An adaptive denoising method for EEG signals in ubiquitous environments

Abstract

The present application discloses an adaptive denoising method, system and device for EEG signals in a ubiquitous environment, which relates to the field of signal processing technology and is used to improve the accuracy of removing noise signals from EEG signals. The method comprises: performing blind source signal separation processing on the signal to be processed to obtain the source component of the signal. According to the information entropy and the first threshold of the source component, the first source component corresponding to the aliasing signal of the noise signal and the EEG signal is determined. The first source component is processed based on the multivariate variational mode decomposition method to obtain the multivariate intrinsic mode function corresponding to the first source component, and the multivariate intrinsic mode function is clustered to determine the intrinsic mode function corresponding to the EEG signal. The intrinsic mode function corresponding to the EEG signal is inversely reconstructed to restore the original EEG signal.

Description

Technical Field

[0001] The present application relates to the field of signal processing technology, and in particular to an adaptive denoising method for electroencephalogram (EEG) signals in a ubiquitous environment. Background Art

[0002] EEG signals are a type of bioelectric signal generated by brain nerve cells. They contain rich information about thinking activities, emotional memory, and behavioral cognition, and play an important role in research fields such as clinical medicine, cognitive science, and brain-computer interfaces. However, the EEG signals collected in a universal environment are very susceptible to noise interference, resulting in a large amount of noise signals in the collected signals, reducing the accuracy of data analysis and processing.

[0003] However, the accuracy of denoising the EEG signals collected in a universal environment is currently not high and needs to be improved. Summary of the invention

[0004] The present invention provides an adaptive denoising method and device for EEG signals in a ubiquitous environment, so as to improve the accuracy of denoising processing for EEG signals in a ubiquitous environment.

[0005] In the first aspect, an embodiment of the present application provides an adaptive denoising method for EEG signals in a ubiquitous environment, comprising: obtaining a signal to be processed, the signal to be processed comprising a noise signal and an EEG signal. Performing blind source signal separation processing on the signal to be processed to obtain multiple source component vectors (SCV). According to the information entropy of the multiple source component components and a first threshold, at least one first source component component is determined from the multiple source component components, the information entropy is used to indicate the complexity of the signal, and the signal corresponding to the first source component component comprises a noise signal and an EEG signal. The first source component component is processed based on a multivariate variational mode decomposition (MVMD) method to obtain multiple multivariate intrinsic mode functions (MIMFs). Clustering processing is performed on the multiple multivariate intrinsic mode functions to determine at least one multivariate intrinsic mode function corresponding to the EEG signal. Inverse reconstruction processing is performed on the multivariate intrinsic mode function corresponding to the EEG signal to obtain the EEG signal.

[0006] By adopting this method, the first device can efficiently and accurately determine the noise signal in the EEG signal collected in the universal environment according to the information entropy of the source component and the first threshold, so as to efficiently and accurately remove the noise signal in the EEG signal collected in the universal environment, and improve the accuracy of denoising the EEG signal collected in the universal environment. In addition, the first device performs clustering processing on the multivariate intrinsic mode function corresponding to the EEG signal, adaptively identifies the noise signal, and efficiently removes the noise signal of the EEG signal.

[0007] In a possible design, the waveform factor and kurtosis factor of the collected signal are obtained, and the signal is determined to be the signal to be processed according to the waveform factor and kurtosis factor of the signal and a preset threshold.

[0008] Based on this design, the noise signal in the frame signal can be accurately determined according to the waveform factor and kurtosis factor of the signal.

[0009] In one possible design, the information entropy of the source component component includes the fuzzy entropy of the source component component.

[0010] In one possible design, the fuzzy entropy of the source component satisfies:

[0011]

[0012] Among them, H(m) represents the fuzzy entropy of the source component, N is the number of subsequences of the source component, U(i,j) represents the fuzzy similarity between the i-th subsequence and the j-th subsequence, and ∑ is the accumulation symbol.

[0013] In one possible design, the energy entropy and autocorrelation coefficient of the multivariate intrinsic mode function in a specified frequency band are obtained, and the autocorrelation coefficient is used to indicate the correlation between the signal and the time-delay signal of the signal. The multidimensional eigenvector of the multivariate intrinsic mode function is determined according to the energy entropy and autocorrelation coefficient of the multivariate intrinsic mode function. The multidimensional eigenvectors of the multiple multivariate intrinsic mode functions are clustered to determine at least one multivariate intrinsic mode function corresponding to the EEG signal. Alternatively, the multidimensional eigenvectors of the multiple multivariate intrinsic mode functions are clustered to determine at least one multivariate intrinsic mode function corresponding to a noise signal and at least one multivariate intrinsic mode function corresponding to an EEG signal, and the multivariate intrinsic mode function corresponding to the noise signal is deleted.

[0014] Based on this design, the signal type corresponding to the multivariate intrinsic mode function is determined according to the energy entropy and autocorrelation coefficient in the specified frequency band, which can improve the accuracy of identifying the signal type and thus improve the accuracy of EEG signal denoising.

[0015] In one possible design, the designated frequency band includes at least one of a first frequency band δ less than 4 Hz, a second frequency band θ greater than or equal to 4 Hz and less than 8 Hz, a third frequency band α greater than or equal to 8 Hz and less than 13 Hz, a fourth frequency band β greater than or equal to 13 Hz and less than 30 Hz, and a fifth frequency band γ greater than or equal to 30 Hz.

[0016] In a second aspect, the embodiment of the present application also provides an adaptive denoising device for EEG signals in a ubiquitous environment, including a communication module and a processing module.

[0017] The communication module is used to obtain the signal to be processed, which includes a noise signal and an electroencephalogram signal.

[0018] The processing module is used to perform blind source signal separation processing on the signal to be processed to obtain multiple source component components.

[0019] The processing module is also used to determine at least one first source component component from the multiple source component components based on the information entropy of the multiple source component components and a first threshold value. The information entropy is used to indicate the complexity of the signal. The signal corresponding to the first source component component includes a noise signal and an electroencephalogram signal.

[0020] The processing module is also used to process the first source component based on a multivariate variational mode decomposition method to obtain a plurality of multivariate intrinsic mode functions.

[0021] The processing module is further used to perform clustering processing on the multiple multivariate intrinsic mode functions to determine at least one multivariate intrinsic mode function corresponding to the EEG signal.

[0022] The processing module is also used to perform inverse reconstruction processing on the multivariate intrinsic mode function corresponding to the EEG signal to obtain the EEG signal.

[0023] In a possible design, the communication module is specifically used to: obtain a waveform factor and a kurtosis factor of a collected signal, and determine that the signal is the signal to be processed according to the waveform factor and the kurtosis factor of the signal and a preset threshold.

[0024] In one possible design, the information entropy of the source component component includes the fuzzy entropy of the source component component.

[0025] In one possible design, the fuzzy entropy of the source component satisfies:

[0026]

[0027] Among them, H(m) represents the fuzzy entropy of the source component, N is the number of subsequences of the source component, U(i,j) represents the fuzzy similarity between the i-th subsequence and the j-th subsequence, and ∑ is the accumulation symbol.

[0028] In one possible design, the processing module is specifically used to: obtain the energy entropy and autocorrelation coefficient of the multivariate intrinsic mode function in a specified frequency band, and the autocorrelation coefficient is used to indicate the correlation between the signal and the time-delay signal of the signal. Determine the multidimensional eigenvector of the multivariate intrinsic mode function based on the energy entropy and autocorrelation coefficient of the multivariate intrinsic mode function. Cluster the multidimensional eigenvectors of the multiple multivariate intrinsic mode functions to determine at least one multivariate intrinsic mode function corresponding to the EEG signal. Alternatively, cluster the multidimensional eigenvectors of the multiple multivariate intrinsic mode functions to determine at least one multivariate intrinsic mode function corresponding to a noise signal and at least one multivariate intrinsic mode function corresponding to an EEG signal, and delete the multivariate intrinsic mode function corresponding to the noise signal.

[0029] In one possible design, the designated frequency band includes at least one of a first frequency band δ less than 4 Hz, a second frequency band θ greater than or equal to 4 Hz and less than 8 Hz, a third frequency band α greater than or equal to 8 Hz and less than 13 Hz, a fourth frequency band β greater than or equal to 13 Hz and less than 30 Hz, and a fifth frequency band γ greater than or equal to 30 Hz.

[0030] In a third aspect, an embodiment of the present application further provides an electronic device, comprising a processor, wherein the processor is configured to implement the method of the first aspect and any one of its designs when executing a computer program stored in a memory.

[0031] In a fourth aspect, an embodiment of the present application further provides a computer-readable storage medium storing a computer program, which, when executed by a processor, implements the method of the first aspect and any one of its designs.

[0032] The technical effects brought about by the second to fourth aspects and any one of their designs can refer to the technical effects brought about by the corresponding design in the first aspect, and will not be repeated here. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0034] Figure 1 A flowchart of an adaptive denoising method for EEG signals in a ubiquitous environment provided in an embodiment of the present application;

[0035] Figure 2 A schematic diagram of a flow chart of another method for adaptive denoising of EEG signals in a ubiquitous environment provided in an embodiment of the present application;

[0036] Figure 3 A schematic diagram of the structure of an adaptive denoising device for EEG signals in a ubiquitous environment provided in an embodiment of the present application;

[0037] Figure 4 A schematic diagram of the structure of another adaptive denoising device for EEG signals in a ubiquitous environment provided by an embodiment of the present application;

[0038] Figure 5 A schematic diagram of the structure of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0039] In order to make the purpose, technical solutions and advantages of the present application clearer, the present application will be described in detail in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present application.

[0040] Next, the denoising method of EEG signals is introduced in combination with the existing technology.

[0041] At present, the denoising scheme of EEG signals in the prior art usually performs filtering, regression, blind source signal separation and frequency decomposition on the collected EEG signals to remove the noise signals in the collected EEG signals.

[0042] Among them, filtering is usually only applicable to removing noise signals in a specific frequency band, and cannot remove noise signals that overlap with the EEG signal spectrum. The scheme of regression processing of EEG signals will cause the problem of "bidirectional contamination" of EEG signals, and other reference signals need to be collected additionally when processing EEG signals. Blind source signal separation algorithms such as independent component analysis (ICA) need to rely on manual identification of noise components, resulting in low efficiency and poor accuracy of EEG signal denoising. Frequency decomposition types will cause modal aliasing problems when processing noise signals with overlapping frequency bands, resulting in low accuracy of EEG signal denoising.

[0043] EEG signals collected in ubiquitous environments usually include noise signals with high amplitude, wide bandwidth and complex components. However, existing EEG signal denoising schemes cannot accurately remove noise signals from EEG signals.

[0044] In order to solve the above-mentioned defects, the present application provides an adaptive denoising method and device for EEG signals in a ubiquitous environment, so as to improve the accuracy of denoising processing on EEG signals.

[0045] In the present application, the method adopted includes: the first device performs blind source signal separation processing on the signal to be processed to obtain the source component of the signal. The first device determines the first source component corresponding to the aliased signal of the noise signal and the electroencephalogram signal based on the information entropy of the source component and the first threshold. The first device processes the first source component based on the multivariate variational mode decomposition method to obtain the multivariate intrinsic mode function corresponding to the first source component. The first device performs clustering processing on the multivariate intrinsic mode function to determine the intrinsic mode function corresponding to the electroencephalogram signal. The first device performs inverse reconstruction processing on the intrinsic mode function corresponding to the electroencephalogram signal, thereby restoring the original electroencephalogram signal.

[0046] By adopting this method, the first device can efficiently and accurately determine the noise signal in the EEG signal collected in the universal environment according to the information entropy of the source component and the first threshold, so as to efficiently and accurately remove the noise signal in the EEG signal collected in the universal environment, and improve the accuracy of denoising the EEG signal collected in the universal environment. In addition, the first device performs clustering processing on the multivariate intrinsic mode function corresponding to the EEG signal, adaptively identifies the noise signal, and efficiently removes the noise signal of the EEG signal.

[0047] In addition, the first device may be included in a computer system for executing the method described in the present application, or may be a processing device in a computer system for executing the method described in the present application, such as a processor or a processing module, etc., which is not specifically limited in the present application.

[0048] Figure 1 A flowchart of a method for adaptively denoising EEG signals in a ubiquitous environment provided by an embodiment of the present invention. Taking the first device as an example, the process may include the following steps:

[0049] S101, a first device obtains a signal to be processed, wherein the signal to be processed includes a noise signal and an electroencephalogram signal, and the signal to be processed may be a mixed signal of the electroencephalogram signal and the noise signal.

[0050] In one or more embodiments, the first device may obtain the waveform factor and kurtosis factor of the collected signal. The first device may determine that the signal is a signal to be processed based on the waveform factor and kurtosis factor of the signal and a preset threshold. For example, Figure 2 A flowchart of an adaptive denoising method for EEG signals in a ubiquitous environment provided in an embodiment of the present application is shown in FIG. Figure 2 As shown, the first device can pass through step a a signal frame containing noise, that is, a signal to be processed.

[0051] As an example, the first device may include a module or device having a signal acquisition function, and the first device may acquire signals through its own module or device. The first device may perform preprocessing such as filtering on the acquired signals to improve the signal-to-noise ratio of the signals. The first device may divide the preprocessed signals into a plurality of frame signals of fixed lengths according to a preset duration (e.g., 10-30 milliseconds).

[0052] The first device may apply a sliding window to the plurality of frame signals respectively according to a preset window function to ensure the continuity of the frame signals of two adjacent frames. The frame signal to which the sliding window is applied may be expressed as y(n), the frame signal may be expressed as s(n), and the window function may be expressed as ω(n). Then the frame signal to which the sliding window is applied, the frame signal and the window function satisfy:

[0053] y(n)=s(n)ω(n).

[0054] The first device can obtain the waveform factor and kurtosis factor of each frame signal. The specific methods of obtaining the waveform factor and kurtosis factor of the signal are all prior arts and are not specifically limited in this application.

[0055] The first device can determine the starting position and ending position of the noise signal in the frame signal according to the waveform factor and kurtosis factor of each frame signal and a preset threshold value, that is, the starting position and ending position of the signal to be processed in the frame signal. For example, the first device can determine the starting position and ending position of the noise signal in the frame signal according to the waveform factor and kurtosis factor of each frame signal based on the double threshold method. Among them, the two threshold values ​​of the double threshold method can be used to set the threshold value based on experience.

[0056] It is understandable that EEG signals have the characteristic of short-term stability, that is, EEG signals in a relatively short period of time can be regarded as stable signals, while noise signals usually do not have the characteristic of short-term stability. When a noise signal is included in an EEG signal, the short-term stability of the EEG signal is poor or does not have short-term stability. In addition, the stability of the waveform factor of the signal is good but the sensitivity is poor, and the sensitivity of the kurtosis factor of the signal is good but the stability is poor. Therefore, the noise signal in the frame signal can be accurately determined based on the waveform factor and kurtosis factor of the signal.

[0057] In addition, the action of obtaining the signal to be processed from the collected signal can also be performed by other devices. For example, Figure 3 A schematic diagram of the structure of an adaptive denoising device for EEG signals in a ubiquitous environment provided in an embodiment of the present application is shown in FIG. Figure 3As shown, the noise detection module can be represented as other devices, or devices in other devices. The noise detection module can perform noise detection on the EEG signal, and determine the starting position and ending position of the noise signal based on the waveform factor and the kurtosis factor. At this time, the first device can be considered as a device independent of the signal acquisition device, or a device in an independent device. The first device and the signal acquisition device can be connected through a wired interface and / or a wireless interface. Correspondingly, the signal acquisition device can send a signal to be processed to the first device through a wired interface and / or a wireless interface. The first device can receive a signal to be processed from the signal acquisition device through a wired interface and / or a wireless interface.

[0058] S102: The first device performs blind source signal separation processing on the signal to be processed to obtain a plurality of source component components.

[0059] In one or more embodiments, the first device may use fast independent vector analysis (IVA) based on Laplace distribution to perform blind source signal separation processing on the signal to be processed to obtain a plurality of source component components corresponding to the signal to be processed.

[0060] It is understandable that the EEG signal is determined based on multiple field potential sources, and the EEG signal and the multiple field potential sources approximately satisfy the linear transformation relationship. The EEG signal can be expressed as X, and the multiple field potential sources can be expressed as S, then the EEG signal and the multiple field potential sources satisfy:

[0061] X = AS;

[0062] Among them, A can be expressed as a mixing matrix, and the mixing matrix can be used to indicate the linear relationship between the EEG signal and multiple field potential sources.

[0063] Since there is a linear transformation relationship between EEG signals and multiple field potential sources, the EEG signals can be processed according to the unmixing matrix to obtain independent source signals. Among them, the unmixing matrix corresponds to the mixing matrix. The unmixing matrix can be expressed as W, and the independent source signal can be expressed as The EEG signal can be expressed as X, then the independent source signal, EEG signal and unmixing matrix satisfy:

[0064]

[0065] As an example, the unmixing matrix can be obtained as follows:

[0066] The cost function of the fast IVA algorithm satisfies:

[0067]

[0068] in, is the estimated spectrum of the kth frequency point of the mth brain source signal, E(·) represents the mathematical expectation, and G(·) is a nonlinear function.

[0069] Furthermore, the nonlinear function satisfies:

[0070]

[0071] in, q(y m ) represents the signal distribution, i.e., the variable Laplace distribution.

[0072] The separation matrix is ​​obtained by:

[0073]

[0074] in, yes The mth column element of * represents conjugate, G'(·) and G”(·) represent the first-order derivative and second-order derivative of the nonlinear function G(·), respectively, and ite is the number of algorithm iterations.

[0075] For example, Figure 3 As shown, the blind source signal separation module can be represented as a first device, or an apparatus in the first device. The blind source signal separation module can be used to perform blind source signal separation on a signal including a noise signal (ie, a signal to be processed) based on fast independent vector analysis of Laplace distribution to obtain multiple source signal matrices.

[0076] Based on this embodiment, the convergence speed of signal processing can be improved, thereby improving the efficiency of obtaining multiple source component components corresponding to the signal to be processed.

[0077] S103, the first device determines at least one first source component from the multiple source component components according to the information entropy of the multiple source component components and the first threshold value. The information entropy is used to indicate the complexity of the signal, and the signal corresponding to the first source component component includes a noise signal and an electroencephalogram signal, or the signal corresponding to the first source component component is an aliased signal of the noise signal and the electroencephalogram signal. The first threshold value can be set in advance according to the needs of the business.

[0078] In one or more embodiments, the information entropy of the source component includes the fuzzy entropy of the source component. In addition, the information entropy of the source component may also include approximate entropy, sample entropy, and permutation entropy.

[0079] The following describes step S103 by taking fuzzy entropy as an example.

[0080] Exemplarily, the first device may obtain the fuzzy entropy of the source component according to multiple subsequences of the source component. The fuzzy entropy may be expressed as H(m), and the fuzzy entropy satisfies:

[0081]

[0082] Among them, H(m) represents the fuzzy entropy of the source component, N is the number of subsequences of the source component, U(i,j) represents the fuzzy similarity between the i-th subsequence and the j-th subsequence, and ∑ is the accumulation symbol.

[0083] As an example, the first device may receive fuzzy entropy of source component components from other devices. Figure 3 As shown, the signal decomposition module can be used to calculate the fuzzy entropy of each source component component in a plurality of source component components. The signal decomposition module can be a device in the first device, and in this case, the first device obtains the fuzzy entropy of the source component component through the signal decomposition module. Alternatively, the signal decomposition module can also be a device independent of the first device, or a device in an independent device. The signal decomposition module can be connected to the first device through a wired interface and / or a wireless interface. Accordingly, the first device can receive the fuzzy entropy from the signal decomposition module through the wired interface and / or the wireless interface.

[0084] The first device can determine whether the fuzzy entropy of the source component is greater than the first threshold value, and determine whether the signal corresponding to the source component component includes a noise signal. When the fuzzy entropy of the source component component is greater than the first threshold value, the signal corresponding to the source component component does not include a noise signal. In other words, the signal corresponding to the source component component only includes an EEG signal. When the fuzzy entropy of the source component component is less than the first threshold value, the signal corresponding to the source component component includes a noise signal. Alternatively, the signal corresponding to the source component component is an aliased signal of a noise signal and an EEG signal.

[0085] For example, Figure 2 As shown, the first device can determine whether the signal corresponding to the source component includes a noise signal through step b.

[0086] It is understandable that EEG signals have the characteristic of high complexity, while the complexity of noise signals is usually low. That is to say, when the signal to be processed includes a noise signal, the complexity corresponding to the signal is low. Signal entropy can be used to indicate the complexity of the signal. Therefore, based on step S103, the source component component corresponding to the signal including the noise signal can be accurately determined from multiple source component components, thereby improving the accuracy of EEG signal denoising.

[0087] S104: The first device processes the first source component based on a multivariate variational mode decomposition method to obtain a plurality of multivariate intrinsic mode functions.

[0088] As an example, the first device may process the first source component in the following manner to obtain a plurality of multivariate intrinsic mode functions corresponding to the first source component. Figure 2 As shown, the first device can process the source component based on the multivariate variational mode decomposition method through step c.

[0089] The first source component can be expressed as x c (t), then the process of solving the first source component based on the multivariate variational mode decomposition method is as follows:

[0090]

[0091] in, is the analytical modulation signal corresponding to mode k, {u k,c (t)} is a set of multi-modulated oscillating signals, {ω k} is a set of center frequencies.

[0092] The augmented Lagrangian function can be constructed by introducing the Lagrangian operator λ(t) to solve the first source component. The augmented Lagrangian function satisfies:

[0093]

[0094] Among them, α can be expressed as a penalty factor, which is used to reduce the interference of Gaussian noise.

[0095] In addition, the above unconstrained variational problem can be solved by the alternating method multiplier method to optimize the mode and center frequency. Among them, the optimized mode satisfies:

[0096]

[0097] The optimized center frequency satisfies:

[0098]

[0099] For example, Figure 3 As shown, the signal decomposition module can be represented as a first device, and the first device can process the source component by multivariate variational mode decomposition to separate the multivariate intrinsic mode functions of each order variable.

[0100] It can be understood that processing the first source component based on the multivariate variational mode decomposition method can ensure the frequency consistency and robustness of the signal decomposition.

[0101] S105: The first device performs clustering processing on the multiple multivariate intrinsic mode functions to determine a multivariate intrinsic mode function corresponding to at least one EEG signal.

[0102] In one or more embodiments, the first device may cluster multiple multivariate intrinsic modal functions according to the characteristics of the multiple multivariate intrinsic modal functions, and determine the multivariate intrinsic modal function corresponding to at least one EEG signal. That is, the first device may perform type identification on multiple multivariate intrinsic modal functions according to the characteristics of the multivariate intrinsic modal functions, and determine the multivariate intrinsic modal function corresponding to the EEG signal and / or the multivariate intrinsic modal function corresponding to the noise signal. The characteristics of the multivariate intrinsic modal function may include the spectral energy entropy and autocorrelation coefficient of the multivariate intrinsic modal function, and the autocorrelation coefficient may be used to indicate the correlation between the signal and the time-delay signal of the signal. The clustering may be hierarchical clustering (HC).

[0103] As an example, the first device may obtain the energy entropy of the multivariate intrinsic mode function in a specified frequency band and the autocorrelation coefficient of the multivariate intrinsic mode function. Exemplarily, the first device may obtain the energy entropy and the autocorrelation coefficient of the multivariate intrinsic mode function in the following manner:

[0104] The energy entropy of the multivariate intrinsic mode function can be expressed as H, then the energy entropy of the multivariate intrinsic mode function satisfies:

[0105]

[0106] P i =E i / E;

[0107] Among them, E i It can be expressed as the energy of a specified frequency band, and E can be expressed as the energy of the entire frequency band.

[0108] The multivariate intrinsic mode function can be expressed as x(t), and the autocorrelation coefficient of the multivariate intrinsic mode function can be expressed as AutoCC. Then the autocorrelation coefficient of the multivariate intrinsic mode function satisfies:

[0109]

[0110] Among them, Cov() is the covariance formula, Var() is the variance formula, and x(t-τ) can be expressed as a lagged signal of the multivariate intrinsic mode function delayed by τ.

[0111] It is understandable that energy entropy can be used to indicate the energy distribution of the signal in the time domain and the energy proportion of the signal component in the frequency domain entropy. Different signals have different energy entropies in a specified frequency band. In addition, due to the different characteristics of EEG signals and noise signals, the autocorrelation coefficient of EEG signals is smaller than the autocorrelation coefficient of noise signals. Therefore, the signal type corresponding to the multivariate intrinsic mode function can be determined based on the energy entropy and autocorrelation coefficient of the multivariate intrinsic mode function, that is, EEG signal or noise signal.

[0112] As an example, the first device may determine a multidimensional eigenvector of the multivariate intrinsic mode function according to the energy entropy and the autocorrelation coefficient of the multivariate intrinsic mode function.

[0113] Optionally, the designated frequency band includes at least one of a first frequency band δ being less than 4 Hz, a second frequency band θ being greater than or equal to 4 Hz and less than 8 Hz, a third frequency band α being greater than or equal to 8 Hz and less than 13 Hz, a fourth frequency band β being greater than or equal to 13 Hz and less than 30 Hz, and a fifth frequency band γ being greater than or equal to 30 Hz.

[0114] Exemplarily, take the case where the designated frequency bands include the first frequency band δ, the second frequency band θ, the third frequency band α, the fourth frequency band β and the fifth frequency band γ. The energy entropy of the multivariate intrinsic modal function in the first frequency band δ can be expressed as f1, the energy entropy of the multivariate intrinsic modal function in the second frequency band θ can be expressed as f2, the energy entropy of the multivariate intrinsic modal function in the third frequency band α can be expressed as f3, the energy entropy of the multivariate intrinsic modal function in the fourth frequency band β can be expressed as f4, the energy entropy of the multivariate intrinsic modal function in the fifth frequency band γ can be expressed as f5, and the autocorrelation coefficient of the multivariate intrinsic modal function can be expressed as f6. The first device can obtain a 6-dimensional eigenvector of the multivariate intrinsic modal function based on the energy entropy and autocorrelation coefficient of the multivariate intrinsic modal function. Among them, the 6-dimensional eigenvector can be expressed as F, then the 6-dimensional eigenvector satisfies:

[0115] F=[f1,f2,f3,f4,f5,f6] T ;

[0116] in,[] T Represented as transposed.

[0117] Furthermore, the first device may perform clustering processing on the multidimensional feature vectors of the plurality of multivariate intrinsic mode functions to determine the multivariate intrinsic mode function corresponding to at least one EEG signal.

[0118] Alternatively, the first device can perform clustering processing on the multidimensional feature vectors of multiple multivariate intrinsic mode functions, determine the multivariate intrinsic mode function corresponding to at least one noise signal and the multivariate intrinsic mode function corresponding to at least one EEG signal, and delete the multivariate intrinsic mode function corresponding to the noise signal.

[0119] That is to say, the first device can determine the multivariate intrinsic mode function whose signal type is an electroencephalogram signal based on the multidimensional characteristic vector of the multivariate intrinsic mode function. Alternatively, the first device can determine the signal type corresponding to multiple multivariate intrinsic mode functions, i.e., electroencephalogram signal or noise signal, based on the multidimensional characteristic vector of the multivariate intrinsic mode function. After determining the signal type corresponding to multiple multivariate intrinsic mode functions, the first device can delete the multivariate intrinsic mode function whose signal type is a noise signal and retain the multivariate intrinsic mode function whose signal type is an electroencephalogram signal.

[0120] Among them, the specific methods of clustering the multidimensional feature vectors of multiple multivariate intrinsic mode functions are all existing technologies and will not be elaborated in this application.

[0121] Still Figure 3 For example, the noise removal module can be represented as a first device, or a device in the first device. The noise removal module can be used to obtain the characteristics of the multivariate intrinsic mode function, and cluster the multivariate intrinsic mode function to determine the signal type corresponding to each multivariate intrinsic mode function. In addition, the noise removal module can also perform zeroing processing on the multivariate intrinsic mode function corresponding to the noise signal to remove the noise signal.

[0122] It can be understood that by using this method to determine the signal type corresponding to the multivariate intrinsic mode function, the signal type corresponding to the multivariate intrinsic mode function can be adaptively determined to improve the accuracy and efficiency of identifying the signal type, thereby improving the accuracy and efficiency of EEG signal denoising.

[0123] S106, the first device performs inverse reconstruction processing on the multivariate intrinsic mode function corresponding to the EEG signal to obtain the EEG signal.

[0124] In one or more embodiments, the first device can perform inverse operation processing on the multivariate intrinsic mode function corresponding to the EEG signal based on the multivariate variational mode decomposition method to obtain the denoised source component. The first device can also perform inverse reconstruction processing on the denoised source component to obtain the denoised EEG signal.

[0125] Still Figure 3 For example, the signal reconstruction module can be represented as the first device, or a device in the first device. The signal reconstruction module can be used to perform an inverse operation of multivariate variational mode decomposition on the multivariate intrinsic mode function corresponding to the EEG signal to obtain the denoised source component, and perform an inverse reconstruction process on the denoised source component to obtain the denoised EEG signal, thereby restoring the original EEG signal.

[0126] In addition, the first device may superimpose the denoised source component with the source component excluding the noise signal, and inversely reconstruct the superimposed source component to obtain the corresponding EEG signal. Figure 2As shown, the first device can obtain a pure EEG signal through step e.

[0127] Alternatively, the first device can perform inverse reconstruction processing on the denoised source component and the source component excluding the noise signal, respectively, and splice the obtained signals to obtain the corresponding EEG signal. The specific methods of performing inverse reconstruction processing on the source component in this application are all prior art and will not be elaborated here.

[0128] Based on the above content and the same concept, the present application provides an adaptive denoising device for EEG signals in a ubiquitous environment. Figure 4 As shown, the device includes a communication module 401 and a processing module 402 .

[0129] The communication module 401 is used to obtain a signal to be processed, where the signal to be processed includes a noise signal and an electroencephalogram signal.

[0130] The processing module 402 is used to perform blind source signal separation processing on the signal to be processed to obtain multiple source component components.

[0131] The processing module 402 is also used to determine at least one first source component component from the multiple source component components based on the information entropy of the multiple source component components and a first threshold value. The information entropy is used to indicate the complexity of the signal. The signal corresponding to the first source component component includes a noise signal and an EEG signal.

[0132] The processing module 402 is further configured to process the first source component based on a multivariate variational mode decomposition method to obtain a plurality of multivariate intrinsic mode functions.

[0133] The processing module 402 is further configured to perform clustering processing on the multiple multivariate intrinsic mode functions to determine at least one multivariate intrinsic mode function corresponding to the EEG signal.

[0134] The processing module 402 is further used to perform inverse reconstruction processing on the multivariate intrinsic mode function corresponding to the EEG signal to obtain the EEG signal.

[0135] In a possible design, the communication module 401 is specifically used to: obtain the waveform factor and kurtosis factor of the collected signal, and determine that the signal is the signal to be processed according to the waveform factor and kurtosis factor of the signal and a preset threshold.

[0136] In one possible design, the information entropy of the source component component includes the fuzzy entropy of the source component component.

[0137] In one possible design, the fuzzy entropy of the source component satisfies:

[0138]

[0139] Among them, H(m) represents the fuzzy entropy of the source component, N is the number of subsequences of the source component, U(i,j) represents the fuzzy similarity between the i-th subsequence and the j-th subsequence, and ∑ is the accumulation symbol.

[0140] In one possible design, the processing module 402 is specifically used to: obtain the energy entropy and autocorrelation coefficient of the multivariate intrinsic mode function in a specified frequency band, and the autocorrelation coefficient is used to indicate the correlation between the signal and the time-delay signal of the signal. Determine the multidimensional eigenvector of the multivariate intrinsic mode function according to the energy entropy and autocorrelation coefficient of the multivariate intrinsic mode function. Cluster the multidimensional eigenvectors of the multiple multivariate intrinsic mode functions to determine at least one multivariate intrinsic mode function corresponding to the EEG signal. Alternatively, cluster the multidimensional eigenvectors of the multiple multivariate intrinsic mode functions to determine at least one multivariate intrinsic mode function corresponding to a noise signal and at least one multivariate intrinsic mode function corresponding to an EEG signal, and delete the multivariate intrinsic mode function corresponding to the noise signal.

[0141] In one possible design, the designated frequency band includes at least one of a first frequency band δ less than 4 Hz, a second frequency band θ greater than or equal to 4 Hz and less than 8 Hz, a third frequency band α greater than or equal to 8 Hz and less than 13 Hz, a fourth frequency band β greater than or equal to 13 Hz and less than 30 Hz, and a fifth frequency band γ greater than or equal to 30 Hz.

[0142] Based on the same inventive concept, an embodiment of the present application provides an electronic device, which can realize the functions of the device discussed above. Figure 5 A schematic diagram of the structure of an electronic device provided in an embodiment of the present application is shown.

[0143] The electronic device in the embodiment of the present application may include a processor 501. The processor 501 is the control center of the device, and various interfaces and lines can be used to connect various parts of the device, by running or executing instructions stored in the memory 503 and calling data stored in the memory 503. Optionally, the processor 501 may include one or more processing units, and the processor 501 may integrate an application processor and a modem processor, wherein the application processor mainly processes operating systems and application programs, etc., and the modem processor mainly processes wireless communications. It is understandable that the above-mentioned modem processor may not be integrated into the processor 501. In some embodiments, the processor 501 and the memory 503 may be implemented on the same chip, and in some embodiments, they may also be implemented separately on independent chips.

[0144] The processor 501 may be a general-purpose processor, such as a central processing unit (CPU), a digital signal processor, an application-specific integrated circuit, a field programmable gate array or other programmable logic device, a discrete gate or transistor logic device, or a discrete hardware component, and may implement or execute the methods, steps, and logic block diagrams disclosed in the embodiments of the present application. A general-purpose processor may be a microprocessor or any conventional processor, etc. The method steps disclosed in the embodiments of the present application may be directly executed by a hardware processor, or may be executed by a combination of hardware and software modules in the processor.

[0145] In the embodiment of the present application, the memory 503 stores instructions that can be executed by at least one processor 501. The at least one processor 501 can be used to execute the method steps disclosed in the embodiment of the present application by executing the instructions stored in the memory 503.

[0146] The memory 503 is a non-volatile computer-readable storage medium that can be used to store non-volatile software programs, non-volatile computer executable programs and modules. The memory 503 may include at least one type of storage medium, such as a flash memory, a hard disk, a multimedia card, a card-type memory, a random access memory (RAM), a static random access memory (SRAM), a programmable read-only memory (PROM), a read-only memory (ROM), an electrically erasable programmable read-only memory (EEPROM), a magnetic memory, a disk, an optical disk, etc. The memory 503 is any other medium that can be used to carry or store a desired program code in the form of an instruction or data structure and can be accessed by a computer, but is not limited thereto. The memory 503 in the embodiment of the present application can also be a circuit or any other device that can realize a storage function, for storing program instructions and / or data.

[0147] In the embodiment of the present application, the device may further include a communication interface 502 , through which the electronic device may transmit data.

[0148] Optional, can be Figure 5 The processor 501 (or the processor 501 and the communication interface 502) implements Figure 4The processing module 402 and / or the communication module 401 shown, that is, the actions of the processing module 402 and / or the communication module 401 can be executed by the processor 501 (or the processor 501 and the communication interface 502).

[0149] Based on the same inventive concept, the embodiment of the present application also provides a computer-readable storage medium, which may store instructions. When the instructions are executed on a computer, the computer executes the operation steps provided in the above method embodiment. The computer-readable storage medium may be Figure 5 The memory 503 is shown.

[0150] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment in combination with software and hardware. Moreover, the present application may adopt the form of a computer program product implemented in one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that contain computer-usable program code.

[0151] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0152] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0153] These computer program instructions can also be loaded onto a computer or other programmable data processing device, so that a series of operation steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby executing the program on the computer or other programmable device.

Claims

1. An adaptive denoising method for EEG signals in a ubiquitous environment, characterized in that: The method comprises: Acquire a signal to be processed, where the signal to be processed is a mixed signal of an electroencephalogram signal and a noise signal; Performing blind source signal separation processing on the signal to be processed by fast independent vector analysis based on Laplace distribution to obtain multiple source component components; Determine at least one first source component from the multiple source component components according to the information entropy of the multiple source component components and a first threshold, wherein the information entropy is used to indicate the complexity of the signal, and the signal corresponding to the first source component includes a noise signal and an electroencephalogram signal; Processing the first source component based on a multivariate variational mode decomposition method to obtain a plurality of multivariate intrinsic mode functions; Performing clustering processing on the multiple multivariate intrinsic mode functions to determine a multivariate intrinsic mode function corresponding to at least one of the EEG signals; Performing inverse reconstruction processing on the multivariate intrinsic mode function corresponding to the EEG signal to obtain the EEG signal; The clustering process of the plurality of multivariate intrinsic mode functions to determine the multivariate intrinsic mode function corresponding to at least one of the electroencephalogram signals includes: Determine a multidimensional feature vector of the multivariate intrinsic mode function in a specified frequency band, wherein the multidimensional feature vector is determined based on a spectrum energy entropy and an autocorrelation coefficient corresponding to the specified frequency band; Performing clustering processing on the plurality of multivariate intrinsic mode functions according to the multidimensional feature vector, and deleting the multivariate intrinsic mode function corresponding to the noise signal; The information entropy of the source component comprises the fuzzy entropy of the source component, and the fuzzy entropy of the source component satisfies: Wherein, H(m) represents the fuzzy entropy of the source component, N is the number of subsequences of the source component, U(i,j) represents the fuzzy similarity between the i-th subsequence and the j-th subsequence, and ∑ is the accumulation symbol; The autocorrelation coefficient of the multivariate intrinsic mode function satisfies: Wherein, x(t) is the multivariate intrinsic mode function, AutoCC is the autocorrelation coefficient, Cov() is the covariance formula, Var() is the variance formula, and x(t-τ) is the lag signal of the multivariate intrinsic mode function delayed by τ.

2. The method according to claim 1, characterized in that The obtaining of the signal to be processed comprises: Obtaining the waveform factor and kurtosis factor of the collected signal; The signal is determined to be the signal to be processed according to the waveform factor and the kurtosis factor of the signal and a preset threshold.

3. The method according to claim 1, characterized in that The designated frequency band includes at least one of the following: The first frequency band δ is less than 4 Hz; The second frequency band θ is greater than or equal to 4 Hz and less than 8 Hz; The third frequency band α is greater than or equal to 8 Hz and less than 13 Hz; The fourth frequency band β is greater than or equal to 13 Hz and less than 30 Hz; The fifth frequency band γ is greater than or equal to 30 Hz.

4. An adaptive denoising device for EEG signals in a ubiquitous environment, characterized in that: The device comprises: A communication module, used for acquiring a signal to be processed, wherein the signal to be processed is a mixed signal of an electroencephalogram signal and a noise signal; A processing module, used for performing blind source signal separation processing on the signal to be processed by fast independent vector analysis based on Laplace distribution to obtain multiple source component components; The processing module is further used to determine at least one first source component component from the multiple source component components according to the information entropy of the multiple source component components and a first threshold, wherein the information entropy is used to indicate the complexity of the signal, and the signal corresponding to the first source component includes a noise signal and an electroencephalogram signal; The processing module is further used to process the first source component based on a multivariate variational mode decomposition method to obtain a plurality of multivariate intrinsic mode functions; The processing module is further used to perform clustering processing on the multiple multivariate intrinsic mode functions to determine a multivariate intrinsic mode function corresponding to at least one of the EEG signals; The processing module is further used to perform inverse reconstruction processing on the multivariate intrinsic mode function corresponding to the EEG signal, and obtain the multidimensional feature vector of the multivariate intrinsic mode function in the specified frequency band determined by the EEG signal, wherein the multidimensional feature vector is determined according to the spectrum energy entropy and autocorrelation coefficient corresponding to the specified frequency band; Performing clustering processing on the plurality of multivariate intrinsic mode functions according to the multidimensional feature vector, and deleting the multivariate intrinsic mode function corresponding to the noise signal; The information entropy of the source component comprises the fuzzy entropy of the source component, and the fuzzy entropy of the source component satisfies: Wherein, H(m) represents the fuzzy entropy of the source component, N is the number of subsequences of the source component, U(i,j) represents the fuzzy similarity between the i-th subsequence and the j-th subsequence, and ∑ is the accumulation symbol; The autocorrelation coefficient of the multivariate intrinsic mode function satisfies: Wherein, x(t) is the multivariate intrinsic mode function, AutoCC is the autocorrelation coefficient, Cov() is the covariance formula, Var() is the variance formula, and x(t-τ) is the lag signal of the multivariate intrinsic mode function delayed by τ.

5. The device according to claim 4, characterized in that The communication module is specifically used for: Obtaining the waveform factor and kurtosis factor of the collected signal; The signal is determined to be the signal to be processed according to the waveform factor and the kurtosis factor of the signal and a preset threshold.

6. The device according to claim 4, characterized in that The designated frequency band includes at least one of the following: The first frequency band δ is less than 4 Hz; The second frequency band θ is greater than or equal to 4 Hz and less than 8 Hz; The third frequency band α is greater than or equal to 8 Hz and less than 13 Hz; The fourth frequency band β is greater than or equal to 13 Hz and less than 30 Hz; The fifth frequency band γ is greater than or equal to 30 Hz.

7. An electronic device, characterized in that: The electronic device comprises a processor, and the processor is used to implement the steps of any one of the methods as claimed in claims 1 to 3 when executing a computer program stored in a memory.

8. A computer-readable storage medium, characterized in that: The device stores a computer program, which implements the steps of the method according to any one of claims 1 to 3 when executed by a processor.

Images