CN106951835A - A kind of EEG signals noise remove method - Google Patents
A kind of EEG signals noise remove method Download PDFInfo
- Publication number
- CN106951835A CN106951835A CN201710123597.2A CN201710123597A CN106951835A CN 106951835 A CN106951835 A CN 106951835A CN 201710123597 A CN201710123597 A CN 201710123597A CN 106951835 A CN106951835 A CN 106951835A
- Authority
- CN
- China
- Prior art keywords
- noise
- eeg signals
- wavelet
- signal
- eigencomponent
- 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
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
- G06F2218/02—Preprocessing
- G06F2218/04—Denoising
- G06F2218/06—Denoising by applying a scale-space analysis, e.g. using wavelet analysis
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Artificial Intelligence (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Signal Processing (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Measurement And Recording Of Electrical Phenomena And Electrical Characteristics Of The Living Body (AREA)
Abstract
The present invention relates to a kind of EEG signals noise remove method based on DTCWT EEMD algorithms, comprise the following steps:Row set empirical mode decomposition is entered to the EEG signals collected using EEMD algorithms, obtains including the eigencomponent of different frequency range information;The main some components for including high-frequency random noises, are decomposed to it using DTCWT algorithms, obtain a series of corresponding wavelet coefficients of noise component(s) in eigencomponent obtained by selection;For the wavelet coefficient corresponding to random noise, wavelet coefficient processing is carried out using Soft thresholding, by random noise correspondence wavelet coefficient zero setting, and then the effect of noise remove is realized;Wavelet coefficient Jing Guo noise removal process is carried out to the reconstruct of DTCWT algorithm for inversions, the eigencomponent after denoising is obtained, then carries out being added reconstruct with the eigencomponent of remaining representation signal useful information, final denoising EEG signals are obtained.The integrality of useful information has been effectively ensured in the present invention while noise remove.
Description
Technical field
The present invention relates to EEG signals noise-removed technology field, more particularly to a kind of EEG signals noise remove method.
Background technology
Due to complexity specific to brain and feature, the mankind thirst for understanding its operating mechanism always, so as to protect big
Brain, treats brain diseases, or even replicates brain, realizes many multi-functional artificial substitutings of brain, therefore, explores brain, discloses big
Brain Cognition Mechanism be the mankind explore self, important science proposition extremely challenging during the knowledge of natural environment.
Brain by substantial amounts of nerve cell control with coordinate language in mankind's daily life, thinking, cognition, mood,
A series of operations such as motion, and the activity of nerve cell is then along with the activity of electricity.Thus people recognize can be by this
The capture and research of electric signal are planted, carrys out a series of artificial control effects for reappearing human brain.So, with leading in recent years
The fast development of the arts such as letter, computer, biomedicine is with mutually merging, and a kind of new brain is communicated with the external world
Technology --- brain-computer interface (Brain Computer Interface, BCI) arises at the historic moment.Brain-computer interface has built one kind
The passage of information exchange is directly carried out with external environment independent of brain surrounding muscles and the brain of nerve fiber.
Developing rapidly for BCI systems is allowed in terms of recovering aid, traffic control, military game, Entertainment all obtain
With extensive use.Such as braking of brain control electric wheelchair, brain electric car, brain electric fatigue drive detection, the electric panzer control of brain, brain electricity
The a series of BCI products such as virtual world are even more to emerge in an endless stream.It can be said that BCI technologies have expanded brain information and external environment
Interactive channel, enhances the understanding to brain Cognitive Mode, helps to explain the essence of mankind's activity, is greatly promoted big
The research in brain cognitive science and Neurobiology field, more conducively provides for the whole society and meets the numerous of human thinking's pattern and surpass just
Victory service, with wide Research Prospects.
However, when realizing some control effects using BCI systems, on the pretreatment step of the EEG signals extracted
Rapid is always the emphasis of researcher's concern, and noise remove problem is exactly the crucial part in EEG signals preprocessing process.
EEG signals are to the spontaneity of cerebral cortex cells group, the electrical activity of rhythmicity using sensor in human brain scalp
Obtained from collection, but because highly sensitive eeg amplifier is easily influenceed by external environment in gatherer process,
Along with the various factors and EEG signals own characteristic of subject itself, so being highly prone to neural source in gatherer process
The interference of noise and non-neural source noise, can pass through for the interference from non-noise during EEG signals are gathered
Eeg signal acquisition device removes some interference, but for the interference of noise, it is necessary to pass through certain noise remove method
To carry out denoising Processing, the signal to noise ratio of signal is improved.
EEMD Denoising Algorithms widely used at present, are mainly decomposed by EEMD and obtain multiple IMF components, preceding several
Component mainly includes the radio-frequency component in signal, and component below is mainly comprising the low-frequency component in signal, and noise intensity is opposite
The increase of IMF levels also can be more and more weaker, i.e. the low-frequency component of signal is based on effective information, and radio-frequency component is then containing a large amount of
Noise.Noise-eliminating method universal at present is that the direct preceding several high-frequency I MF components removed during EEMD is decomposed carry out de-noising, still,
Effective information composition therein can be so eliminated while the noise contribution in removing high frequency, it is impossible to ensure EEG signals
Completeness and efficiency.
The content of the invention
The technical problems to be solved by the invention are to provide a kind of EEG signals noise remove method, and this method is based on
DTCWT-EEMD blending algorithms, it largely compensate for the defect of traditional denoising method, while realizing noise remove
The integrality of useful information has been effectively ensured.
The technical solution adopted for the present invention to solve the technical problems is:A kind of EEG signals noise remove method is provided,
Comprise the following steps:
(1) row set empirical mode decomposition is entered to the EEG signals collected using EEMD algorithms, obtains including different frequencies
The eigencomponent of segment information;
(2) the main some components for including high-frequency random noises in the eigencomponent obtained by choosing, using DTCWT algorithms pair
It is decomposed, and obtains a series of corresponding wavelet coefficients of noise component(s);
(3) wavelet coefficient corresponding to random noise, carries out wavelet coefficient processing using Soft thresholding, will make an uproar at random
Sound correspondence wavelet coefficient zero setting, and then realize the effect of noise remove;
(4) wavelet coefficient Jing Guo noise removal process is carried out to the reconstruct of DTCWT algorithm for inversions, obtains intrinsic after denoising
Component, then carry out being added reconstruct with the eigencomponent of remaining representation signal useful information, obtain final denoising EEG signals.
The step (1) is specially:White noise is added into signal, noise mixture is constituted;Noise mixture is carried out
EMD is decomposed, and resolves into intrinsic component combination;Repeat the above steps so that the white noise added every time is differed, obtain multiple
Eigencomponent is combined, and obtains decomposition result to the combination averaging of multiple eigencomponents.
In the step (2) decomposition and reconstruct of signal, institute are realized using two parallel real wavelet transform trees
It is respectively real part tree and imaginary part tree to state two parallel real wavelet transform trees, and a different real number filter has been used respectively
Ripple device group, the real part coefficient and imaginary part coefficient of dual-tree complex wavelet are respectively obtained by the two wave filter groups.
Beneficial effect
As a result of above-mentioned technical scheme, the present invention compared with prior art, has the following advantages that and actively imitated
Really:The present invention can effectively remove noise while, to the greatest extent can remain useful signal composition, it is ensured that the validity of signal
And integrality.Meanwhile, the present invention is stable effective, goes for the noise remove work of any EEG signals, also eliminates people
It is cumbersome that work is intervened.
Brief description of the drawings
Fig. 1 is the flow chart of the present invention;
Fig. 2 is EMD algorithm flow charts;
Fig. 3 is dual-tree complex wavelet decomposing schematic representation;
Fig. 4 is brain electricity emulation signal schematic representation;
Fig. 5 is noisy brain electricity emulation signal schematic representation;
Fig. 6 is the denoising effect schematic diagram using the present invention;
Fig. 7 is that algorithm compares root-mean-square error performance indications schematic diagram;
Fig. 8 is that algorithm compares signal-to-noise performance index schematic diagram.
Embodiment
With reference to specific embodiment, the present invention is expanded on further.It should be understood that these embodiments are merely to illustrate the present invention
Rather than limitation the scope of the present invention.In addition, it is to be understood that after the content of the invention lectured has been read, people in the art
Member can make various changes or modifications to the present invention, and these equivalent form of values equally fall within the application appended claims and limited
Scope.
Embodiments of the present invention are related to a kind of EEG signals noise remove method based on DTCWT-EEMD algorithms, such as scheme
Shown in 1, comprise the following steps:Row set empirical mode decomposition is entered to the EEG signals collected using EEMD algorithms, wrapped
The eigencomponent of the information containing different frequency range;The main some components for including high-frequency random noises in eigencomponent obtained by selection,
It is decomposed using DTCWT algorithms, a series of corresponding wavelet coefficients of noise component(s) are obtained;For corresponding to random noise
Wavelet coefficient, wavelet coefficient processing is carried out using Soft thresholding, by random noise correspondence wavelet coefficient zero setting, and then realizes and makes an uproar
The effect that sound is removed;Wavelet coefficient Jing Guo noise removal process is carried out to the reconstruct of DTCWT algorithm for inversions, the sheet after denoising is obtained
Component is levied, then carries out being added reconstruct with the eigencomponent of remaining representation signal useful information, final denoising brain telecommunications is obtained
Number.
Present invention relates generally to two kinds of algorithms of DTCWT and EEMD, it is specifically described as follows:
EEMD algorithms are the evolution algorithms developed by classical EMD, and the purpose of EMD algorithms is by the bad letter of performance
The intrinsic mode functions IMF of one group of better performances number is decomposed into, when decompositing each IMF components come and containing the difference of original signal
Between yardstick local feature information.Wherein, each IMF components have following feature:
(1) from global property, extreme value points must it is consistent with zero passage points or at most difference one.
(2) in some partial points, maximum envelope and minimum envelope the point value arithmetic average and be zero.
The idiographic flow of EMD algorithms is as shown in Figure 2.Divide when a non-stationary signal is decomposed into IMF one by one by EMD
During amount, it is possible to specific signal component is analyzed and processed according to these IMF components, however, when the extreme point point of signal
When cloth is uneven, it is possible to cause to occur multiple magnitude signals in an IMF component or the signal distributions of a similar scale exist
In different IMF components, i.e. modal overlap phenomenon.
In order to overcome modal overlap phenomenon, EEMD algorithms are further provided, the principle of EEMD algorithms is in original signal
White noise several times is added, using the combination of signal and noise as a signal band decomposed signal, is tiled using white noise uniform
The characteristic of distribution, when signal loading, which spreads all over this time frequency space, to be distributed in consistent white noise background, different time scales
It is suitable with reference on yardstick that signal can be distributed in automatically, and the signal for loading white noise now is carried out into EMD decomposition again, avoided
Modal overlap, and due to the characteristic of white noise zero-mean, obtained each IMF components are decomposed to EMD and seek population mean, with regard to that can disappear
Except the influence of added white noise, the true mode approached.Idiographic flow is as follows:
(1) toward addition white noise k × σ in signal x (t)x× n (t), wherein, n (t) is normalized white noise, σxFor letter
Number standard deviation, k is the ratio of white noise standard deviation and signal standards difference, constitutes following noise mixture:
X (t)=x (t)+k × σx×n(t) (1)
(2) noise mixture is carried out EMD decomposition, resolves into IMF combinations:
In above formula, cjJth IMF eigencomponents are represented, m are had;rmRepresent to decompose gained remainder.
(3) repeat step 1 and step 2, add different white noises every time:
Xi(t)=x (t)+k × ni(t) (3)
Resolve into IMF:
(4) n times are repeated, each IMF is averaging:
Last decomposition result is:
Due to the zero mean characteristic of white noise, after the result averaged that these are repeatedly decomposed, noise most at last by
The effect of elimination is offset and reached to greatest extent, and the result of population mean can just regard actual signal.
To sum up, when being decomposed using EEMD to EEG signals, multiple IMF components can be obtained, preceding several components are main
Comprising the radio-frequency component in signal, component below is mainly comprising the low-frequency component in signal, and noise intensity is with IMF levels
Increase also can be more and more weaker, i.e. the low-frequency component of signal is based on effective information, and radio-frequency component then contains substantial amounts of noise.In
This, you can choose high-frequency I MF compositions and handled, reach the purpose of de-noising.
, it is necessary to take dual-tree complex wavelet transform (DTCWT) to carry out deep layer when further handling high-frequency I MF components
Secondary signal component is decomposed, and dual-tree complex wavelet transform here is developed by wavelet transform here.
The flexible rectangular projection different with a series of resolution ratio of translation composition that wavelet transform passes through wavelet basis function
The corresponding base of space machine, with this group of basis representation or approaches a certain signal, thus has good frequency domain resolution in low frequency, in height
Frequency has good temporal resolution, also by feat of spies such as good Time-Frequency Localization characteristic, multi-resolution characteristics, decorrelations
Point, there is good performance in signal noise silencing field.The multiresolution analysis characteristic of small echo can be carried out signal under different scale
The decomposition of multiresolution, and the mixed signal that the various different frequencies of weave in are constituted resolves into the son letter of different frequency range
Number, thus there is the ability handled by frequency band to signal, so, the basic step of de-noising is exactly as needed, will contain noise
Signal decomposed under a certain yardstick in different frequency bands, then the frequency band zero setting residing for noise is then subjected to small echo again
Reconstruct, so as to reach the purpose of de-noising.
But, wavelet transform may produce serious frequency alias phenomenon in actual signal analyzing and processing.When
When primary signal contains the periodic signal of several different frequencies, the frequency for the different levels signal that wavelet transform is decomposited
Rate may can include other frequency contents.And dual-tree complex wavelet is showed during decomposed signal than wavelet transform
It is more thorough, the generation of frequency alias phenomenon can be more effectively avoided, the frequency information of detail section is showed well.So,
When carrying out denoising to EEG signals, different frequency bands decomposition should be carried out to signal using dual-tree complex wavelet transform.
The exploded pictorial of dual-tree complex wavelet transform is as shown in Figure 3.
In figure 3, the decomposition and reconstruct of signal are realized using two parallel real wavelet transform trees, be referred to as respectively
For real part tree and imaginary part tree, two wavelet transform trees have used a different real filter group respectively, by this
Two wave filter groups can respectively obtain the real part coefficient and imaginary part coefficient of dual-tree complex wavelet.
The wavelet coefficient of real part treeAnd scale coefficientIt is as follows respectively:
Wherein, j cuts for scale factor, j=1,2 ..., J, ψh() represents the corresponding wavelet basis of real part high-pass filter
Function, φh() represents that the corresponding wavelet basis function of real part low pass filter, t represent that time coefficient, n represent wavelet decomposition
The number of plies.
Similarly, the wavelet coefficient of imaginary part treeAnd scale coefficientIt is as follows respectively:
Wherein, ψg() represents the corresponding wavelet basis function of imaginary part high-pass filter, φg() represents imaginary part low pass
The corresponding wavelet basis function of wave filter.
Dual-tree complex wavelet transform decompose after each layer wavelet coefficient and scale coefficient be real part, imaginary part two parts coefficient it
With:
Noisy primary signal is after dual-tree complex wavelet transform, and the energy of useful signal is concentrated mainly on limited coefficient
In, and the Energy distribution of noise is in whole wavelet field, so, signal is after decomposition, and the wavelet coefficient of signal is more than noise
Coefficient, chooses appropriate threshold value, it is possible to eliminate noise and stick signal wavelet coefficient.
So far, the high-frequency I MF components of selection are carried out after soft-threshold processing using DTCWT, you can realize the effective of noise
Remove, while remaining useful signal composition as much as possible.
Hereafter, the wavelet coefficient of each subband can be reconstructed by following two formula, MF points of high-frequency I after denoising is recovered
Amount:
Finally, using EEMD reverse process, the high-frequency I MF components after denoising are entered with reference to remaining low frequency IMF components
Row is added reconstruct, obtains final EEG signals.
It is high-visible also for denoising effect is made in order to verify effectiveness of the invention and accuracy, brain electricity is taken here
The simulated function of signal as signal source, signal hint as shown in figure 4, loading random Gaussian obtain signals and associated noises as shown in figure 5,
Gained signals and associated noises are handled by DTCWT-EEMD Denoising Algorithms, denoising result are obtained as shown in Figure 6.
In order to highlight effectiveness of the invention and advance, here using wavelet transformation (CWT), set empirical mode decomposition
(EEMD), dual-tree complex wavelet transform (DTCWT), the set empirical mode decomposition (CWT-EEMD) based on wavelet transformation and the present invention
Signals and associated noises are handled by the DTCWT-EEMD algorithms of proposition respectively, and are weighed using mean square deviation error and signal to noise ratio as performance
Figureofmerit.
Change the signal to noise ratio of signals and associated noises, and test and average for 100 times, obtain the final impact of performance and compare such as Fig. 7
With shown in Fig. 8.
As can be seen from Figures 7 and 8, for root-mean-square error, DTCWT-EEMD is minimum all the time, and for signal to noise ratio,
DTCWT-EEMD is maximum all the time, so illustrate guarantor of the algorithm proposed by the present invention in noise remove effect and useful signal integrality
In terms of card, current existing Processing Algorithm is superior to, DTCWT-EEMD has more high efficiency and accuracy, can reliablely and stablely applied
During the EEG signals denoising of BCI brain control systems.
Claims (3)
1. a kind of EEG signals noise remove method, it is characterised in that comprise the following steps:
(1) row set empirical mode decomposition is entered to the EEG signals collected using EEMD algorithms, obtained comprising different frequency range letter
The eigencomponent of breath;
(2) the main some components for including high-frequency random noises, are entered using DTCWT algorithms to it in the eigencomponent obtained by choosing
Row is decomposed, and obtains a series of corresponding wavelet coefficients of noise component(s);
(3) wavelet coefficient corresponding to random noise, carries out wavelet coefficient processing, by random noise pair using Soft thresholding
Wavelet coefficient zero setting is answered, and then realizes the effect of noise remove;
(4) wavelet coefficient Jing Guo noise removal process is carried out to the reconstruct of DTCWT algorithm for inversions, intrinsic point after denoising is obtained
Amount, then carry out being added reconstruct with the eigencomponent of remaining representation signal useful information, obtain final denoising EEG signals.
2. EEG signals noise remove method according to claim 1, it is characterised in that the step (1) is specially:It is past
White noise is added in signal, noise mixture is constituted;Noise mixture is carried out EMD decomposition, intrinsic component combination is resolved into;Weight
Multiple above-mentioned steps so that the white noise added every time is differed, obtain multiple eigencomponent combinations, and to multiple eigencomponents
Combination averaging obtains decomposition result.
3. EEG signals noise remove method according to claim 1, it is characterised in that use two in the step (2)
Individual parallel real wavelet transform tree realizes the decomposition and reconstruct of signal, described two parallel real wavelet transform trees
Respectively real part tree and imaginary part tree, have used a different real filter group, have passed through the two wave filter components respectively
The real part coefficient and imaginary part coefficient of dual-tree complex wavelet are not obtained.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710123597.2A CN106951835A (en) | 2017-03-03 | 2017-03-03 | A kind of EEG signals noise remove method |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710123597.2A CN106951835A (en) | 2017-03-03 | 2017-03-03 | A kind of EEG signals noise remove method |
Publications (1)
Publication Number | Publication Date |
---|---|
CN106951835A true CN106951835A (en) | 2017-07-14 |
Family
ID=59467735
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201710123597.2A Pending CN106951835A (en) | 2017-03-03 | 2017-03-03 | A kind of EEG signals noise remove method |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN106951835A (en) |
Cited By (10)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107657868A (en) * | 2017-10-19 | 2018-02-02 | 重庆邮电大学 | A kind of teaching tracking accessory system based on brain wave |
CN107766793A (en) * | 2017-09-20 | 2018-03-06 | 天津大学 | MEMS gyroscope signal denoising processing method based on mixed method |
CN108095722A (en) * | 2018-01-31 | 2018-06-01 | 南京邮电大学 | Improvement EEMD algorithms based on EEG signals |
CN109498000A (en) * | 2018-12-25 | 2019-03-22 | 杭州电子科技大学 | Epileptic attack prediction technique early period based on EMD Yu WPD Fusion Features |
CN109567743A (en) * | 2018-10-24 | 2019-04-05 | 加康康健有限公司 | A kind of signal reconfiguring method based on EMD, device, terminal device and storage medium |
CN111657936A (en) * | 2020-06-16 | 2020-09-15 | 河南大学 | Signal denoising method based on wavelet transformation and total variation regularization |
CN112730712A (en) * | 2020-12-31 | 2021-04-30 | 杭州谱育科技发展有限公司 | Method for improving LC-MS data signal-to-noise ratio |
CN113289250A (en) * | 2021-06-22 | 2021-08-24 | 河北启智教育科技有限公司 | Brain pulse chip |
CN114757236A (en) * | 2022-06-13 | 2022-07-15 | 广东工业大学 | Electroencephalogram signal denoising optimization method and system based on TQWT and SVMD |
CN117312965A (en) * | 2023-11-30 | 2023-12-29 | 国网辽宁省电力有限公司 | Unknown fault self-learning method for GIS high-voltage isolating switch |
-
2017
- 2017-03-03 CN CN201710123597.2A patent/CN106951835A/en active Pending
Cited By (13)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107766793A (en) * | 2017-09-20 | 2018-03-06 | 天津大学 | MEMS gyroscope signal denoising processing method based on mixed method |
CN107657868A (en) * | 2017-10-19 | 2018-02-02 | 重庆邮电大学 | A kind of teaching tracking accessory system based on brain wave |
CN108095722B (en) * | 2018-01-31 | 2021-05-28 | 南京邮电大学 | Improved EEMD algorithm based on electroencephalogram signals |
CN108095722A (en) * | 2018-01-31 | 2018-06-01 | 南京邮电大学 | Improvement EEMD algorithms based on EEG signals |
CN109567743A (en) * | 2018-10-24 | 2019-04-05 | 加康康健有限公司 | A kind of signal reconfiguring method based on EMD, device, terminal device and storage medium |
CN109498000A (en) * | 2018-12-25 | 2019-03-22 | 杭州电子科技大学 | Epileptic attack prediction technique early period based on EMD Yu WPD Fusion Features |
CN111657936A (en) * | 2020-06-16 | 2020-09-15 | 河南大学 | Signal denoising method based on wavelet transformation and total variation regularization |
CN112730712A (en) * | 2020-12-31 | 2021-04-30 | 杭州谱育科技发展有限公司 | Method for improving LC-MS data signal-to-noise ratio |
CN112730712B (en) * | 2020-12-31 | 2022-06-07 | 杭州谱育科技发展有限公司 | Method for improving LC-MS data signal-to-noise ratio |
CN113289250A (en) * | 2021-06-22 | 2021-08-24 | 河北启智教育科技有限公司 | Brain pulse chip |
CN114757236A (en) * | 2022-06-13 | 2022-07-15 | 广东工业大学 | Electroencephalogram signal denoising optimization method and system based on TQWT and SVMD |
CN114757236B (en) * | 2022-06-13 | 2022-10-11 | 广东工业大学 | Electroencephalogram signal denoising optimization method and system based on TQWT and SVMD |
CN117312965A (en) * | 2023-11-30 | 2023-12-29 | 国网辽宁省电力有限公司 | Unknown fault self-learning method for GIS high-voltage isolating switch |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN106951835A (en) | A kind of EEG signals noise remove method | |
CN106709469B (en) | Automatic sleep staging method based on electroencephalogram and myoelectricity multiple characteristics | |
CN102697493B (en) | Method for rapidly and automatically identifying and removing ocular artifacts in electroencephalogram signal | |
CN102697495B (en) | Second-generation wavelet electromyographic signal noise eliminating method based on ensemble empirical mode decomposition | |
CN105677035B (en) | Mental imagery EEG Signal Denoising method based on EEMD and wavelet threshold | |
CN105342605B (en) | A kind of method of Muscle artifacts in removal EEG signals | |
CN103961092B (en) | EEG Noise Cancellation based on adaptive thresholding | |
Sobahi | Denoising of EMG signals based on wavelet transform | |
CN102835955A (en) | Method of automatically removing ocular artifacts from electroencephalogram signal without setting threshold value | |
WO2006034024A2 (en) | Method for adaptive complex wavelet based filtering of eeg signals | |
CN108309290A (en) | The automatic removal method of Muscle artifacts in single channel EEG signals | |
CN114533086A (en) | Motor imagery electroencephalogram decoding method based on spatial domain characteristic time-frequency transformation | |
CN110151175A (en) | Surface electromyogram signal noise-eliminating method based on CEEMD and improvement wavelet threshold | |
CN108042130A (en) | One kind is based on empirical mode decomposition(EMD)EEG signals preprocess method | |
CN110059564B (en) | Feature extraction method based on power spectral density and cross-correlation entropy spectral density fusion | |
CN114081503A (en) | Method for removing ocular artifacts in electroencephalogram signals | |
Liu et al. | An efficient and robust muscle artifact removal method for few-channel EEG | |
CN112294340A (en) | Method, system, storage medium and computer equipment for quickly and automatically removing myoelectric artifacts | |
Feng et al. | Feature extraction algorithm based on csp and wavelet packet for motor imagery eeg signals | |
CN106859641B (en) | A method of based on nuclear-magnetism artefact in automatic ICA removal EEG signal | |
CN114403897A (en) | Human body fatigue detection method and system based on electroencephalogram signals | |
Feng et al. | A novel SSA-CCA framework for muscle artifact removal from ambulatory EEG | |
CN117462145A (en) | FastICA-based eye artifact self-identification clearing method, device, equipment and medium | |
CN114757236B (en) | Electroencephalogram signal denoising optimization method and system based on TQWT and SVMD | |
Rong-Yi et al. | Blind source separation of multichannel electroencephalogram based on wavelet transform and ICA |
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 | ||
RJ01 | Rejection of invention patent application after publication | ||
RJ01 | Rejection of invention patent application after publication |
Application publication date: 20170714 |