CN112244876A - Electroencephalogram signal time sequence random noise removing method of adaptive filter based on variable forgetting factor - Google Patents
Electroencephalogram signal time sequence random noise removing method of adaptive filter based on variable forgetting factor Download PDFInfo
- Publication number
- CN112244876A CN112244876A CN202011061254.6A CN202011061254A CN112244876A CN 112244876 A CN112244876 A CN 112244876A CN 202011061254 A CN202011061254 A CN 202011061254A CN 112244876 A CN112244876 A CN 112244876A
- Authority
- CN
- China
- Prior art keywords
- noise
- eeg
- signal
- electroencephalogram
- time sequence
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Classifications
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/72—Signal processing specially adapted for physiological signals or for diagnostic purposes
- A61B5/7203—Signal processing specially adapted for physiological signals or for diagnostic purposes for noise prevention, reduction or removal
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/72—Signal processing specially adapted for physiological signals or for diagnostic purposes
- A61B5/7225—Details of analog processing, e.g. isolation amplifier, gain or sensitivity adjustment, filtering, baseline or drift compensation
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/72—Signal processing specially adapted for physiological signals or for diagnostic purposes
- A61B5/7235—Details of waveform analysis
- A61B5/725—Details of waveform analysis using specific filters therefor, e.g. Kalman or adaptive filters
Abstract
The invention relates to an electroencephalogram signal time sequence random noise removing method of an adaptive filter based on a variable forgetting factor, which is used for removing time sequence random noise in an electroencephalogram signal. The method comprises model training and model testing, wherein the model training comprises the step of acquiring a mixed electroencephalogram signal containing a time sequence random signal as training data; setting initial parameters of the model; generating input mixed signal sequence segment of current iteration according to mixed brain electrical signal
Generating an input reference noise segment x for a current iterationt(ii) a Updating a forgetting factor; computing noise-free electroencephalogram signal segments for a current iteration
Updating the intermediate parameters; repeating the steps until the condition is met and the training is finished; and finally, carrying out actual denoising by using the trained model. Hair brushObviously makes up for the defect that the artifact removal research on stronger time sequence random noise in the electroencephalogram signals is less, and compared with the prior method, the experimental result shows that the time sequence random strong noise removal effect on the motor imagery data set is obvious.
Description
Technical Field
The invention relates to the field of time sequence random noise removal of brain electrical signals in a motor imagery task and the field of bioinformatics.
Background
Extracranial Electroencephalography (EEG) contains a large amount of physiological and pathological information, but may also contain a large amount of artifacts, and the acquired electroencephalography is seriously affected by the artifact noise. In recent years, many artifact removal methods have emerged. At present, most of electroencephalogram artifact removal methods are focused on removing artifacts such as ocular artifacts, electrocardio artifacts, motor motion artifacts, baseline drift artifacts, power frequency interference and the like, however, models and parameters for common artifact removal need to be optimized, and the artifact removal research on strong timing random noise in electroencephalogram signals is less.
The adaptive filter is a digital filter capable of automatically and continuously adjusting parameters according to an input signal, and can solve the defect of parameter fixation of a non-adaptive filter. The electroencephalogram is a random signal, time sequence random noise added into the electroencephalogram is complex and changeable, and parameters required to be set in a model are not known in advance, so that the purpose of artifact removal can be achieved by adopting a self-adaptive filter to self-adaptively adjust the parameters. Common adaptive filters are divided into the basic LMS (Least Mean Square) and RLS (recursive Least Square) and their modified algorithms. For the self-adaptive filter, the size of the forgetting factor has a close relation with the convergence rate and stability of the algorithm,
disclosure of Invention
Against the background, the invention provides an adaptive filter VFFRLS (variable shaping Factor RLS) with a variable Forgetting Factor for removing time sequence random noise in an electroencephalogram signal, and is used for solving the problem of removing the time sequence random noise in the electroencephalogram signal.
The specific technical scheme is as follows:
comprises two stages of model training and model actual measurement, wherein,
the model training stage is used for training a model and parameters of the VFFRLS;
the model actual measurement is to carry out noise filtering on the electroencephalogram signal to be denoised by utilizing a VFFRLS model which is trained,
the model training phase specifically comprises the following steps:
1) obtaining mixed EEG signal containing time sequence random signal
2) Setting various initial parameters of VFFRLS
3) An input mixed signal sequence segment is generated.
4) An input reference noise segment is generated.
5) The forgetting factor is updated.
6) And estimating the electroencephalogram signal segment without noise.
7) A noise segment is estimated.
8) Other intermediate parameters are updated.
9) If the stop condition is not satisfied, the updating is continuously repeated.
10) And finishing training after the stopping condition is met.
Advantageous effects
The invention provides a method for removing time sequence random noise in an electroencephalogram signal by adopting a variable Forgetting Factor self-adaptive filter VFFRLS (variable shaping Factor RLS), and the removal effect of different algorithms after stronger time sequence random noise is mixed in an electroencephalogram signal of a moving object under a laboratory environment is compared. Compared with other 7 self-adaptive filtering methods, the self-adaptive filtering algorithm of the VFFRLS variable forgetting factor has the best effect of removing time sequence random noise in the electroencephalogram signal of the motor imagery, and the average Root Mean Square Error (RMSE) of the VFFRLS algorithm is the minimum and the SNR is the maximum. The VFFRLS effectively completes artifact removal on the electroencephalogram signal, and finally the electroencephalogram signal without obvious noise is obtained.
Drawings
FIG. 1 is a block diagram of a process according to the present invention;
Detailed Description
The invention is further described with reference to the accompanying drawings and the detailed description. In this embodiment, the method for removing the artifacts from the electrooculogram or electromyogram signals in the hospital comprises the following steps:
training phase
Step 1) selecting an electroencephalogram data set.
Adopting a second Ia data set of BCI (brain Computer interface) brain-Computer interface competition, and recording samples of 561 electroencephalogram signals as EEGorgThere were 6 spot electrode sample segments in each sample, 896 sample spots in each segment, and a useful time period was 3.5 s.
Step 2) generating random noise.
noise＝a0·frnd(m,n)
In the formula: a is0M is 1 and n is 896, which is the amplitude coefficient of the noise. f. ofrnd(m, n) is a function for generating random numbers between 0 and 1 in m × n dimensions, and noise is a row vector for randomly generating 1 × 896 dimensions.
Step 3) generating a mixed electroencephalogram signal containing a time sequence random signal in the following mode:
EEGnoise(i)＝EEGorg(i)
+1·a1·noise(i)
+1i＞1·a2·noise(i-1)
+1i＞2·noise(i-2)
in the formula: EEG (electroencephalogram)org(i) Representing the ith sample point in a point electrode sample segment, 1 ≦ i ≦ n, EEGnoiseIs a mixed signal added with time sequence random strong noise, adds different weight noises in a delay way to embody the time sequence, a1And a2Is a weight coefficient, where 0 < a1＜1，0＜a2Is less than 1. Both are 1 × 896 dimensional row vectors, 1 ≦ i ≦ 896, 1i＞1This means that when i > 1, 1 is taken, and conversely 0 is taken.
And 4) setting various initial parameters of the VFFRLS.
Initializing forgetting factor lambda0、δ、w0、y0、M、P0Parameter, wherein λ0＝0.99，δ＝0.001，M＝3，l＝M-1＝2，P0＝δ·M，y0、w0Is a 0 vector of 1 x 3 dimensions,
step 5) EEG (electroencephalogram) based on mixed EEG signalsnoiseGenerating the input mixed signal sequence segment of the current t-th iteration
Generating an input mixed signal sequence segment
In the formula:
is a 1 × (l +1) -dimensional vector, EEGnoise(t) expression of vector EEGnoiseData at index t, such as when t is 5, then
Is derived from EEGnoiseStarts to truncate backward a sequence fragment of length l + 1.
Step 6) generating an input reference noise segment xt。
xt＝[noise(t),…,noise(t+l)]
In the formula: x is a vector of dimension 1 x (l +1), noise (t) represents the data of the vector noise at index t,
step 7) updating the forgetting factor lambdat。
If the iteration number is more than or equal to 1 and t is less than or equal to lambdanum-1 updating the formula if λnumN-M +1 skipping formula with-1 < t ≦.
In the formula: lambda [ alpha ]numAnd taking 90 and taking N as 896. The forgetting factor has important influence on the algorithm and is too stableThe method has the advantages that the method is good in qualitative performance, but poor in tracking capability, too small in tracking capability, but poor in stability, generally speaking, the forgetting factor is 0.9-1.0, the algorithm effect is good, the initial value is 0.99, the relatively small value is started, the tracking speed is improved, the forgetting factor is gradually improved along with the increase of time, and the stability is improved.
Step 8) estimating the noise-free EEG signal segment
Wherein, the initial value is as follows,
the update is as follows,
EEGest(t+l)＝EEGnoise(t+l)-noiseest(t+l)
wherein the content of the first and second substances,
step 9), updating the intermediate parameter wt。
Step 10), updating the intermediate parameter wt、Kt、PtAnd other intermediate parameters.
Step 11), updating the intermediate parameter Pt。
In the formula: EEG (electroencephalogram)estIs an estimated EEG signal, EEG, free of random noiseest(t + l) representing EEGestData at the t + l index in the vector, PtThe same applies to the small brackets and subscripts of the other variables in the above formula, which are t, for the updated P value at the t-th iteration.
Step 12), if t is more than 1 and less than or equal to N-M +1, the steps 5-12) are repeatedly updated, and after one-time noise filtering is finished, the denoised EEG and pure EEG are calculatedorgAnd whether the Root Mean Square Error (RMSE) and the signal-to-noise ratio (SNR) between the two meet the requirements, if so, finishing the training, otherwise, continuing the training.
Step 13), after model training is completed, in order to enable the actual measurement result to be more accurate, the ocular or electromyographic signals of the hospital are required to be used for artifact removal and fine adjustment of parameters, and at the moment, the mixed EEG signal EEG containing time sequence random signalsnoiseThe EEG signals are pure EEG signals of EEG lead signals which are actually collected by a hospital and are influenced by the electro-oculogram EOG signals or the electromyogram EMG signals; the noise is a reference eye EOG signal or an electromyographic EMG signal acquired in a hospital, or is converted into a signal component which is solved by various algorithms and is most related to the EOG signal or the ECG signal; clean EEG when root mean square error RMSE and signal-to-noise ratio SNR are calculatedorgEEG lead signals produced for a hospital that are unaffected by an electrooculographic EOG signal (e.g., not blinking) or electromyographic EMG signals (e.g., not having significant muscle tremor).
And step 14) performing electrocardio and myo-electric artifact removal on the electroencephalogram signals containing the artifacts actually acquired by the hospital by using the trained model.
The invention includes, but is not limited to, artifact removal of hospital measured electro-ocular or electromyographic signals.
The effectiveness of the VFFRLS algorithm of the present invention can be evaluated by calculating the root mean square error RMSE of a single sample and calculating the signal-to-noise ratio SNR of the single sample. Smaller RMSE is better, smaller values indicate that the estimated noise-free signal is closer to the original noise-free signal, while larger SNR values are better.
RMSE is rootRoot Mean Square Error (Root Mean Square Error) can be used for measuring the removal effect of time sequence random strong noise in the motor imagery electroencephalogram on the time sequence and evaluating the calculated EEG (electroencephalogram) without noise signalsestAnd original EEG without noise signalsorgThe smaller the difference, the better the value is, the closer the estimated noise-free signal is to the original noise-free signal, which is expressed by the following formula:
in the formula: n is the number of sample points, EEGorgEEG (electroencephalogram) signal without joint noise and conducted by certain electrode in certain sampleestThe noise-free EEG signal of the electrode lead is estimated by various adaptive filters.
Calculating the SNR of a single sample
SNR is the Signal-to-Noise Ratio (Signal Noise Ratio) and is originally the Ratio of Signal to Noise. Representing raw noise-free signal EEG by signal-to-noise ratio hereinorgWith an estimated EEG signal free of noiseestThe larger the value, the better the relationship between the errors of (1).
In the formula: n, EEGorg、EEGestThe meaning of the parameters is consistent with that of the parameters in RMSE.
To verify the validity of the VFFRLS algorithm, tests were performed in a laboratory environment. During testing, a second Ia data set of BCI (brain Computer interface) brain-Computer interface competition is adopted, and samples of 561 electroencephalogram signals are recorded as EEGorgThere were 6 spot electrode sample segments in each sample, 896 sample spots in each segment, and a useful time period was 3.5 s. Generating random noise, noise ═ a0·frnd(m, n), wherein: a is0M is 1 and n is 896, which is the amplitude coefficient of the noise. f. ofrnd(m, n) is a function generating random numbers between 0 and 1 in m n dimensions, noise is a randomly generated 1 x 896 dimensional row vector. Generating a mixed electroencephalogram signal containing a time sequence random signal in the following mode:
EEGnoise(i)＝EEGorg(i)
+1·a1·noise(i)
+1i＞1·a2·noise(i-1)
+1i＞2·noise(i-2)
in the formula: EEG (electroencephalogram)org(i) Representing the ith sample point in a point electrode sample segment, 1 ≦ i ≦ n, EEGnoiseIs a mixed signal added with time sequence random strong noise, adds different weight noises in a delay way to embody the time sequence, a1And a2Is a weight coefficient, where 0 < a1＜1，0＜a2Is less than 1. The mixed signal after the random noise is added is very different from the original brain wave signal. Both are 1 × 896 dimensional row vectors, 1 ≦ i ≦ 896, 1i＞1This means that when i > 1, 1 is taken, and conversely 0 is taken.
TABLE 1 comparison of the mean RMSE values (in. mu.V) for different adaptive filter algorithms
TABLE 2 comparison of average SNR values (in db) for different adaptive filtering algorithms
In tables 1 and 2, the horizontal direction represents different adaptive filtering algorithms, and the vertical direction represents different electrodes. As can be seen from Table 1, in the 8 kinds of adaptive filters, the mean root mean square error RMSE of the pure EEG signal estimated by VFFRLS in 6 electrodes and the original EEG signal without random noise is the minimum. As can be seen from Table 2, the average SNR between the clean EEG signal estimated by the VFFRLS adaptive filter on 6 electrodes and the original EEG signal without timing random noise is the largest. The smaller the RMSE, the better the evaluation index, and the larger the SNR, the better the evaluation index. Experiments prove that the VFFRLS algorithm adopted by the invention can effectively remove the artifact of the electroencephalogram signal.
Claims (9)
1. A method for removing EEG signal time sequence random noise based on self-adaptive filter of variable forgetting factor is characterized by comprising two stages of model training and model actual measurement, wherein,
the model training stage is used for training a model and parameters of the VFFRLS;
in the model actual measurement stage, noise filtering is carried out on the electroencephalogram signal to be denoised by utilizing the VFFRLS model which is trained, so that the denoised electroencephalogram signal is obtained;
the model training stage specifically comprises the following steps:
1) obtaining mixed EEG (electroencephalogram) signal containing time sequence random signalnoiseAs training data;
2) setting initial parameters of VFFRLS, including initializing forgetting factor lambda0、λnum、δ、w0、y0、M、K0、P0A parameter;
3) from mixed EEG signalsnoiseGenerating the input mixed signal sequence segment of the current t-th iteration
4) Generating an input reference noise segment x for a current t-th iterationtFor generating an estimated noise segment;
5) updating a forgetting factor;
6) computing a noise-free EEG segment for a current t-th iteration
The estimated pure electroencephalogram signal is obtained;
7) updating the intermediate parameters;
8) if t is more than 1 and less than or equal to N-M +1, N is equal to N, and N is the number of sampling points in one sample, repeating the updating steps 3) to 8) until the completionOne iteration of all results
The denoised electroencephalogram signal is obtained; calculating EEG (electroencephalogram) of denoised EEG (electroencephalogram) signal and pure EEG signalorgWhether the Root Mean Square Error (RMSE) and the signal-to-noise ratio (SNR) between the two meet the requirements or not is judged, if so, the training is ended, otherwise, the training is continued;
2. the electroencephalogram signal time sequence random noise removing method based on the adaptive filter with the variable forgetting factor, according to claim 1, is characterized in that: hybrid EEG for the model training phasenoiseThe construction method comprises the following steps:
step 1.1) selecting a plurality of noise-free electroencephalogram signal samples to be recorded as EEGorgEach sample comprises m point electrode sample sections, and each section comprises n sample points;
step 1.2) is pure EEGorgEach point of the electrode sample segment generates a random noise,
noise＝a0·frnd(m,n)
in the formula: a is0For the amplitude coefficient of noise, m is 1, frnd(m, n) is a function for generating random numbers between 0 and 1 in m × n dimensions, and noise is a row vector for randomly generating 1 × n dimensions;
step 1.3) generating mixed EEG (electroencephalogram) signals containing time sequence random signals for each point electrode sample sectionnoise＝{EEGnoise(1),...,EEGnoise(n) }, wherein, EEGnoise(i) The generation method is as follows:
EEGnoise(i)＝EEGorg(i)+1·a1·noise(i)+1i＞1·a2·noise(i-1)+1i＞2·noise(i-2)
in the formula: EEG (electroencephalogram)org(i) Representing the ith sample point in a point electrode sample segment, i is more than or equal to 1 and less than or equal to n, noise (i-1) and noise (i-2) represent time delay, noise with different weights is added in the time delay to represent the time sequence, a1And a2Is a weight coefficient, where 0 < a1＜1，0＜a2＜1，1i＞1This means that when i > 1, 1 is taken, and conversely 0 is taken.
3. The electroencephalogram signal time sequence random noise removing method based on the adaptive filter with the variable forgetting factor, according to claim 1, is characterized in that: further, before the model test, the model is finely tuned and transferred, and when the ocular or electromyographic signals of the hospital are finely tuned and transferred in the artifact removal mode, the EEG (electroencephalogram) signal containing the time sequence random signals is mixednoiseThe EEG lead signal is a pure EEG signal affected by the eye electrical EOG signal or the electromyographic EMG signal collected in a hospital, the noise is a reference eye electrical EOG signal or the electromyographic EMG signal collected in the hospital, and the EEG is a reference eye electrical EOG signal or the electromyographic EMG signal collected in the hospitalorgThe EEG lead signals which are collected by hospitals and are not influenced by the eye electrical EOG signals or the electromyographic EMG signals.
4. The EEG signal time sequence random noise removing method of the adaptive filter based on the variable forgetting factor as claimed in claim 1, wherein the input mixed signal sequence segment of the current t-th iteration in step 3
Comprises the following steps: starting from t, from mixed EEG signalsnoiseThe truncated signal segment of length l +1 is shown below
In the formula: EEG (electroencephalogram)noise(t) representation of Mixed electroencephalogram EEGnoiseData at index t.
5. The EEG signal time sequence random noise removing method based on the adaptive filter with variable forgetting factor of claim 1, wherein the input reference noise segment x of the current t-th iteration in step 4tComprises the following steps: cutting from randomly generated noise by taking t as a starting pointSignal fragment of length l +1, represented as follows:
xt＝[noise(t),…,noise(t+l)]
in the formula: noise (t) represents the data of noise at index t.
6. The EEG signal time sequence random noise removing method of the adaptive filter based on the variable forgetting factor as claimed in claim 1, wherein the step 5 of updating the forgetting factor λtThe specific method of (3) is as follows;
if the iteration number is more than or equal to 1 and t is less than or equal to lambdanum-1, updating according to the following formula;
if λnumT is more than-1 and less than or equal to N-M +1, and a formula is skipped;
7. the method for removing the random noise of the electroencephalogram signal time sequence of the adaptive filter based on the variable forgetting factor according to the claim 1, characterized in that the step 6 is
The initial values are as follows:
the update is as follows,
EEGest(t+l)＝EEGnoise(t+l)-noiseest(t+l)
wherein the content of the first and second substances,
8. the electroencephalogram signal time sequence random noise removal method based on the adaptive filter with the variable forgetting factor, according to claim 1, wherein the step 7 specifically comprises the following contents:
updating the intermediate parameter wt，
Updating the intermediate parameter Kt，
Updating the intermediate parameter Pt，
9. The EEG signal time sequence random noise removing method based on the adaptive filter with the variable forgetting factor as claimed in claim 1, wherein when artifact removal is performed on the measured eye electrical or electromyographic signals of a hospital, EEG signals to be denoised in the measured stage are used for EEGnoiseThe EEG lead signals are collected by a hospital and influenced by the eye electric EOG signals or the electromyographic EMG signals, and the noise is the reference eye electric EOG signals or the electromyographic EMG signals collected by the hospital.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011061254.6A CN112244876A (en) | 2020-09-30 | 2020-09-30 | Electroencephalogram signal time sequence random noise removing method of adaptive filter based on variable forgetting factor |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011061254.6A CN112244876A (en) | 2020-09-30 | 2020-09-30 | Electroencephalogram signal time sequence random noise removing method of adaptive filter based on variable forgetting factor |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN112244876A true CN112244876A (en) | 2021-01-22 |
Family
ID=74233595
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202011061254.6A Pending CN112244876A (en) | 2020-09-30 | 2020-09-30 | Electroencephalogram signal time sequence random noise removing method of adaptive filter based on variable forgetting factor |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112244876A (en) |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113208614A (en) * | 2021-04-30 | 2021-08-06 | 南方科技大学 | Electroencephalogram noise reduction method and device and readable storage medium |
| CN114098764A (en) * | 2021-11-05 | 2022-03-01 | 北京大学 | Data processing method and device, electronic equipment and storage medium |
| CN114190953A (en) * | 2021-12-09 | 2022-03-18 | 四川新源生物电子科技有限公司 | Training method and system of electroencephalogram signal noise reduction model for electroencephalogram acquisition equipment |
| CN115358367A (en) * | 2022-08-09 | 2022-11-18 | 浙江大学 | Dynamic self-adaptive brain-computer interface decoding method based on multi-model learning integration |
| CN116602691A (en) * | 2023-07-14 | 2023-08-18 | 北京元纽科技有限公司 | Denoising method and device for electroencephalogram signals, electronic equipment and storage medium |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20060284600A1 (en) * | 2005-06-21 | 2006-12-21 | Verbrugge Mark W | Method for control and monitoring using a state estimator having variable forgetting factors |
| WO2007137484A1 (en) * | 2006-05-11 | 2007-12-06 | Shanghai Jiao Tong University | A channel estimation method and the device thereof |
| CN109726501A (en) * | 2019-01-11 | 2019-05-07 | 武汉理工大学 | RLS lithium battery model parameter on-line identification method based on variable forgetting factor |
-
2020
- 2020-09-30 CN CN202011061254.6A patent/CN112244876A/en active Pending
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20060284600A1 (en) * | 2005-06-21 | 2006-12-21 | Verbrugge Mark W | Method for control and monitoring using a state estimator having variable forgetting factors |
| WO2007137484A1 (en) * | 2006-05-11 | 2007-12-06 | Shanghai Jiao Tong University | A channel estimation method and the device thereof |
| CN109726501A (en) * | 2019-01-11 | 2019-05-07 | 武汉理工大学 | RLS lithium battery model parameter on-line identification method based on variable forgetting factor |
Non-Patent Citations (2)
| Title |
|---|
| JUNFENG WANG: "A Variable Forgetting Factor RLS Adaptive Filtering Algorithm", 2009 3RD IEEE INTERNATIONAL SYMPOSIUM ON MICROWAVE, ANTENNA, PROPAGATION AND EMC TECHNOLOGIES FOR WIRELESS COMMUNICATIONS, pages 1127 - 1130 * |
| SRIDHAR CHINTALA 等: "GVFF-RLS Adaptive Algorithm for Elimination of Ocular Artifacts from EEG Signals", 2019 IEEE 16TH INDIA COUNCIL INTERNATIONAL CONFERENCE (INDICON), pages 1 - 3 * |
Cited By (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113208614A (en) * | 2021-04-30 | 2021-08-06 | 南方科技大学 | Electroencephalogram noise reduction method and device and readable storage medium |
| CN114098764A (en) * | 2021-11-05 | 2022-03-01 | 北京大学 | Data processing method and device, electronic equipment and storage medium |
| CN114098764B (en) * | 2021-11-05 | 2024-02-20 | 北京大学 | Data processing method, device, electronic equipment and storage medium |
| CN114190953A (en) * | 2021-12-09 | 2022-03-18 | 四川新源生物电子科技有限公司 | Training method and system of electroencephalogram signal noise reduction model for electroencephalogram acquisition equipment |
| CN115358367A (en) * | 2022-08-09 | 2022-11-18 | 浙江大学 | Dynamic self-adaptive brain-computer interface decoding method based on multi-model learning integration |
| CN115358367B (en) * | 2022-08-09 | 2023-04-18 | 浙江大学 | Dynamic self-adaptive brain-computer interface decoding method based on multi-model learning integration |
| CN116602691A (en) * | 2023-07-14 | 2023-08-18 | 北京元纽科技有限公司 | Denoising method and device for electroencephalogram signals, electronic equipment and storage medium |
| CN116602691B (en) * | 2023-07-14 | 2023-10-10 | 北京元纽科技有限公司 | Denoising method and device for electroencephalogram signals, electronic equipment and storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN112244876A (en) | Electroencephalogram signal time sequence random noise removing method of adaptive filter based on variable forgetting factor | |
| CN108714026B (en) | Fine-grained electrocardiosignal classification method based on deep convolutional neural network and online decision fusion | |
| Tracey et al. | Nonlocal means denoising of ECG signals | |
| CN108158573B (en) | Electrocardiosignal noise reduction method based on adaptive threshold wavelet transformation | |
| Smital et al. | Adaptive wavelet wiener filtering of ECG signals | |
| CN110353672B (en) | System and method for removing eye artifacts in electroencephalogram signals | |
| CN110680308B (en) | Electrocardiosignal denoising method based on fusion of improved EMD and threshold method | |
| Jafarifarmand et al. | Real-time ocular artifacts removal of EEG data using a hybrid ICA-ANC approach | |
| US5908393A (en) | Reducing noise in a biological signal | |
| Rahman et al. | Noise cancellation in ECG signals using computationally simplified adaptive filtering techniques: Application to biotelemetry | |
| CN111631710B (en) | Method for eliminating myoelectric artifacts in state-related dynamic electroencephalogram signals | |
| CN114190944B (en) | Robust emotion recognition method based on electroencephalogram signals | |
| CN113456086A (en) | Medical auxiliary system based on brain waves | |
| Al-Naggar et al. | Performance of adaptive noise cancellation with normalized last-mean-square based on the signal-to-noise ratio of lung and heart sound separation | |
| CN111493821B (en) | PPG signal real-time denoising method based on MODWT and median filtering | |
| CN111543981B (en) | Dynamic electrocardiogram real-time filtering method based on segmented MODWT and adaptive threshold | |
| CN117158999A (en) | Electroencephalogram signal denoising method and system based on PPMC and self-adaptive VMD | |
| JPH11318843A (en) | Electroencephalography | |
| CN111493864B (en) | EEG signal mixed noise processing method, equipment and storage medium | |
| CN115462803A (en) | BG-Attention-based electroencephalogram signal denoising method, device and storage medium | |
| Lin et al. | Online recursive ICA algorithm used for motor imagery EEG signal | |
| CN110477916B (en) | Method and device for identifying motion starting time of surface electromyographic signal | |
| Biswas et al. | Removing baseline wander from ECG signal using wavelet transform | |
| Shahbakhti et al. | Elimination of blink from EEG by adaptive filtering without using artifact reference | |
| JP2002153435A (en) | Brain wave recording processing method |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |