whitening matrix P The transpose of 1.

[0135] (8) Determine a second transformation matrix according to the second whitening matrix and the second covariance matrix: S 2: ,in, Represents the second whitening matrix P The transpose of 2.

[0136] (9) Perform eigenvalue decomposition on the first transformation matrix to determine the third diagonal matrix and the third eigenvector matrix. That is, use the formula Determine the third diagonal matrix and the third eigenvector matrix U 3.

[0137] (10) Perform eigenvalue decomposition on the second transformation matrix to determine the fourth diagonal matrix and the fourth eigenvector matrix. That is, the formula is used Determine the fourth diagonal matrix and the third eigenvector matrix U 4. Above U 3= U 4. .

[0138] (11) Determine a first spatial filter matrix according to the third eigenvector matrix and the first whitening matrix W 1: .

[0139] (12) Determine a second spatial filter matrix according to the fourth eigenvector matrix and the second whitening matrix: W 2: .

[0140] (13) Determine a first spatial feature according to the first spatial filter matrix and the rapid breathing EEG signal Z 1: .

[0141] (14) Determine a second spatial feature according to the second spatial filter matrix and the slow breathing EEG signal Z 2: .

[0142] (15) Determine a target spatial feature matrix based on the first spatial feature and the second spatial feature. Z 1 and second space characteristics Z 2 merged into the target space feature matrix Z .

[0143] S32: Perform smoothing filtering on the target space feature matrix to obtain filtered EEG data features.

[0144] When collecting EEG signals, the system noise and background noise inevitably cause isolated points or abnormal values ​​in the EEG signals. To address this problem, this embodiment proposes a filtering method based on feature information. C The mean of the values ​​is used instead of j values, j > C , perform smoothing filtering on the target space feature matrix Z. The specific formula is:

[0145] ;

[0146] in, Y j is the first feature in the filtered EEG data j The eigenvalues ​​of the features, C is the filter order, Z j is the first j value.

[0147] like C =5, indicating 5th order filtering. j =8 means the 8th value, Y 8=1 / 5( Z 8+ Z 7+ Z 6+ Z 5+ Z 4+ Z 3); and so on. j =9 o'clock, Y 9=1 / 5( Z 9+ Z 8+ Z 7+ Z 6+ Z 5+ Z 4) In summary, C =5, the filtered eigenvalue = (current eigenvalue + first four eigenvalues) / 5.

[0148] S33: Calculate the stable value of the filtered EEG data feature according to the proportion of data in the filtered EEG data feature.

[0149] Furthermore, S33 specifically includes:

[0150] The filtered EEG data features are divided into two parts to obtain the first feature set A And the second feature set E .

[0151] According to the maximum value and the minimum value in the filtered EEG data features, the first feature set and the second feature set are respectively divided into a plurality of intervals. In this embodiment, the first feature set and the second feature set are respectively divided into 10 intervals with the same distance.

[0152] The proportion of each interval data in the first feature set and the proportion of each interval data in the second feature set are calculated respectively.

[0153] According to the proportion of each interval data in the first feature set and the proportion of each interval data in the second feature set, the stable value of the filtered EEG data feature is determined. Specifically, the stable value of the filtered EEG data feature is determined using the following formula:

[0154] ;

[0155] in, PSI is the stable value of the filtered EEG data feature, A i The first feature set i The proportion of interval data, E i The second feature set i The proportion of interval data, n is the number of intervals of the first feature set; the number of intervals of the first feature set is equal to the number of intervals of the second feature set.

[0156] S34: Determine whether the stability value is less than a set threshold. If the stability value is less than the set threshold, re-collect the EEG signal. In this embodiment, the set threshold is 0.1.

[0157] S35: If the stability value is greater than or equal to the set threshold, then according to the filtered EEG data features, based on the identity recognition model, determine the matching degree between the filtered EEG data features and each identity identifier stored in the database. The identity recognition model is obtained by pre-training the support vector machine using a training sample set. The training sample set includes multiple groups of sample filtered EEG data features and unique identity identifiers corresponding to the sample filtered EEG data features.

[0158] SVM (Support Vector Machine) is a discriminant classifier defined by a separating hyperplane. The best hyperplane is output based on the training sample data. That is, there will be a hyperplane that separates the training samples, which can be described as:

[0159] ;

[0160] Among them, “·” is the dot product operation, w yes m dimensional vector, b is the offset.

[0161] The hyperplane is not unique, and the optimal hyperplane is the plane that maximizes the distance between the vector closest to each type of data and the hyperplane. Figure 6 As shown, H 1Cannot separate the categories, H 2 can but only with a small interval, H 3 can separate the categories with the largest margin, so H 3 is the optimal hyperplane.

[0162] The optimal hyperplane can be obtained by solving the following quadratic optimization problem:

[0163] ;

[0164] Satisfy the constraints:

[0165] ;

[0166] in, x i It is i input vector, y i yes x i The category label of .

[0167] In the present invention, the features of sample filtered EEG data in the training sample set are used as input, and the categories (identity tags) are used as labels to train the SVM to obtain an identity recognition model.

[0168] The identity recognition of the present invention is a multi-classification problem. The number of identities in the database is the number of categories. The use of SVM is essentially a multiple binary classification problem. One of the identity identifiers is regarded as one category, and all the other identity identifiers are regarded as another category, and so on for multi-classification. The extracted filtered EEG data features are input into the trained identity recognition model for prediction. The model outputs the matching degree between the filtered EEG data features and the existing categories (identity identifiers) in the database, and then the best matching category in the database is obtained to complete the identity recognition.

[0169] S36: If the matching degree between the filtered EEG data feature and each identity identifier stored in the database is less than a first matching degree threshold, identity recognition fails.

[0170] S37: If the maximum value of the matching degree between the filtered EEG data feature and each identity stored in the database is greater than the second matching degree threshold, the identity recognition is successful. The second matching degree threshold>the first matching degree threshold.

[0171] The various embodiments in this specification are described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the various embodiments can be referenced to each other.

[0172] The principles and implementation methods of the present invention are described in this article using specific examples. The description of the above embodiments is only used to help understand the method and core idea of ​​the present invention. At the same time, for those skilled in the art, according to the idea of ​​the present invention, there will be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as limiting the present invention.

Claims

1. An identity recognition device based on breathing induced EEG signals, characterized in that: The identity recognition device based on breathing induced EEG signals comprises: The outer shell is fixedly wrapped around the user's head during operation; Electrodes, arranged on the housing, for collecting initial EEG signals induced by the user's breathing; A data processing unit, arranged on the housing and connected to the electrodes, for filtering and amplifying the initial EEG signal to obtain a target EEG signal; The data processing unit comprises: an electrostatic protection module, connected to the electrode, for performing electrostatic protection on the initial EEG signal to obtain a first EEG signal; A filtering module, connected to the electrostatic protection module, configured to filter and reduce noise on the first EEG signal to obtain a second EEG signal; A preamplifier module, connected to the filtering module, for amplifying the second EEG signal to obtain a target EEG signal; A main control module, connected to the preamplifier module and the host computer, and used to send the target EEG signal to the host computer; A host computer, connected to the data processing unit, for identifying the identity of the user according to the target EEG signal; The host computer includes: a spatial feature extraction unit, a smoothing filter unit, a stable value calculation unit, a judgment unit, a re-collection unit and an identity matching unit; Wherein, the spatial feature extraction unit is connected to the data processing unit, and the spatial feature extraction unit is used to extract the spatial features of the target EEG signal to obtain a target spatial feature matrix; The smoothing filter unit is connected to the spatial feature extraction unit, and the smoothing filter unit is used to perform smoothing filtering on the target spatial feature matrix to obtain filtered EEG data features; The stable value calculation unit is connected to the smoothing filter unit, and the stable value calculation unit is used to calculate the stable value of the filtered EEG data feature according to the proportion of data in the filtered EEG data feature; The judging unit is connected to the stable value calculating unit, and the judging unit is used to judge whether the stable value is less than a set threshold value; The re-collection unit is connected to the judgment unit, and when the stable value is less than the set threshold value, the re-collection unit re-collects the EEG signal; The identity matching unit is connected to the judgment unit, and the identity matching unit is used to determine the matching degree between the filtered EEG data features and each identity identifier stored in the database based on the identity recognition model according to the filtered EEG data features when the stable value is greater than or equal to the set threshold; the identity recognition model is obtained by pre-training the support vector machine with a training sample set; the training sample set includes multiple groups of sample filtered EEG data features and unique identity identifiers corresponding to the sample filtered EEG data features; if the matching degree between the filtered EEG data features and each identity identifier stored in the database is less than a first matching degree threshold, identity recognition fails; if the maximum value of the matching degree between the filtered EEG data features and each identity stored in the database is greater than a second matching degree threshold, identity recognition is successful; the second matching degree threshold> the first matching degree threshold; The working process of the identity recognition device based on breathing induced EEG signals is as follows: the user puts on the identity recognition device, turns on the power switch, and performs fast breathing and slow breathing respectively according to voice commands, collects the user's breathing induced EEG signals, and tests the quality of the EEG signals. After the quality is qualified, a database comparison analysis is performed. If the identity recognition fails, you can choose whether to re-identify.

2. The identity recognition device based on breathing induced EEG signals according to claim 1, characterized in that: The shell is in the shape of a strip.

3. The identity recognition device based on breathing induced EEG signals according to claim 1, characterized in that: The data processing unit also includes: a voice prompt module; The main control module is also used to generate breathing voice instructions; The voice prompt module is connected to the main control module, and is used to send fast breathing and slow breathing voice signals to the user according to the breathing voice instructions.

4. A method for identifying an individual based on a respiration-induced electroencephalogram signal, applied to an identity identification device based on a respiration-induced electroencephalogram signal as claimed in any one of claims 1 to 3, characterized in that: The identity recognition method based on breathing induced EEG signals comprises: Collecting initial EEG signals induced by the user's breathing; Performing filtering and amplification processing on the initial EEG signal to obtain a target EEG signal; The identity of the user is identified according to the target EEG signal.

5. The method for identifying an individual based on a breathing-induced EEG signal according to claim 4, characterized in that: The identifying of the user according to the target EEG signal specifically includes: Extracting spatial features of the target EEG signal to obtain a target spatial feature matrix; Performing smoothing filtering on the target space feature matrix to obtain filtered EEG data features; Calculating a stable value of the filtered EEG data feature according to the proportion of data in the filtered EEG data feature; Determining whether the stable value is less than a set threshold; If the stable value is less than the set threshold, the EEG signal is collected again; If the stable value is greater than or equal to the set threshold, then according to the filtered EEG data features, based on the identity recognition model, the matching degree between the filtered EEG data features and each identity identifier stored in the database is determined; the identity recognition model is obtained by pre-training the support vector machine using a training sample set; the training sample set includes multiple groups of sample filtered EEG data features and unique identity identifiers corresponding to the sample filtered EEG data features; If the matching degree between the filtered EEG data feature and each identity identifier stored in the database is less than a first matching degree threshold, then the identity recognition fails; If the maximum value of the matching degree between the filtered EEG data feature and each identity stored in the database is greater than the second matching degree threshold, the identity recognition is successful; the second matching degree threshold>the first matching degree threshold.

6. The method for identifying an individual based on a breathing-induced EEG signal according to claim 5, characterized in that: The spatial feature extraction of the target EEG signal to obtain a target spatial feature matrix specifically includes: Extracting the EEG signal induced by rapid breathing and the EEG signal induced by slow breathing from the target EEG signal to obtain a rapid breathing EEG signal and a slow breathing EEG signal; Determine a first covariance matrix according to the rapid breathing EEG signal; Determine a second covariance matrix according to the slow breathing EEG signal; Performing eigenvalue decomposition on the first covariance matrix and the second covariance matrix respectively to determine a first diagonal matrix, a first eigenvector matrix, a second diagonal matrix and a second eigenvector matrix; the first diagonal matrix is ​​a diagonal matrix formed by the eigenvalues ​​of the first covariance matrix; the first eigenvector matrix is ​​an eigenvector matrix of the first diagonal matrix; the second diagonal matrix is ​​a diagonal matrix formed by the eigenvalues ​​of the second covariance matrix; the second eigenvector matrix is ​​an eigenvector matrix of the second diagonal matrix; Determine a first whitening matrix according to the first diagonal matrix and the first eigenvector matrix; Determine a second whitening matrix according to the second diagonal matrix and the second eigenvector matrix; Determine a first transformation matrix according to the first whitening matrix and the first covariance matrix; Determine a second transformation matrix according to the second whitening matrix and the second covariance matrix; Performing eigenvalue decomposition on the first transformation matrix to determine a third diagonal matrix and a third eigenvector matrix; Performing eigenvalue decomposition on the second transformation matrix to determine a fourth diagonal matrix and a fourth eigenvector matrix; Determine a first spatial filter matrix according to the third eigenvector matrix and the first whitening matrix; Determine a second spatial filter matrix according to the fourth eigenvector matrix and the second whitening matrix; Determining a first spatial feature according to the first spatial filter matrix and the rapid breathing EEG signal; Determining a second spatial feature according to the second spatial filter matrix and the slow breathing EEG signal; A target spatial feature matrix is ​​determined according to the first spatial feature and the second spatial feature.

7. The method for identifying an individual based on a breathing-induced EEG signal according to claim 6, characterized in that: The fast breathing EEG signal and the slow breathing EEG signal are both matrices; The first covariance matrix is ​​determined using the following formula: Among them, R1 is the first covariance matrix, X1 is the rapid breathing EEG signal, is the transpose of the rapid breathing EEG signal X1, and trace() represents the sum of the elements on the diagonal of the matrix.

8. The method for identifying an individual based on a breathing-induced EEG signal according to claim 6, characterized in that: The step of calculating the stable value of the filtered EEG data feature according to the proportion of data in the filtered EEG data feature specifically includes: Dividing the filtered EEG data features into two parts to obtain a first feature set and a second feature set; According to the maximum value and the minimum value in the filtered EEG data features, the first feature set and the second feature set are divided into a plurality of intervals respectively; Respectively calculating the proportion of data in each interval in the first feature set and the proportion of data in each interval in the second feature set; The stable value of the filtered EEG data feature is determined according to the proportion of each interval data in the first feature set and the proportion of each interval data in the second feature set.

9. The method for identifying an individual based on a breathing-induced EEG signal according to claim 8, characterized in that: The following formula is used to determine the stable value of the filtered EEG data feature: Among them, PSI is the stable value of the filtered EEG data feature, A i is the proportion of the data in the i-th interval in the first feature set, E i is the proportion of the i-th interval data in the second feature set, n is the number of intervals in the first feature set; the number of intervals in the first feature set is equal to the number of intervals in the second feature set.

Images