Brain pattern recognition headband device for synchronously collecting electroencephalogram signals and near-infrared signals

Abstract

The invention discloses a brain grain recognition headband device for synchronously collecting electroencephalogram signals and near-infrared signals, and relates to the field of brain grain recognition. Comprising a bandage, a fixing buckle A and a fixing buckle B which are matched with each other are arranged at the two ends of the bandage respectively, a plurality of electrode assemblies are arranged in the bandage, and a flexible chip electrically connected with the electrode assemblies is fixed in the bandage. The electrode assembly comprises a base and a plurality of electrode claws, the electrode claws penetrate through the bandage and are attached to the brain skin, and the working method comprises the following steps that preparation work is conducted, the brain of a testee is cleaned, the head-mounted device is worn, then near-infrared spectrum imaging equipment is arranged, a light source of the near-infrared spectrum imaging equipment is adjusted, and the near-infrared spectrum imaging equipment is used for imaging the brain of the testee. The area where the head-mounted device is located is covered. According to the invention, the electroencephalogram signal and the near-infrared signal are synchronously acquired, and then the computer is used for feature extraction and fusion, so that the identification and verification of the brain electric potential and the brain grain of the living body can be realized, and the use convenience is also improved by adopting a head-mounted mode.

Description

Technical Field

[0001] The present invention relates to a field, and in particular to a brain pattern recognition headband device for synchronously collecting electroencephalogram signals and near-infrared signals. Background Art

[0002] The concept of brainprint was proposed by Sarah Laszlo's research group at Binghamton University in the United States in 2014. It refers to a unique, collectible and persistent EEG signal feature produced by the brain that can be used for identity recognition and verification. Unlike biometric features such as fingerprints that are easily affected by external factors, EEG signals are relatively stable.

[0003] Research on identity recognition and authentication based on EEG signals can be roughly divided into four categories: resting state, visual evoked, motor imagery, and event-related potentials. Resting state recognition is based on the phenomenon that the EEG signals generated by different people's brains in a completely resting state are different. Research on personal identity recognition using EEG signals in a resting state. Paranjape et al. first used EEG signals with eyes open and closed in a resting state as biometrics for identity recognition, with an accuracy rate of over 80%. Liu Quanying et al. proposed an identity recognition method based on single-channel EEG in a resting state acquisition environment, with an accuracy rate of over 90%. Visual evoked potential refers to a specific activity generated by the nervous system receiving visual stimulation. Visual evoked potential occurs at a specific time and location, is relatively easy to detect, and is suitable for brain-computer interfaces. In 2009, Das et al. used EEG signals based on visual evoked potentials to conduct identity feature recognition research, with an identification accuracy rate of 91%. Paranjape et al. extracted feature vectors through power spectrum, used K-nearest neighbor algorithm and Elman neural network for classification, and the average recognition accuracy rates were 96.13% and 98.12% respectively. Motor imagery EEG is the EEG pattern when imagining a certain limb movement. Marcel et al. proposed a statistical framework based on Gaussian mixture model and maximum a posteriori probability model to conduct statistics on the false acceptance rate and false rejection rate, with an average error rate of 7.1%. Hu Jianfeng et al. used autoregressive model, signal power spectrum, complexity and other methods to extract features from motor imagery EEG signals, and finally used BP neural network to identify the feature vector after dimension reduction. The average recognition rates of imagining left hand, right hand, leg and tongue movements were 81.2%, 82.1%, 82.8% and 90.6% respectively. Event-related potential is a special evoked potential. It is the brain potential recorded from the surface of the skull through average superposition when a person performs cognitive processing on an object. It reflects the neuroelectrophysiological changes of the brain during the cognitive process. Yeom et al. proposed an EEG identity recognition paradigm with special response to self-face pictures, with an average accuracy of 86.1%. With the introduction of the concept of brainprints, Armstrong et al. used ERP components as biometric features for identity recognition, with an accuracy rate of 82%-97%. Six months later, the EEG recognition accuracy of some participants remained unchanged. Liu Jing et al. from Tianjin University verified the feasibility of identity recognition using event-related potentials induced by the Farwell paradigm, and the results showed that the accuracy of the ERP features (P300) of target and non-target stimuli could reach 92.50% and 93.27%, respectively.

[0004] Therefore, how to achieve living identity recognition through brain pattern information analysis based on brain potential is a technical problem that needs to be solved urgently.

[0005] To this end, the present invention proposes a brain pattern recognition headband device for synchronously collecting electroencephalogram signals and near-infrared signals. Summary of the invention

[0006] The purpose of the present invention is to solve the shortcomings of the prior art and to propose a brain pattern recognition headband device for synchronously collecting electroencephalogram signals and near-infrared signals.

[0007] In order to achieve the above object, the present invention adopts the following technical solutions:

[0008] A brain pattern recognition headband device for synchronously collecting electroencephalogram signals and near-infrared signals comprises a strap, two ends of which are respectively provided with a fixing buckle A and a fixing buckle B that cooperate with each other, a plurality of electrode assemblies are arranged inside the strap, and a flexible chip electrically connected to the electrode assemblies is fixed inside the strap.

[0009] Preferably, the electrode assembly comprises a base and a plurality of electrode claws, and the electrode claws penetrate through the strap to fit the brain skin.

[0010] Preferably: the working method comprises the following steps:

[0011] S1: Preparation: clean the subject's brain and put on the head-mounted device, then set up the near-infrared spectral imaging device and adjust the light source of the spectral imaging device to cover the area where the head-mounted device is located;

[0012] S2: Signal acquisition: start the head-mounted device and the near-infrared spectral imaging device, connect them to the computer, let the subjects perform cognitive tasks, and the computer collects signals from the head-mounted device and the near-infrared spectral imaging device;

[0013] S3: data preprocessing, preprocessing the signal;

[0014] S4: feature extraction, feature extraction of the preprocessed signal;

[0015] S5: Fusion and classification, using data fusion technology to fuse the two signal features into a comprehensive feature vector;

[0016] S6: Compare the brainprint features of the individual to be identified with the brainprint features in the database, and determine the identity based on the comparison results.

[0017] Preferably: in the step S2, the cognitive task is any one of resting state, visual induction, motor imagery, and event-related.

[0018] Preferably: in the step S3, data preprocessing includes the following steps:

[0019] S31: filtering the data of the head mounted device;

[0020] S32: Then remove the eye movement and electromyography artifacts;

[0021] S33: De-noise near-infrared spectral imaging equipment to improve the quality of blood oxygen-dependent signals;

[0022] S34: Align the timing of the two signals to ensure data synchronization.

[0023] Preferably: in the step S4, the feature extraction wavelet transform performs multi-scale analysis on the signal, and its formula is W(a,b)=\frac{1}{\sqrt{a}}\int_{-\infty}^{\infty}x(t)\psi\le ft(\frac{tb}{a}\right)dt\], wherein $x(t)$ is the original EEG signal, $\psi(t)$ is the wavelet function, $a$ is the scale factor, and $b$ is the translation factor.

[0024] Preferably: in the step S5, feature fusion includes the following steps:

[0025] S51: Data standardization: standardize the data so that they are within the same numerical range;

[0026] S52: Calculate the covariance matrix, construct the covariance matrix of the data, and form the correlation between different features. The formula is

[0027] S53: Solve for eigenvalues ​​and eigenvectors, solve |C-\λI|=0 to obtain eigenvalues, and then solve (C-λI)v=0 to obtain eigenvector v;

[0028] S54: Select the principal component, select the principal component according to the size of the eigenvalue, and the eigenvalue is calculated according to the formula Calculate, where mm is the number of original features;

[0029] S55: Construct a new feature space, project the original data onto the selected principal component, and form the reduced-dimensional data. The formula is y k =V k1 x1+V k2 x2+......+V kn x n ;

[0030] S56: Fusion process: Combine feature sets from different sources and reduce the dimension through PCA to output the fused new features.

[0031] Preferably: in the step S51, the formula for data standardization is x′ is the data after standardization, x is the data before standardization, μ is the variance of the data before standardization, and σ is the mean.

[0032] Preferably: in the step S54, p∈(0.9, 1).

[0033] The beneficial effects of the present invention are:

[0034] 1. The present invention collects EEG signals and near-infrared signals synchronously, and then uses a computer to extract and fuse features, so as to realize the recognition and verification of living brain potential and brain pattern, and adopts a head-mounted approach, which also increases the convenience of use. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] Figure 1 This is an overall schematic diagram of a brain pattern recognition headband device for synchronously collecting EEG signals and near-infrared signals proposed by the present invention;

[0036] Figure 2 A schematic diagram of an electrode assembly of a brain pattern recognition headband device for synchronously collecting EEG signals and near-infrared signals proposed by the present invention;

[0037] Figure 3 This is a working logic diagram of a brain pattern recognition headband device for synchronously collecting EEG signals and near-infrared signals proposed by the present invention;

[0038] In the figure: 1- strap, 2- fixing buckle A, 3-, flexible chip, 4- electrode assembly, 5- fixing buckle B, 6- base, 7- electrode claw. DETAILED DESCRIPTION

[0039] The technical solution of the present invention is further described in detail below in conjunction with specific implementation methods.

[0040] In the description of the present invention, it should be noted that, unless otherwise clearly specified and limited, the terms "installed", "connected", "connected", and "set" should be understood in a broad sense, for example, they can be fixedly connected or set, or detachably connected or set, or integrally connected or set. For ordinary technicians in this field, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0041] Embodiment 1:

[0042] A brain pattern recognition headband device for synchronously collecting electroencephalogram signals and near-infrared signals comprises a binding strap (1), wherein two ends of the binding strap (1) are respectively provided with a fixing buckle A (2) and a fixing buckle B (5) that cooperate with each other, a plurality of electrode assemblies (4) are arranged inside the binding strap (1), and a flexible chip (3) electrically connected to the electrode assemblies (4) is fixed inside the binding strap (1).

[0043] The electrode assembly (4) comprises a base (6) and a plurality of electrode claws (7), and the electrode claws (7) penetrate the bandage (1) and adhere to the brain skin.

[0044] When the present embodiment is in use, the strap (1) can be wrapped around the head and then fastened together by the fixing buckle A (2) and the fixing buckle B (5). The electrode claw (7) is attached to the scalp to collect brainprint signals, which are then transmitted to the flexible chip (3) for analysis.

[0045] Embodiment 2:

[0046] A headband device for brain pattern recognition for synchronously collecting electroencephalogram signals and near-infrared signals, the working method of which comprises the following steps:

[0047] S1: Preparation: clean the subject's brain and put on the head-mounted device, then set up the near-infrared spectral imaging device and adjust the light source of the spectral imaging device to cover the area where the head-mounted device is located;

[0048] S2: Signal acquisition: start the head-mounted device and the near-infrared spectral imaging device, connect them to the computer, let the subjects perform cognitive tasks, and the computer collects signals from the head-mounted device and the near-infrared spectral imaging device;

[0049] S3: data preprocessing, preprocessing the signal;

[0050] S4: feature extraction, feature extraction of the preprocessed signal;

[0051] S5: Fusion and classification, using data fusion technology to fuse the two signal features into a comprehensive feature vector;

[0052] S6: Compare the brainprint features of the individual to be identified with the brainprint features in the database, and determine the identity based on the comparison results.

[0053] In the step S2, the cognitive task is in a resting state.

[0054] In the step S3, data preprocessing includes the following steps:

[0055] S31: filtering the data of the head mounted device;

[0056] S32: Then remove the eye movement and electromyography artifacts;

[0057] S33: De-noise near-infrared spectral imaging equipment to improve the quality of blood oxygen-dependent signals;

[0058] S34: Align the timing of the two signals to ensure data synchronization.

[0059] In the step S4, the feature extraction wavelet transform performs multi-scale analysis on the signal, and its formula is W(a,b)=\frac{1}{\sqrt{a}}\int_{-\infty}^{\infty}x(t)\psi\le ft(\frac{tb}{a}\right)dt\], where $x(t)$ is the original EEG signal, $\psi(t)$ is the wavelet function, $a$ is the scale factor, and $b$ is the translation factor.

[0060] In the step S5, feature fusion includes the following steps:

[0061] S51: Data standardization: standardize the data so that they are within the same numerical range;

[0062] S52: Calculate the covariance matrix, construct the covariance matrix of the data, and form the correlation between different features. The formula is

[0063] S53: Solve for eigenvalues ​​and eigenvectors, solve |C-\λI|=0 to obtain eigenvalues, and then solve (C-λI)v=0 to obtain eigenvector v;

[0064] S54: Select the principal component, select the principal component according to the size of the eigenvalue, and the eigenvalue is calculated according to the formula Calculate, where mm is the number of original features;

[0065] S55: Construct a new feature space, project the original data onto the selected principal component, and form the reduced-dimensional data. The formula is y k =V k1 x1+V k2 x2+......+V kn x n ;

[0066] S56: Fusion process: Combine feature sets from different sources and reduce the dimension through PCA to output the fused new features.

[0067] In the step S51, the formula for data standardization is: x′ is the data after standardization, x is the data before standardization, μ is the variance of the data before standardization, and σ is the mean.

[0068] In the step S54, p∈(0.9, 1).

[0069] Embodiment 3:

[0070] A headband device for brain pattern recognition for synchronously collecting EEG signals and near-infrared signals, a headband device for brain pattern recognition for synchronously collecting EEG signals and near-infrared signals, and a working method thereof comprises the following steps:

[0071] S1: Preparation: clean the subject's brain and put on the head-mounted device, then set up the near-infrared spectral imaging device and adjust the light source of the spectral imaging device to cover the area where the head-mounted device is located;

[0072] S2: Signal acquisition: start the head-mounted device and the near-infrared spectral imaging device, connect them to the computer, let the subjects perform cognitive tasks, and the computer collects signals from the head-mounted device and the near-infrared spectral imaging device;

[0073] S3: data preprocessing, preprocessing the signal;

[0074] S4: feature extraction, feature extraction of the preprocessed signal;

[0075] S5: Fusion and classification, using data fusion technology to fuse the two signal features into a comprehensive feature vector;

[0076] S6: Compare the brainprint features of the individual to be identified with the brainprint features in the database, and determine the identity based on the comparison results.

[0077] In the step S2, the cognitive task is visual induction.

[0078] In the step S3, data preprocessing includes the following steps:

[0079] S31: filtering the data of the head mounted device;

[0080] S32: Then remove the eye movement and electromyography artifacts;

[0081] S33: De-noise near-infrared spectral imaging equipment to improve the quality of blood oxygen-dependent signals;

[0082] S34: Align the timing of the two signals to ensure data synchronization.

[0083] In the step S4, the feature extraction wavelet transform performs multi-scale analysis on the signal, and its formula is W(a,b)=\frac{1}{\sqrt{a}}\int_{-\infty}^{\infty}x(t)\psi\le ft(\frac{tb}{a}\right)dt\], where $x(t)$ is the original EEG signal, $\psi(t)$ is the wavelet function, $a$ is the scale factor, and $b$ is the translation factor.

[0084] In the step S5, feature fusion includes the following steps:

[0085] S51: Data standardization: standardize the data so that they are within the same numerical range;

[0086] S52: Calculate the covariance matrix, construct the covariance matrix of the data, and form the correlation between different features. The formula is

[0087] S53: Solve for eigenvalues ​​and eigenvectors, solve |C-\λI|=0 to obtain eigenvalues, and then solve (C-λI)v=0 to obtain eigenvector v;

[0088] S54: Select the principal component, select the principal component according to the size of the eigenvalue, and the eigenvalue is calculated according to the formula Calculate, where mm is the number of original features;

[0089] S55: Construct a new feature space, project the original data onto the selected principal component, and form the reduced-dimensional data. The formula is y k =V k1 x1+V k2 x2+......+V kn x n ;

[0090] S56: Fusion process: Combine feature sets from different sources and reduce the dimension through PCA to output the fused new features.

[0091] In the step S51, the formula for data standardization is: x′ is the data after standardization, x is the data before standardization, μ is the variance of the data before standardization, and σ is the mean.

[0092] In the step S54, p∈(0.9, 1).

[0093] Embodiment 4:

[0094] A headband device for brain pattern recognition for synchronously collecting EEG signals and near-infrared signals, a headband device for brain pattern recognition for synchronously collecting EEG signals and near-infrared signals, and a working method thereof comprises the following steps:

[0095] S1: Preparation: clean the subject's brain and put on the head-mounted device, then set up the near-infrared spectral imaging device and adjust the light source of the spectral imaging device to cover the area where the head-mounted device is located;

[0096] S2: Signal acquisition: start the head-mounted device and the near-infrared spectral imaging device, connect them to the computer, let the subjects perform cognitive tasks, and the computer collects signals from the head-mounted device and the near-infrared spectral imaging device;

[0097] S3: data preprocessing, preprocessing the signal;

[0098] S4: feature extraction, feature extraction of the preprocessed signal;

[0099] S5: Fusion and classification, using data fusion technology to fuse the two signal features into a comprehensive feature vector;

[0100] S6: Compare the brainprint features of the individual to be identified with the brainprint features in the database, and determine the identity based on the comparison results.

[0101] In the step S2, the cognitive task is motor imagery.

[0102] In the step S3, data preprocessing includes the following steps:

[0103] S31: filtering the data of the head mounted device;

[0104] S32: Then remove the eye movement and electromyography artifacts;

[0105] S33: De-noise near-infrared spectral imaging equipment to improve the quality of blood oxygen-dependent signals;

[0106] S34: Align the timing of the two signals to ensure data synchronization.

[0107] In the step S4, the feature extraction wavelet transform performs multi-scale analysis on the signal, and its formula is W(a,b)=\frac{1}{\sqrt{a}}\int_{-\infty}^{\infty}x(t)\psi\le ft(\frac{tb}{a}\right)dt\], where $x(t)$ is the original EEG signal, $\psi(t)$ is the wavelet function, $a$ is the scale factor, and $b$ is the translation factor.

[0108] In the step S5, feature fusion includes the following steps:

[0109] S51: Data standardization: standardize the data so that they are within the same numerical range;

[0110] S52: Calculate the covariance matrix, construct the covariance matrix of the data, and form the correlation between different features. The formula is

[0111] S53: Solve for eigenvalues ​​and eigenvectors, solve |C-\λI|=0 to obtain eigenvalues, and then solve (C-λI)v=0 to obtain eigenvector v;

[0112] S54: Select the principal component, select the principal component according to the size of the eigenvalue, and the eigenvalue is calculated according to the formula Calculate, where mm is the number of original features;

[0113] S55: Construct a new feature space, project the original data onto the selected principal component, and form the reduced-dimensional data. The formula is y k =Vk1 x1+V k2 x2+......+V kn x n ;

[0114] S56: Fusion process: Combine feature sets from different sources and reduce the dimension through PCA to output the fused new features.

[0115] In the step S51, the formula for data standardization is: x′ is the data after standardization, x is the data before standardization, μ is the variance of the data before standardization, and σ is the mean.

[0116] In the step S54, p∈(0.9, 1).

[0117] Embodiment 5:

[0118] A headband device for brain pattern recognition for synchronously collecting EEG signals and near-infrared signals, a headband device for brain pattern recognition for synchronously collecting EEG signals and near-infrared signals, and a working method thereof comprises the following steps:

[0119] S1: Preparation: clean the subject's brain and put on the head-mounted device, then set up the near-infrared spectral imaging device and adjust the light source of the spectral imaging device to cover the area where the head-mounted device is located;

[0120] S2: Signal acquisition: start the head-mounted device and the near-infrared spectral imaging device, connect them to the computer, let the subjects perform cognitive tasks, and the computer collects signals from the head-mounted device and the near-infrared spectral imaging device;

[0121] S3: data preprocessing, preprocessing the signal;

[0122] S4: feature extraction, feature extraction of the preprocessed signal;

[0123] S5: Fusion and classification, using data fusion technology to fuse the two signal features into a comprehensive feature vector;

[0124] S6: Compare the brainprint features of the individual to be identified with the brainprint features in the database, and determine the identity based on the comparison results.

[0125] In the step S2, the cognitive task is event-related.

[0126] In the step S3, data preprocessing includes the following steps:

[0127] S31: filtering the data of the head mounted device;

[0128] S32: Then remove the eye movement and electromyography artifacts;

[0129] S33: De-noise near-infrared spectral imaging equipment to improve the quality of blood oxygen-dependent signals;

[0130] S34: Align the timing of the two signals to ensure data synchronization.

[0131] In the step S4, the feature extraction wavelet transform performs multi-scale analysis on the signal, and its formula is W(a,b)=\frac{1}{\sqrt{a}}\int_{-\infty}^{\infty}x(t)\psi\le ft(\frac{tb}{a}\right)dt\], where $x(t)$ is the original EEG signal, $\psi(t)$ is the wavelet function, $a$ is the scale factor, and $b$ is the translation factor.

[0132] In the step S5, feature fusion includes the following steps:

[0133] S51: Data standardization: standardize the data so that they are within the same numerical range;

[0134] S52: Calculate the covariance matrix, construct the covariance matrix of the data, and form the correlation between different features. The formula is

[0135] S53: Solve for eigenvalues ​​and eigenvectors, solve |C-\λI|=0 to obtain eigenvalues, and then solve (C-λI)v=0 to obtain eigenvector v;

[0136] S54: Select the principal component, select the principal component according to the size of the eigenvalue, and the eigenvalue is calculated according to the formula Calculate, where mm is the number of original features;

[0137] S55: Construct a new feature space, project the original data onto the selected principal component, and form the reduced-dimensional data. The formula is y k =V k1 x1+V k2 x2+......+V kn x n ;

[0138] S56: Fusion process: Combine feature sets from different sources and reduce the dimension through PCA to output the fused new features.

[0139] In the step S51, the formula for data standardization is: x′ is the data after standardization, x is the data before standardization, μ is the variance of the data before standardization, and σ is the mean.

[0140] In the step S54, p∈(0.9, 1).

[0141] The above description is only a preferred specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can make equivalent replacements or changes according to the technical scheme and inventive concept of the present invention within the technical scope disclosed by the present invention, which should be covered by the protection scope of the present invention.

Claims

1. A headband device for brain pattern recognition that synchronously collects EEG signals and near-infrared signals, characterized in that: The invention comprises a binding strap (1), wherein two ends of the binding strap (1) are respectively provided with a fixing buckle A (2) and a fixing buckle B (5) which cooperate with each other, a plurality of electrode assemblies (4) are arranged inside the binding strap (1), and a flexible chip (3) which is electrically connected to the electrode assemblies (4) is fixed inside the binding strap (1).

2. The headband device for brain pattern recognition for synchronous acquisition of EEG signals and near infrared signals according to claim 1, characterized in that: The electrode assembly (4) comprises a base (6) and a plurality of electrode claws (7), and the electrode claws (7) penetrate the bandage (1) and adhere to the brain skin.

3. A headband device for brain pattern recognition for synchronous acquisition of EEG signals and near infrared signals according to claim 1 or 2, characterized in that: Its working method includes the following steps: S1: Preparation: clean the subject's brain and put on the head-mounted device, then set up the near-infrared spectral imaging device and adjust the light source of the spectral imaging device to cover the area where the head-mounted device is located; S2: Signal acquisition: start the head-mounted device and the near-infrared spectral imaging device, connect them to the computer, let the subjects perform cognitive tasks, and the computer collects signals from the head-mounted device and the near-infrared spectral imaging device; S3: data preprocessing, preprocessing the signal; S4: feature extraction, feature extraction of the preprocessed signal; S5: Fusion and classification, using data fusion technology to fuse the two signal features into a comprehensive feature vector; S6: Compare the brainprint features of the individual to be identified with the brainprint features in the database, and determine the identity based on the comparison results.

4. The headband device for brain pattern recognition for synchronous acquisition of EEG signals and near infrared signals according to claim 3, characterized in that: In the step S2, the cognitive task is any one of resting state, visual induction, motor imagery, and event-related.

5. The headband device for brain pattern recognition for synchronous acquisition of EEG signals and near infrared signals according to claim 3, characterized in that: In the step S3, data preprocessing includes the following steps: S31: filtering the data of the head mounted device; S32: Then remove the eye movement and electromyography artifacts; S33: De-noise near-infrared spectral imaging equipment to improve the quality of blood oxygen-dependent signals; S34: Align the timing of the two signals to ensure data synchronization.

6. The headband device for brain pattern recognition for synchronous acquisition of EEG signals and near infrared signals according to claim 3, characterized in that: In the step S4, the feature extraction wavelet transform performs multi-scale analysis on the signal, and its formula is W(a,b)=\frac{1}{\sqrt{a}}\int_{-\infty}^{\infty}x(t)\psi\le ft(\frac{tb}{a}\right)dt\], where $x(t)$ is the original EEG signal, $\psi(t)$ is the wavelet function, $a$ is the scale factor, and $b$ is the translation factor.

7. The headband device for brain pattern recognition for synchronous acquisition of EEG signals and near infrared signals according to claim 3, characterized in that: In the step S5, feature fusion includes the following steps: S51: Data standardization: standardize the data so that they are within the same numerical range; S52: Calculate the covariance matrix, construct the covariance matrix of the data, and form the correlation between different features. The formula is S53: Solve for eigenvalues ​​and eigenvectors, solve |C-\λI|=0 to obtain eigenvalues, and then solve (C-λI)v=0 to obtain eigenvector v; S54: Select the principal component, select the principal component according to the size of the eigenvalue, and the eigenvalue is calculated according to the formula Calculate, where mm is the number of original features; S55: Construct a new feature space, project the original data onto the selected principal component, and form the reduced-dimensional data. The formula is y k =V k1 x1+V k2 x2+......+V kn x n ; S56: Fusion process: Combine feature sets from different sources and reduce the dimension through PCA to output the fused new features.

8. The headband device for brain pattern recognition for synchronous acquisition of EEG signals and near infrared signals according to claim 7, characterized in that: In the step S51, the formula for data standardization is: x′ is the data after standardization, x is the data before standardization, μ is the variance of the data before standardization, and σ is the mean.

9. The headband device for brain pattern recognition for synchronous acquisition of EEG signals and near infrared signals according to claim 7, characterized in that: In the step S54, p∈(0.9, 1).

Images