Electroencephalogram signal amplification method, system and terminal based on Gaussian mixture distribution model

Abstract

The invention belongs to the technical field of data amplification, and relates to an electroencephalogram signal amplification method and system based on a Gaussian mixture distribution model and a terminal. The amplification method comprises the following steps: collecting and obtaining an original electroencephalogram signal comprising a lead signal; obtaining a plurality of extreme points of each lead signal, and segmenting the lead signal; constructing a phase deviation Gaussian mixture model of each extreme point for each lead signal, and randomly extracting a phase difference from the phase deviation Gaussian mixture model corresponding to the extreme point; after the phase difference is applied to the corresponding extreme point, a sampling point is adjusted through increasing / undersampling to achieve peak offset, an initial amplification signal corresponding to each lead signal is obtained, smooth filtering is carried out on the initial amplification signal, and a filtered amplification signal is obtained; constructing an amplitude scaling Gaussian mixture model for the original electroencephalogram signals, and sampling amplitude change parameters from the amplitude scaling Gaussian mixture model; and for each lead signal, applying an amplitude variation parameter corresponding to the filtered amplification signal to the filtered amplification signal so as to generate amplitude variation, and finally obtaining the amplification signal.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of data augmentation, and particularly relates to an electroencephalogram signal augmentation method, system and terminal based on a Gaussian mixture distribution model. BACKGROUND

[0002] A brain-computer interface (BCI) bypasses the peripheral nervous system and muscles to directly transmit brain activity to external devices. Compared with the brain-computer interface system implanted with electrodes, the non-invasive brain-computer interface based on electroencephalography (EEG) has the advantages of safety, economy and easy operation, and therefore has wide application potential in applications such as typing, rehabilitation and robot control. However, the low signal-to-noise ratio and spatial resolution of EEG seriously limit the application of brain-computer interface.

[0003] In order to overcome the interference of background noise on task-evoked potentials, the traditional non-invasive visual brain-computer interface (v-BCI) usually uses methods such as increasing the size of the stimulation block or the intensity of the stimulation to induce a large neuron group response and produce obvious electroencephalogram features. However, for users, they are only interested in the task performed by the brain-computer interface, and are not interested in the stimulation of evoked potentials. In addition, long-term direct vision of strong visual stimulation will make the subjects feel visual fatigue, nervousness and even headache, and large-size stimulation will occupy a large amount of visual resources. However, reducing the stimulation area and lowering the stimulation intensity will make the electroencephalogram features not obvious, or even submerged in the background noise, which will greatly reduce the performance of the brain-computer interface system and bring challenges to the recognition of user's intention.

[0004] According to the principle of retinal mapping, the spatial pattern of visual evoked potentials (VEPs) is related to the position of visual stimulation presented in the user's visual field, which is manifested as that the lateral stimulation can induce P1-N1 waveform with larger difference in the contralateral hemisphere than in the ipsilateral hemisphere, i.e. asymmetric VEPs (aVEPs). According to the contralateral dominant property, the brain-computer interface paradigm designed based on aVEPs can eliminate the visual occupation of the user by the flickering stimulation by making the user not look directly at the stimulation block, while improving the user's interactive comfort. At present, there are related paradigm and algorithm researches on aVEPs, Xu et al. designed a character speller using aVEPs and realized the decoding of 0.5 μV level electroencephalogram signals, Xiao et al. designed an algorithm that can better filter out background noise and decode aVEPs by using the bilateral symmetry of brain activity, and Zhou et al. improved the DCPM and designed a brain-computer interface paradigm for tracking the fixation point based on aVEPs. However, as a kind of event-related potentials (ERPs), although aVEPs have the advantage of visual friendliness, they also have various typical shortcomings of ERPs, such as weak signal, large variation, and difficulty in stable extraction. For this situation, the traditional method is usually to collect multiple trials and directly classify using time domain waveform. In the case of sufficient training samples, satisfactory classification results can be obtained. However, collecting sufficient electroencephalogram data requires the subject to look at the screen for a long time, which is not only time-consuming and laborious, but also the long-time acquisition usually makes the subject mentally tired, which leads to weaker or changed signal characteristics, affecting the accuracy of the classification model.

[0005] A promising method is to augment the sample data with artificially generated signals, i.e. data augmentation. At present, there are related researchers who have applied data augmentation to the field of electroencephalogram, which can be roughly divided into deep learning methods and geometric methods. Deep learning methods themselves require a certain number of samples to train the model, such as generative adversarial networks and variational autoencoders, etc. However, these methods are generally used to learn electroencephalogram signals with very fixed features, and for the classification task of aVEPs with poor lock-in phase, they may even have the opposite effect. Geometric methods usually use random cropping, translation, etc. to augment the signal, but this method may damage some features of the original data. SUMMARY

[0006] The application aims to provide a brain electrical signal amplification method, system and terminal based on a Gaussian mixture distribution model, which collects non-locked asymmetric evoked potential brain electrical signals, and uses a Chebyshev filter of 0.5-20 Hz to denoise the brain electrical data; then the positions of the extreme points of each lead data are found, and the extreme points are segmented into small data segments according to the extreme points; then the left or right phase difference of each extreme point is fitted using Gaussian distribution, and the two Gaussian distributions are weighted and mixed to obtain the phase offset Gaussian mixture model of the extreme point; then the phase difference of each extreme point is extracted from the phase offset Gaussian mixture model of each extreme point, and the offset and smoothing of each wave peak / trough are performed using the up-sampling / down-sampling and local weighted smoothing algorithm; finally, the same method is used to weight and mix to obtain the amplitude scaling Gaussian mixture model, and the amplitude scaling of each wave peak / trough is performed based on the model. A new idea is provided in the field of non-locked data amplification.

[0007] To achieve the above-mentioned purpose, the application adopts the following technical solutions: In a first aspect, the application provides a brain electrical signal amplification method based on a Gaussian mixture distribution model, comprising the following steps: S10. Collect and obtain original brain electrical signals, wherein the original brain electrical signals include lead signals equal in number to the number of leads; S20. Obtain a plurality of extreme points of each lead signal, and segment the lead signal based on the extreme points; S30. Construct a phase offset Gaussian mixture model of each extreme point for each lead signal, randomly extract a phase difference from the phase offset Gaussian mixture model corresponding to each extreme point of each lead signal; after applying the phase difference to the corresponding extreme point, adjust the sampling points using up-sampling / down-sampling to realize peak offset, obtain an initial amplification signal corresponding to each lead signal, and perform smoothing filtering on the initial amplification signal to obtain a filtered amplification signal; S40. Construct an amplitude scaling Gaussian mixture model for the original brain electrical signal, and sample amplitude variation parameters therefrom; S50. For each lead signal, apply the filtered amplification signal to the corresponding amplitude variation parameter to produce amplitude variation, and finally obtain an amplification signal.

[0008] As a possible implementation manner, the phase offset Gaussian mixture model of each extreme point is constructed in the following manner: The mean and variance of the Gaussian distribution of the left segment and the right segment of the extreme point are calculated respectively; The corresponding Gaussian distribution is fitted based on the mean and variance of each group of Gaussian distributions; The Gaussian distributions are weighted and summed to obtain the Gaussian mixture model of the extreme point; Rewrite the weight in the Gaussian mixture model as the ratio of the difference between adjacent extreme points and the difference between interval extreme points, and obtain the phase shift Gaussian mixture model.

[0009] As a possible implementation, the weights corresponding to the two Gaussian distributions are respectively denoted as 、 , and are further rewritten as: wherein, is the kth extreme point of the jth lead signal; is the kth extreme point of the jth lead signal; is the kth extreme point of the jth lead signal.

[0010] As a possible implementation, the lead signal on the left side of the extreme point is oversampled, and the lead signal on the right side of the extreme point is undersampled; or, the lead signal on the left side of the extreme point is undersampled, and the lead signal on the right side of the extreme point is oversampled.

[0011] As a possible implementation, the amplitude scaling Gaussian mixture model of the original electroencephalogram signal is constructed by the following method: Set the maximum scaling amplitude and the minimum scaling amplitude; Calculate the expectation and variance of the Gaussian distribution corresponding to the maximum scaling amplitude and the minimum scaling amplitude respectively, fit the amplitude scaling Gaussian mixture model according to the above two sets of expectation and variance.

[0012] As a possible implementation, the initial amplification signal is smoothed by using the local weighted linear regression algorithm to obtain the filtered amplification signal.

[0013] In a second aspect, the present application provides an electroencephalogram signal amplification system based on Gaussian mixture distribution model, comprising: a signal acquisition unit for collecting and obtaining an original electroencephalogram signal, wherein the original electroencephalogram signal includes lead signals equal to the number of leads; a segmentation unit for obtaining a plurality of extreme points of each lead signal and segmenting the lead signal based on the extreme points; ​​​​​​The phase amplification unit constructs a phase-shift Gaussian mixture model for each extreme point of each lead signal. It randomly extracts the phase difference from the phase-shift Gaussian mixture model corresponding to each extreme point of each lead signal. After applying the phase difference to the corresponding extreme point, it adjusts the sampling point by upsampling / undersampling to achieve peak shift, thereby obtaining the initial amplified signal corresponding to each lead signal. The initial amplified signal is then smoothed and filtered to obtain the filtered amplified signal. The amplitude amplification unit constructs an amplitude-scaled Gaussian mixture model for the original EEG signal and samples amplitude variation parameters from it. The phase amplitude amplification unit applies a corresponding amplitude change parameter to the filtered amplified signal for each lead signal to generate an amplitude change, ultimately obtaining the amplified signal.

[0014] As one possible implementation, the first The first lead signal The Gaussian mixture model of phase shifts at each extreme point is denoted as . , ;in, For the first The first lead signal There are several extreme points; For the first The first lead signal There are several extreme points; For the first The first lead signal There are several extreme points; The distribution is a Gaussian distribution on the left side of the extreme point; It is a Gaussian distribution for the segment to the right of the extreme point.

[0015] As one possible implementation, the magnitude-scaled Gaussian mixture model is denoted as... , ;in, This represents the maximum scaling range; Minimum scaling factor; The Gaussian distribution corresponding to the minimum scaling factor; This represents the Gaussian distribution corresponding to the maximum scaling magnitude.

[0016] Thirdly, the present invention provides a terminal including a processor and a communication interface coupled to the processor, the processor being used to run computer programs or instructions to implement the EEG signal amplification method based on the Gaussian mixture distribution model provided in the first aspect.

[0017] Compared with the prior art, the beneficial effects of the present invention are as follows: 1.The electroencephalogram signal augmentation method based on a Gaussian mixture distribution model, which is a new method for generating artificial signals based on original signals, uses a Gauss Mixed Sampling (GMS) algorithm to expand electroencephalogram data samples and improve single-test aVEPs classification accuracy.

[0018] 2.The electroencephalogram signal augmentation method based on a Gaussian mixture distribution model essentially belongs to a geometric method and does not require model training.

[0019] 3.The electroencephalogram signal augmentation method based on a Gaussian mixture distribution model can generate a large amount of data from a small sample, and the GMS algorithm involves multiple sampling processes from a Gaussian mixture model, so theoretically, any number of artificial signals can be generated without repetition; the generated signals only change the characteristics between single tests and do not change the overall characteristics. BRIEF DESCRIPTION OF DRAWINGS

[0020] The accompanying drawings, which are included to provide a further understanding of the application, constitute a part of this application and illustrate embodiments of the application and its description, which do not constitute an improper limitation on the application. In the drawings: Figure 1 The electroencephalogram signal augmentation method based on a Gaussian mixture distribution model in the embodiments of the application is a flowchart; Figure 2 The original signal (red) and the signal generated by the Gaussian mixture model (blue); Figure 3 The flowchart of the asymmetric evoked potential data augmentation algorithm based on a Gaussian mixture distribution model; Figure 4 Part of the resampling of the Gauss Mixed Sampling algorithm and the weight instances corresponding to each sampling point; Figure 5 The evaluation index of the overall dispersion degree of each type of data before and after data augmentation and data visualization; Figure 6 The four-class accuracy comparison chart before and after data augmentation. DETAILED DESCRIPTION

[0021] In order to clearly describe the technical solutions of the embodiments of the application, in the embodiments of the application, the same items or similar items with basically the same functions and effects are distinguished by using "first", "second", etc. For example, the first threshold and the second threshold are only used to distinguish different thresholds, and do not limit the order. Those skilled in the art can understand that "first", "second", etc. do not limit the number and execution order, and "first", "second", etc. also do not necessarily mean different.

[0022] It should be noted that in the present application, the words "exemplary" or "for example" are used to mean serving as an example, instance, or illustration. Any embodiment or design described in this application as "exemplary" or "for example" should not be construed as being preferred or advantageous over other embodiments or designs. Rather, the embodied words are used to present concepts in a concrete manner.

[0023] In the present application, "at least one" means one or more, and "multiple" means two or more. The association relationship of "and / or" describes the association relationship of the associated objects, which means that there can be three relationships, for example, A and / or B, which can represent the cases of A alone, A and B together, B alone, where A and B can be singular or plural. The character " / " generally represents an "or" relationship between the associated objects before and after it. The following at least one or similar expressions mean any combination of these items, including any combination of single or multiple items. For example, at least one of a, b or c can represent: a, b, c, the combination of a and b, the combination of a and c, the combination of b and c, or the combination of a, b and c, where a, b and c can be single or multiple.

[0024] The embodiment of the present application aims to provide a brain electrical signal amplification method, system and terminal based on a Gaussian mixture distribution model, apply a Gauss Mixed Sampling (GMS) algorithm to expand the brain electrical data sample and improve the single trial aVEPs classification accuracy. The specific implementation is as follows: In a first aspect, the present application provides a brain electrical signal amplification method based on a Gaussian mixture distribution model, as shown in Figure 1 , comprising the following steps: S10. Collect and obtain the original brain electrical signal, which includes lead signals equal to the number of leads; For example, the brain electrical signal of the first trial is collected, denoted as , wherein is the number of leads, is the one-dimensional discrete data sampling point of the first lead, wherein . The brain electrical data is denoised using a Chebyshev filter of 0.5-20Hz; as shown in Figure 2 , the red waveform in the figure is an example waveform of the EEG of part of the leads in the present application.

[0025] S20. Obtain a plurality of extreme points of each lead signal, and segment the lead signal based on the extreme points; For example, the signal of the first lead is segmented into a plurality of segments, and each segment is denoted as The segmentation method of the present application is as follows: wherein represents the position of the time point of the th peak value, wherein , and respectively represent the position of the first point and the last point of the one-dimensional data . represents all sampling points from to .

[0026] S30. Construct a phase shift Gaussian mixture model of each extreme point for each lead signal, randomly extract a phase difference from the phase shift Gaussian mixture model corresponding to each extreme point of each lead signal; after applying the phase difference to the corresponding extreme point, adjust the sampling points by upsampling / downsampling to realize peak shift, obtain an initial amplified signal corresponding to each lead signal, perform smoothing filtering on the initial amplified signal to obtain a filtered amplified signal; see Figure 3 .

[0027] As a possible implementation manner, the phase shift Gaussian mixture model of each extreme point is constructed in the following manner: Calculate the mean and variance of the Gaussian distribution of the left segment and the right segment of the extreme point respectively; Exemplarily, according to the position of the extreme point , the mean and variance of the two Gaussian distributions of the left segment and the right segment are calculated as follows: wherein, and are the mean and variance calculated according to and , and are the mean and variance calculated according to and .

[0028] Fit the corresponding Gaussian distribution based on the mean and variance of each group of Gaussian distribution; Exemplarily, according to the two groups of mean and variance, the corresponding Gaussian distributions and can be fitted.

[0029] Sum the Gaussian distributions by weight to obtain the Gaussian mixture model of the extreme point; As a possible implementation manner, the weights corresponding to the two Gaussian distributions are respectively denoted as , , and Further rewritten as follows: in, For the first The first lead signal There are several extreme points; For the first The first lead signal There are several extreme points; For the first The first lead signal There are several extreme points.

[0030] For example, see Figure 3 (b) Weighted combination of the two Gaussian distributions to form the extreme points. Phase-shifted Gaussian mixture model: in, and These are the weights for the corresponding Gaussian distributions. The range of values ​​for each Gaussian distribution is as follows: .

[0031] The weights in the Gaussian mixture model are rewritten as the ratio of the difference between adjacent extreme points to the difference between phase-separated extreme points, thus obtaining the phase-shifted Gaussian mixture model. For example, to avoid generating a large number of distorted artificial signals, the algorithm needs to sample the phase difference in the opposite direction of a smaller offset interval. The two weight parameters in the phase-shifted Gaussian mixture model are determined by the number of sampling points in the two data segments. The phase-shifted Gaussian mixture model is obtained by weighting and fusing the two Gaussian distributions as follows. The expression is as follows: ; The setting of this weight is described in the Gaussian mixture sampling algorithm. Figure 3 (b) The area of ​​the probability fusion of the blue Gaussian distribution in the figure is smaller than that of the green Gaussian distribution.

[0032] The phase difference is randomly extracted from the Gaussian mixture model corresponding to each extreme point of each lead signal; For example, from No. Phase shift Gaussian mixture model of each peak Randomly select possible phase differences .

[0033] After the phase difference is applied to the corresponding extreme value point, the sampling points are adjusted by upsampling / downsampling to realize peak shift, to obtain an initial amplified signal corresponding to each lead signal, and the initial amplified signal is subjected to smoothing filtering to obtain a filtered amplified signal; As a possible implementation manner, the lead signal of the left segment of the extreme value point is subjected to upsampling, and the lead signal of the right segment of the extreme value point is subjected to downsampling; or, The lead signal of the left segment of the extreme value point is subjected to downsampling, and the lead signal of the right segment of the extreme value point is subjected to upsampling.

[0034] Exemplarily, the phase difference is applied to the first peak of the extreme value point , and the sampling points are adjusted by upsampling / downsampling to realize peak shift, to obtain an initial amplified signal corresponding to the lead signal . However, upsampling / downsampling may cause the waveform to appear in a stepped and non-smooth manner as shown in (a), and in a stepped and non-smooth manner as shown in Figure 4 (b). Then, the local weighted linear regression algorithm LOESS is used to perform smoothing filtering on the result, as shown in Figure 4 (c), and the specific steps are as follows: Figure 3 wherein, is the local weight of the i-th sampling point, is the order sequence of the sampling point position, is the number of sampling points, is the polynomial coefficient fitted according to the local weight , is a hyperparameter for adjusting the smoothing degree, and the smaller the parameter is, the more the high-frequency part of the signal is retained, which is sequentially brought into each sampling point to obtain the smoothed result . Referring to (c), the red original sampling points, the blue waveform without LOESS smoothing, and the green waveform with LOESS smoothing are shown in the figure; Figure 4 (d) is the weight of each sampling point, and in order to clearly show the effect, a group of weights is drawn every 10 sampling points; and Figure 4 (e) shows the result after the original waveform is multiplied by the weight of each sampling point.

[0035] S40. Constructing an amplitude scaling Gaussian mixture model for the original electroencephalogram signal, and sampling amplitude variation parameters therefrom; As a possible implementation manner, the amplitude scaling Gaussian mixture model of the original electroencephalogram signal is constructed in the following manner: ​setting a maximum scaling amplitude and a minimum scaling amplitude; Exemplarily, the minimum and maximum scaling amplitudes are set as and ; Further, when the minimum and maximum scaling amplitudes are set as and , the change of amplitude can be more comprehensively covered.

[0036] The expectation and variance of the Gaussian distribution corresponding to the maximum scaling amplitude and the minimum scaling amplitude are respectively calculated; Exemplarily, the expectation and variance of the Gaussian distribution corresponding to the maximum scaling amplitude and the minimum scaling amplitude are expressed as follows: The amplitude scaling Gaussian mixture model is fitted according to the above two sets of expectation and variance; Exemplarily, S50. For each lead signal, the filtered amplified signal is applied with the amplitude change parameter corresponding thereto to generate amplitude change, and finally the amplified signal is obtained; As a possible implementation manner, the local weighted linear regression algorithm is adopted to perform smoothing filtering on the initial amplified signal to obtain the filtered amplified signal.

[0037] Exemplarily, the first peak amplitude change parameter is sampled from the amplitude scaling Gaussian mixture model ; ; The amplitude change parameter is applied to generate amplitude change, and finally the artificial signal with both amplitude and phase difference is obtained, and the expression is as follows: .

[0038] The final output result is shown in the blue waveform in Figure 2 , and compared with the red original waveform in Figure 2 , each blue generated signal has difference in peak position and amplitude from the red input waveform.

[0039] Next, simulation experiments are performed to evaluate the feasibility and effectiveness of the amplification method proposed in the embodiment.

[0040] ​Design evaluation index (Angle Squared and Euclidean distances and Variances, ASEV); use t-SNE dimension reduction method to reduce all test times of electroencephalogram data to 3-dimensional space, then calculate the coordinates of the center points of each sample in the three-dimensional space, and finally calculate the Euclidean distance between these points, the variance in each dimension, and the angle square sum of these points relative to their average vectors. Sum these parameters to obtain the dispersion degree of each category. Figure 5 (a) is the unexpanded data of the four-class aVEPs, ASEV=0.656, and after expansion, the dispersion degree increases obviously, such as Figure 5 (b) ASEV=4.236. At the same time Figure 5 (c), (d), (e), and (f) are the results of ASEV of the original data of the four categories randomly divided into two groups respectively, and the ASEV between the original data and the artificially generated data. From Figure 5 It can be seen from the data in that the ASEV value of the artificial signal and the original signal is not much different from the ASEV value of the original signal and the original signal, which shows that the algorithm proposed in the patent can make the single test exist difference and the overall characteristics of the signal change little. Figure 6 The four subgraphs of show the classification accuracy of single test times of four subjects before and after data expansion. With the increase of the expansion number, the accuracy increases.

[0041] The electroencephalogram signal expansion method based on the Gaussian mixture distribution model provided by the application essentially belongs to a geometric method and does not need to train a model; a small sample can generate a large amount of data by applying the method, and there are multiple sampling processes from the Gaussian mixture model in the GMS algorithm, so theoretically, any number of artificial signals can be generated without repetition; the generated signal only changes the characteristics between single tests and does not change the overall characteristics.

[0042] In the second aspect, the application provides an electroencephalogram signal expansion system based on a Gaussian mixture distribution model, comprising: A signal acquisition unit acquires and obtains original electroencephalogram signals, and the original electroencephalogram signals include lead signals equal in number to the number of leads; A segmentation unit acquires a plurality of extreme points of each lead signal and segments the lead signal based on the extreme points; A phase expansion unit constructs a phase offset Gaussian mixture model of each extreme point for each lead signal, randomly extracts a phase difference from the phase offset Gaussian mixture model corresponding to each extreme point of each lead signal, applies the phase difference to the corresponding extreme point, adjusts the sampling points by upsampling or downsampling to realize peak value offset, obtains an initial expansion signal corresponding to each lead signal, and performs smoothing filtering on the initial expansion signal to obtain a filtered expansion signal; The amplitude amplification unit constructs an amplitude-scaled Gaussian mixture model for the original EEG signal and samples amplitude variation parameters from it. The phase amplitude amplification unit applies a corresponding amplitude change parameter to the filtered amplified signal for each lead signal to generate an amplitude change, ultimately obtaining the amplified signal.

[0043] As one possible way to achieve this, the first The first lead signal The Gaussian mixture model of phase shifts at each extreme point is denoted as . , ;in, For the first The first lead signal There are several extreme points; For the first The first lead signal There are several extreme points; For the first The first lead signal There are several extreme points; The distribution is a Gaussian distribution on the left side of the extreme point; It is a Gaussian distribution for the segment to the right of the extreme point.

[0044] As one possible implementation, the magnitude-scaled Gaussian mixture model is denoted as... , ;in, This represents the maximum scaling range; Minimum scaling factor; The Gaussian distribution corresponding to the minimum scaling factor; This represents the Gaussian distribution corresponding to the maximum scaling magnitude.

[0045] Thirdly, the present invention provides a terminal, including a processor and a communication interface coupled to the processor, wherein the processor is used to run computer programs or instructions to implement the electroencephalogram signal amplification method based on the Gaussian mixture distribution model proposed in this invention.

[0046] Although the invention has been described herein in conjunction with various embodiments, those skilled in the art will understand and implement other variations of the disclosed embodiments by reviewing the accompanying drawings, the disclosure, and the description of the drawings, in carrying out the claimed invention. In this specification, the word "comprising" does not exclude other components or steps, and "a" or "an" does not exclude multiple components. A single processor or other unit can implement several of the functions listed in the specification. While certain measures are described in different embodiments, this does not mean that these measures cannot be combined to produce good results.

[0047] Although the present application has been described in connection with the preferred embodiments thereof with reference to the specific content thereof, it will be apparent to those skilled in the art that various modifications and changes can be made thereto without departing from the spirit and scope of the application. Accordingly, the description and drawings are to be regarded as illustrative in nature and are not restrictive. It will be apparent that those skilled in the art can modify and adapt the application without departing from the spirit and scope of the application. Accordingly, such modifications and variations are intended to fall within the scope of the application and its equivalents.

Claims

1. A method for amplifying electroencephalogram (EEG) signals based on a Gaussian mixture distribution model, characterized in that, Includes the following steps: S10. Collect and obtain raw EEG signals, which include lead signals equal to the number of leads; S20. Obtain multiple extreme points for each lead signal and segment the lead signal based on the extreme points; S30. For each lead signal, a phase-shift Gaussian mixture model is constructed for each extreme point. The phase difference is randomly extracted from the phase-shift Gaussian mixture model corresponding to each extreme point of each lead signal. After applying the phase difference to the corresponding extreme point, the sampling point is adjusted by upsampling / undersampling to achieve peak shift, and the initial amplified signal corresponding to each lead signal is obtained. The initial amplified signal is smoothed and filtered to obtain the filtered amplified signal. S40. Construct an amplitude-scaled Gaussian mixture model for the raw EEG signal and sample amplitude variation parameters from it; S50. For each lead signal, apply the corresponding amplitude variation parameter to the filtered amplified signal to generate an amplitude variation, and finally obtain the amplified signal.

2. The EEG signal amplification method based on a Gaussian mixture distribution model according to claim 1, characterized in that, The phase-shifted Gaussian mixture model for each extreme point is constructed as follows: Calculate the mean and variance of the Gaussian distribution for the left and right segments of the extreme point, respectively; A corresponding Gaussian distribution is fitted based on the mean and variance of each Gaussian distribution. By weighted summation of the Gaussian distributions, a Gaussian mixture model of the extreme points is obtained; The weights in the Gaussian mixture model are rewritten as the ratio of the difference between adjacent extreme points to the difference between phase-separated extreme points, thus obtaining the phase-shifted Gaussian mixture model.

3. The EEG signal amplification method based on a Gaussian mixture distribution model according to claim 2, characterized in that, Let the weights corresponding to the two Gaussian distributions be denoted as follows: , , and Further rewritten as follows: in, For the first The first lead signal There are several extreme points; For the first The first lead signal There are several extreme points; For the first The first lead signal There are several extreme points.

4. The EEG signal amplification method based on a Gaussian mixture distribution model according to claim 1, characterized in that, Upsample the lead signal to the left of the extreme point and undersample the lead signal to the right of the extreme point; or, Undersample the lead signal on the left side of the extreme point and upsample the lead signal on the right side of the extreme point.

5. The EEG signal amplification method based on a Gaussian mixture distribution model according to claim 1, characterized in that, An amplitude-scaled Gaussian mixture model of the original EEG signal was constructed as follows: Set the maximum and minimum zoom levels; Calculate the expectation and variance of the Gaussian distributions corresponding to the maximum and minimum scaling values, respectively. Based on the above two sets of expectations and variances, an amplitude-scaled Gaussian mixture model is fitted.

6. The EEG signal amplification method based on a Gaussian mixture distribution model according to claim 1, characterized in that, The initial amplified signal is smoothed and filtered using a locally weighted linear regression algorithm to obtain the filtered amplified signal.

7. A brainwave signal amplification system based on a Gaussian mixture distribution model, characterized in that, include: The signal acquisition unit collects and obtains raw EEG signals, which include lead signals equal to the number of leads. The segmentation unit acquires multiple extreme points of each lead signal and segments the lead signal based on the extreme points; The phase amplification unit constructs a phase-shift Gaussian mixture model for each extreme point of each lead signal. It randomly extracts the phase difference from the phase-shift Gaussian mixture model corresponding to each extreme point of each lead signal. After applying the phase difference to the corresponding extreme point, it adjusts the sampling point by upsampling / undersampling to achieve peak shift, thereby obtaining the initial amplified signal corresponding to each lead signal. The initial amplified signal is then smoothed and filtered to obtain the filtered amplified signal. The amplitude amplification unit constructs an amplitude-scaled Gaussian mixture model for the original EEG signal and samples amplitude variation parameters from it. The phase amplitude amplification unit applies a corresponding amplitude change parameter to the filtered amplified signal for each lead signal to generate an amplitude change, ultimately obtaining the amplified signal.

8. The EEG signal amplification system based on a Gaussian mixture distribution model according to claim 7, characterized in that, No. The first lead signal The Gaussian mixture model of phase shifts at each extreme point is denoted as . , ;in, For the first The first lead signal There are several extreme points; For the first The first lead signal There are several extreme points; For the first The first lead signal There are several extreme points; The distribution is a Gaussian distribution on the left side of the extreme point; It is a Gaussian distribution for the segment to the right of the extreme point.

9. The EEG signal amplification system based on a Gaussian mixture distribution model according to claim 7, characterized in that, Amplitude scaling Gaussian mixture model is denoted as , ;in, This represents the maximum scaling range; Minimum scaling factor; The Gaussian distribution corresponding to the minimum scaling factor; This represents the Gaussian distribution corresponding to the maximum scaling magnitude.

10. A terminal, comprising a processor and a communication interface coupled to the processor, the processor being configured to run a computer program or instructions to implement the electroencephalogram signal amplification method based on a Gaussian mixture distribution model as described in any one of claims 1 to 6.

Images