CN109214325A - A kind of movement related potential detection method based on space filtering and stencil matching - Google Patents

A kind of movement related potential detection method based on space filtering and stencil matching Download PDF

Info

Publication number
CN109214325A
CN109214325A CN201810988364.3A CN201810988364A CN109214325A CN 109214325 A CN109214325 A CN 109214325A CN 201810988364 A CN201810988364 A CN 201810988364A CN 109214325 A CN109214325 A CN 109214325A
Authority
CN
China
Prior art keywords
data
filtering
filter
template
space
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.)
Granted
Application number
CN201810988364.3A
Other languages
Chinese (zh)
Other versions
CN109214325B (en
Inventor
明东
王坤
许敏鹏
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Tianjin University
Original Assignee
Tianjin University
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Tianjin University filed Critical Tianjin University
Priority to CN201810988364.3A priority Critical patent/CN109214325B/en
Priority to PCT/CN2018/125926 priority patent/WO2020042511A1/en
Publication of CN109214325A publication Critical patent/CN109214325A/en
Application granted granted Critical
Publication of CN109214325B publication Critical patent/CN109214325B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

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
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
    • G06F18/213—Feature extraction, e.g. by transforming the feature space; Summarisation; Mappings, e.g. subspace methods
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
    • G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
    • G06V10/74—Image or video pattern matching; Proximity measures in feature spaces
    • G06V10/75—Organisation of the matching processes, e.g. simultaneous or sequential comparisons of image or video features; Coarse-fine approaches, e.g. multi-scale approaches; using context analysis; Selection of dictionaries
    • G06V10/758—Involving statistics of pixels or of feature values, e.g. histogram matching
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
    • G06F2218/08—Feature extraction
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
    • G06F2218/12—Classification; Matching

Abstract

The movement related potential detection method based on space filtering and stencil matching that the invention discloses a kind of, it the described method comprises the following steps: constructing task Related Component spatial filter using pretreated eeg data, eeg data is the first filtering data after the filtering of task Related Component spatial filter；Spatial mode filter is constructed using pretreated eeg data, eeg data is the second filtering data after spatial mode filter filters；Two template datas are constructed respectively according to the first filtering data and the second filtering data, and test data carries out the matching analysis with two template datas after task Related Component spatial filter, discriminant space mode filter respectively respectively, finally carries out Decision Classfication.The present invention combines the method for discriminant space pattern analysis using the analysis of task based access control Related Component, it can be achieved that movement related potential accurately identifies for the first time.

Description

A kind of movement related potential detection method based on space filtering and stencil matching
Technical field
The present invention relates to movement related potential detection field more particularly to a kind of fortune based on space filtering and stencil matching Dynamic related potential detection method.
Background technique
Motor cortex related potential (Movement-related cortical potentials, MRCPs) is that human body exists Generated low frequency current potential when being called when executing indicative or activity in brain to the relevant cognitive resources of movement.Pass through Detection MPCPs can be used for controlling brain-computer interface (Brain-computer interface, BCI) output.BCI be one will in Pivot nervous system activity is converted into the system manually exported, it can substitute, repair, enhance, supplement or improve maincenter The normal output of nervous system, so as to improve the reciprocation between central nervous system and internal and external environment.It is detected based on MRCPs BCI system can be used for a variety of occasions such as rehabilitation training, intelligent artificial limb and mechanical exoskeleton control, thus obtained researchers Extensive concern.
MRCPs can be regarded as event related potential relevant to movement.Since the signal-to-noise ratio of EEG signals is low, single examination time MRCP wave character be difficult effectively to extract.Common method is superposed average, improves signal-to-noise ratio.Scalp EEG signals can be seen Work is superposition of the signal in scalp of source generation different in brain.The main thought of space filtering is: by original brain electricity Each road signal distributes weight, source signal ingredient interested, cancelling noise signal is extracted, to improve the signal-to-noise ratio of characteristic signal. The main thought of template matching is to find matching degree highest by matching characteristic signal with the template signal of different mode Template as recognition mode.
Movement related potential is Motion Evoked Potential, and amplitude is small, and signal-to-noise ratio is low, and common method is superposed average, is improved Signal-to-noise ratio.But the method for superposed average need user that same task is performed a plurality of times after output one as a result, time cost compared with It is high.The motor task correlation EEG wave character of single examination time is difficult effectively to extract.
Summary of the invention
The present invention provides a kind of the movement related potential detection method based on space filtering and stencil matching, purport of the present invention EEG signals are improved in the spatial filtering method using a kind of analysis of task based access control Related Component and discrimination model spatial analysis to believe It makes an uproar and compares, be compared with the traditional method, movement related potential detection efficiency can be improved, further research can open up for the development of BCI New developing direction is expected to obtain considerable Social benefit and economic benefit, described below:
A kind of movement related potential detection method based on space filtering and stencil matching, the method includes following steps It is rapid:
Task Related Component spatial filter is constructed using pretreated eeg data, eeg data is related by task It is the first filtering data after component space filter filtering；
Spatial mode filter is constructed using pretreated eeg data, eeg data is filtered by spatial mode filter It is the second filtering data after wave；
Two template datas are constructed respectively according to the first filtering data and the second filtering data, and test data is respectively by appointing The matching analysis is carried out respectively with two template datas after business Related Component spatial filter, spatial mode filter, is finally carried out Decision Classfication.
It is described to construct task Related Component spatial filter using pretreated eeg data specifically:
Training setTask Related Component analysis filter is obtained by calculation
N subfilter is obtained；Wherein NcIndicate the dimension of filter, NtIndicate interception Signal length；
To subfilter dimensionality reductionSubfilter combination after dimensionality reduction is become
To the template signal and test data of training sample after subspace filters, obtainAnd
Wherein, N 'cFor the dimension of dimensionality reduction postfilter,For the transposition of template signal,For the i-th class TRCA after dimensionality reduction Filter.
It is described to construct spatial mode filter using pretreated eeg data specifically:
DSP filter is calculated to every two classes training set, then is obtainedA subfilter, to subfilter dimensionality reductionSubfilter combination after dimensionality reduction is become
To the template signal of training sampleAnd test data carries out space filtering, obtainsAnd
Wherein,Subfilter after passing through the dimensionality reduction obtained after calculating for the i-th class training data and jth class data； It is the filter group for including all subfilters.
Wherein, the test data is respectively after task Related Component spatial filter, spatial mode filter with two A template data carries out the process of the matching analysis respectively are as follows:
CCA analysis is carried out to the test data after TRCA space filtering and each template and calculates skin in new projector space The process of Ademilson related coefficient；
Specific formula for calculation is as follows:
Calculate the Pearson correlation coefficients of test data and each template after DSP space filtering:
Calculate the Euclidean distance of test data and each template after DSP space filtering:
CCA analysis is carried out to the test data after DSP space filtering and each template and calculates skin in new projector space Ademilson related coefficient:
Test data and the template matching of every a kind of training data have obtained feature vector [ρ after calculating as a result,i1,ρi2, ρi3,ρi4,ρi5,ρi6,ρi7,ρi8]T, ρi∈R8×1；
Wherein, CCA (*) indicates CCA analysis, and corr (*) indicates Pearson correlation coefficients, Ai,BiFor linear projection matrix, ρi1For first characteristic element of the i-th category feature vector, ρi2For second characteristic element of the i-th category feature vector, ρi3It is i-th The third characteristic element of category feature vector, ρi4For the 4th characteristic element of the i-th category feature vector, ρi5For the i-th category feature to 5th characteristic element of amount, ρi6For the 6th characteristic element of the i-th category feature vector, ρi7It is the 7th of the i-th category feature vector the A characteristic element, ρi8For the 8th characteristic element of the i-th category feature vector.
Further, the process of the Decision Classfication is the process that characteristic value compares, specifically:
More matched by comparing the size discriminating test sample of characteristic value with which kind of training sample, weight coefficient is added ω=[ω1,ω2,ω3,ω4,ω5,ω6,ω7,ω8], ω ∈ R1×8, Optimized Matching result:
Choose maximum ω ρi, then the recognition result of test data is I type games associative mode.
The beneficial effect of the technical scheme provided by the present invention is that:
1, the present invention for the first time using task based access control Related Component analysis combine discriminant space pattern analysis method, it can be achieved that Movement related potential accurately identifies；
2, method proposed by the present invention can be used for the brain-computer interface system detected based on motion intention, utilize this method energy Brain-computer interface technology is further improved, the technology is promoted to convert to application achievements.
Detailed description of the invention
Fig. 1 is a kind of flow chart of movement related potential detection method based on space filtering and stencil matching.
Specific embodiment
To make the object, technical solutions and advantages of the present invention clearer, embodiment of the present invention is made below further Ground detailed description.
Spatial filtering method involved in the embodiment of the present invention includes: task Related Component analysis (Task-related Component analysis, TRCA) and discriminant space pattern analysis (Discriminative spatial pattern, DSP).TRCA is intended to extract task Related Component from the multiple time series of linear weighted function.Its main thought is to maximize together Covariance or correlation between the examination time of one generic task difference.The purpose of DSP is to make feature difference between different mode most Bigization.This method is expansion of the Fisher linear discriminant analysis thought in terms of spatial analysis, and deviation projects most between emphasizing class Deviation projects and obtains spatial filter in the smallest situation in big and class.The template matching method of the embodiment of the present invention is related to Pierre Gloomy related coefficient, Euclidean distance and canonical correlation analysis (Canonical correlation analysis, CCA) method and spy Value indicative comparative approach.
Embodiment 1
The movement related potential detection method based on space filtering and stencil matching that the embodiment of the invention provides a kind of.Fortune Dynamic correlation electrocortical potential (MRCPs) be people generated when imagining or executing body kinematics rich in motion information, have stringent lock When and locking phase EEG signals mode.Because it has in terms of disclosing human motion neuromechanism, instructing Important researching value receives the extensive concern of researchers.The embodiment of the present invention devises the movement based on space filtering Related potential detection method, can be improved recognition correct rate.
Its techniqueflow is: constructing task Related Component spatial filter, data 1 using pretreated eeg data 1 It is data 2 after the filtering of task Related Component spatial filter, constructs discriminant space mould using pretreated eeg data 1 Formula filter, data 1 are data 3 after the filtering of discriminant space mode filter, construct mould according to data 2 and data 3 respectively Plate data, test data respectively after task Related Component spatial filter, discriminant space mode filter with two templates Data are into row the matching analysis respectively (carrying out the process of coherent analysis), and finally carrying out Decision Classfication, (i.e. characteristic value compares Process).
In conclusion TRCA, DSP spatial filtering method combination template matching method are used to transport by the embodiment of the present invention for the first time The detection of dynamic related potential is beneficial to improve recognition efficiency, moves towards practical tool to the BCI based on movement related potential detection It is significant.
Embodiment 2
The scheme in embodiment 1 is further introduced below with reference to Fig. 1 and specific calculation formula, is detailed in It is described below:
Fig. 1 is the flow chart that the embodiment of the present invention calculates, and can be used for moving related potential detection.Assuming that Xi={ x(m)}iFor I-th type games mode related brain electricity training data set (i=1,2 ..., n；M=1,2 ..., M), wherein every a kind of training There is M sample in data acquisition systemX (m) indicates m-th of sample,For test sample, wherein NcExpression is adopted Collect the port number of brain electricity, NtIndicate intercept signal length.All samples of training set are averaged to obtain the template letter of training sample NumberAs shown in formula (1).
Data processing mainly includes data prediction, building spatial filter and divides data space filtering, correlation Four steps of feature and template matching are extracted in analysis.
One, data prediction
EEG signals are pre-processed first.The sample frequency of usual EEG signals is 1000Hz or higher, to guarantee It is saved under the premise of signal quality and calculates cost, be downsampled to 200Hz first.0.5 is done to signal using Chebyshev filter The filtering of~45Hz.And zero-mean processing is carried out all in time scale to training set and test set.
Two, space filtering
In embodiments of the present invention, space filtering is divided into two steps, be first TRCA filtering, extract task correlation at Point, second step is DSP filtering, maximizes the difference of two class data.
1, TRCA spatial filter is constructed
TRCA algorithm belongs to a kind of spatial filter, it is therefore an objective to maximize different in the same task try by filter V The sum of covariance between secondary, shown in the calculation method of the sum of the covariance such as formula (2),Wherein NcIndicate acquisition The port number of brain electricity, NtIndicate intercept signal length, NsIndicate training set number of samples, m1With m2Indicate the number of sample, k1With k2Indicate the number of lead,Indicate m1A sample lead k1Signal, x is sample set.
Wherein, Cov is covariance；Respectively k1, k2The corresponding filter of lead；V is electric-wave filter matrix；S is examination Covariance matrix between secondary.
Constraint condition such as formula (3) is arranged in the solution limited in order to obtain:
Wherein, covariance matrix of the Q between lead.
Therefore final majorized function such as formula (4):
Wherein, optimal solution V is matrix Q-1The feature vector of S.
For the embodiment of the present invention, training setTRCA filter is obtained by calculation N subfilter is obtained.
It, can be to subfilter dimensionality reduction to obtain optimal filterThe dimension N of selectionc' can rule of thumb really It is fixed, it can also be adjusted according to algorithm optimization.Subfilter combination after dimensionality reduction is become
To the template signal and test data of training sample after subspace filters, obtainAnd
Wherein, Nc' be dimensionality reduction postfilter dimension,For the transposition of template signal,For the i-th class TRCA after dimensionality reduction Filter.
2, DSP spatial filter is constructed
DSP is a kind of spatial filtering method, it is therefore an objective to maximize the feature difference between different mode.This method is Expansion of the Fisher linear discriminant analysis thought in terms of spatial analysis, deviation projection is maximum between emphasizing class throws with deviation in class Spatial filter is obtained in the smallest situation of shadow.Wherein class scatter matrix SbAs shown in formula (6),WithTo average to obtain the template signal of two class training samples to the i-th class and all samples of jth class training set.
Scatter Matrix S in classwAs shown in formula (7):
Wherein,For the i-th class training set,For jth class training set.
Therefore final objective function such as formula (8):
Wherein, optimal solution U is matrixFeature vector, u be the corresponding filter of each lead.
For the embodiment of the present invention, two class training setsWithBy obtaining DSP filtering after calculating Device(i=1,2 ..., n；J=1,2 ..., n and j ≠ i).
DSP filter is calculated to every two classes training set, then is obtainedA subfilter.It, can to obtain optimal filter To subfilter dimensionality reductionThe dimension N of selectionc' can be empirically determined, it can also be adjusted according to algorithm optimization.To drop Subfilter combination after dimension becomes
To the template signal of training sampleAnd test data carries out space filtering, obtainsAnd
Wherein,Subfilter after passing through the dimensionality reduction obtained after calculating for the i-th class training data and jth class data； It is the filter group for including all subfilters.
Three, correlation analysis
The method for the correlation analysis being used in the present invention is Pearson correlation coefficients, Euclidean distance and canonical correlation point (Canonical correlation analysis, the CCA) method of analysis.
Pearson correlation coefficients is defined as: the pearson correlation property coefficient of two continuous variables X, Y are equal between them Covariance cov (X, Y) divided by each standard deviation product σXσY.The value range of coefficient is [- 1,1], close to 0 change Amount is referred to as non-correlation, and being referred to as close to 1 or -1 has strong correlation.Being used herein corr (*) indicates Pearson's phase Relationship number.
Euclidean distance is the abbreviation of euclidean metric, is referred in the actual distance in m-dimensional space between two points, in this hair In bright embodiment, test sample is calculated at a distance from two classes training template, it is believed that distance is closer, and correlation is stronger.Herein Dist (*) indicates Euclidean distance.
The basic principle of CCA algorithm is: in order to hold the correlativity between two groups of indexs on the whole, respectively at two groups It extracts representational two generalized variables A and B (linear combination of each variable in respectively two set of variables), utilizes in variable Correlativity between the two generalized variables is come the overall relevancy that reflects between two groups of indexs.It can be incited somebody to action using CCA algorithm Data projection after space filtering to new space and calculates correlation.Being used herein CCA (*) indicates CCA analysis.
The following are the detailed descriptions for calculating feature.
The Pearson correlation coefficients of test data and each template after TRCA space filtering are calculated first:
Calculate the Euclidean distance of test data and each template after TRCA space filtering:
CCA analysis is carried out to the test data after TRCA space filtering and each template and calculates skin in new projector space Ademilson related coefficient:
Calculate the Pearson correlation coefficients of test data and each template after DSP space filtering:
Calculate the Euclidean distance of test data and each template after DSP space filtering:
CCA analysis is carried out to the test data after DSP space filtering and each template and calculates skin in new projector space Ademilson related coefficient:
Test data and the template matching of every a kind of training data have obtained feature vector ρ after calculating as a result,i=[ρi1, ρi2,ρi3,ρi4,ρi5,ρi6,ρi7,ρi8]T, ρi∈R8×1。
Wherein, CCA (*) indicates CCA analysis, and corr (*) indicates Pearson correlation coefficients, Ai,BiFor linear projection matrix, ρi1For first characteristic element of the i-th category feature vector, ρi2For second characteristic element of the i-th category feature vector, ρi3It is i-th The third characteristic element of category feature vector, ρi4For the 4th characteristic element of the i-th category feature vector, ρi5For the i-th category feature to 5th characteristic element of amount, ρi6For the 6th characteristic element of the i-th category feature vector, ρi7It is the 7th of the i-th category feature vector the A characteristic element, ρi8For the 8th characteristic element of the i-th category feature vector.
Four, characteristic value compares
As it was noted above, the characteristic present extracted in the embodiment of the present invention is correlation, therefore the bigger expression of characteristic value It is a kind of more similar to certain.Which kind of more matched by comparing the size discriminating test sample of characteristic value with training sample.At this In, weight coefficient ω=[ω is added1,ω2,ω3,ω4,ω5,ω6,ω7,ω8], ω ∈ R1×8, Optimized Matching result.
Choose maximum ω ρi, then the recognition result of test data is I type games associative mode.Weight coefficient ω can root According to empirically determined, can also be adjusted according to algorithm optimization.
In conclusion being compared with the traditional method, movement related potential detection efficiency is can be improved in this method.In addition the present invention mentions Detection method out further studies the available perfect brain-computer interface system based on motion intention detection, for deformity People, special population auxiliary output carry out information exchange with the external world, and in fields such as electronic entertainment, Industry Controls, being expected to acquisition can The Social benefit and economic benefit of sight.
It will be appreciated by those skilled in the art that attached drawing is the schematic diagram of a preferred embodiment, the embodiments of the present invention Serial number is for illustration only, does not represent the advantages or disadvantages of the embodiments.
The foregoing is merely presently preferred embodiments of the present invention, is not intended to limit the invention, it is all in spirit of the invention and Within principle, any modification, equivalent replacement, improvement and so on be should all be included in the protection scope of the present invention.

Claims (5)

1. a kind of movement related potential detection method based on space filtering and stencil matching, which is characterized in that the method packet Include following steps:
Task Related Component spatial filter is constructed using pretreated eeg data, eeg data passes through task Related Component It is the first filtering data after spatial filter filtering；
Spatial mode filter is constructed using pretreated eeg data, eeg data is filtered by discriminant space mode filter It is the second filtering data after wave；
Two template datas are constructed respectively according to the first filtering data and the second filtering data, and test data passes through task phase respectively The matching analysis is carried out respectively with two template datas after pass component space filter, discriminant space mode filter, is finally carried out Decision Classfication.
2. a kind of movement related potential detection method based on space filtering and stencil matching according to claim 1, It is characterized in that, it is described to construct task Related Component spatial filter using pretreated eeg data specifically:
Training setTask Related Component analysis filter is obtained by calculation；
I=1,2 ..., n subfilter is obtained in n；Wherein NcIndicate the dimension of filter, NtIndicate interception letter Number length；
To subfilter dimensionality reductionSubfilter combination after dimensionality reduction is become
To the template signal and test data of training sample after subspace filters, obtainAnd
Wherein, N 'cFor the dimension of dimensionality reduction postfilter,For the transposition of template signal,For the i-th class TRCA filtering after dimensionality reduction Device.
3. a kind of movement related potential detection method based on space filtering and stencil matching according to claim 2, It is characterized in that, it is described to construct spatial mode filter using pretreated eeg data specifically:
DSP filter is calculated to every two classes training set, then is obtainedA subfilter, to subfilter dimensionality reduction Subfilter combination after dimensionality reduction is become
To the template signal of training sampleAnd test data carries out space filtering, obtainsAnd
Wherein,Subfilter after passing through the dimensionality reduction obtained after calculating for the i-th class training data and jth class data；It is to include The filter group of all subfilters.
4. a kind of movement related potential detection method based on space filtering and stencil matching according to claim 3, It is characterized in that, the test data is respectively after task Related Component spatial filter, discriminant space mode filter with two A template data carries out the process of the matching analysis respectively are as follows: carries out to the test data after TRCA space filtering and each template CCA analysis simultaneously calculates the process of Pearson correlation coefficients in new projector space；
Specific formula for calculation is as follows:
Calculate the Pearson correlation coefficients of test data and each template after DSP space filtering:
Calculate the Euclidean distance of test data and each template after DSP space filtering:
CCA analysis is carried out to the test data after DSP space filtering and each template and calculates Pearson in new projector space Related coefficient:
Test data and the template matching of every a kind of training data have obtained feature vector [ρ after calculating as a result,i1, ρi2, ρi3, ρi4, ρi5, ρi6, ρi7, ρi8]T, ρi∈R8×1；
Wherein, CCA (*) indicates CCA analysis, and corr (*) indicates Pearson correlation coefficients, Ai, BiFor linear projection matrix, ρi1For First characteristic element of the i-th category feature vector, ρi2For second characteristic element of the i-th category feature vector, ρi3For the i-th class spy Levy the third characteristic element of vector, ρi4For the 4th characteristic element of the i-th category feature vector, ρi5For the i-th category feature vector 5th characteristic element, ρi6For the 6th characteristic element of the i-th category feature vector, ρi7For the 7th spy of the i-th category feature vector Levy element, ρi8For the 8th characteristic element of the i-th category feature vector.
5. a kind of movement related potential detection method based on space filtering and stencil matching according to claim 4, It being characterized in that, the process of the Decision Classfication is the process that characteristic value compares, specifically:
More matched by comparing the size discriminating test sample of characteristic value with which kind of training sample, addition weight coefficient ω= [ω1, ω2, ω3, ω4, ω5, ω6, ω7, ω8], ω ∈ R1×8, Optimized Matching result:
Choose maximum ω ρi, then the recognition result of test data is I type games associative mode.
CN201810988364.3A 2018-08-28 2018-08-28 Motion-related potential detection method based on spatial filtering and template matching Active CN109214325B (en)

Priority Applications (2)

Application Number Priority Date Filing Date Title
CN201810988364.3A CN109214325B (en) 2018-08-28 2018-08-28 Motion-related potential detection method based on spatial filtering and template matching
PCT/CN2018/125926 WO2020042511A1 (en) 2018-08-28 2018-12-30 Motion potential brain-machine interface encoding and decoding method based on spatial filtering and template matching

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201810988364.3A CN109214325B (en) 2018-08-28 2018-08-28 Motion-related potential detection method based on spatial filtering and template matching

Publications (2)

Publication Number Publication Date
CN109214325A true CN109214325A (en) 2019-01-15
CN109214325B CN109214325B (en) 2022-04-29

Family

ID=64986162

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201810988364.3A Active CN109214325B (en) 2018-08-28 2018-08-28 Motion-related potential detection method based on spatial filtering and template matching

Country Status (1)

Country Link
CN (1) CN109214325B (en)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112861629A (en) * 2021-01-07 2021-05-28 天津大学 Multi-window distinguishing typical pattern matching method and brain-computer interface application

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101068498A (en) * 2004-10-04 2007-11-07 旗帜健康公司 Methodologies linking patterns from multi-modality datasets
CN101433460A (en) * 2008-07-25 2009-05-20 天津大学 Spatial filtering method of lower limb imaginary action potential
CN105705093A (en) * 2013-10-07 2016-06-22 Mc10股份有限公司 Conformal sensor systems for sensing and analysis
CN105824418A (en) * 2016-03-17 2016-08-03 天津大学 Brain-computer interface communication system based on asymmetric visual evoked potential

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101068498A (en) * 2004-10-04 2007-11-07 旗帜健康公司 Methodologies linking patterns from multi-modality datasets
CN101433460A (en) * 2008-07-25 2009-05-20 天津大学 Spatial filtering method of lower limb imaginary action potential
CN105705093A (en) * 2013-10-07 2016-06-22 Mc10股份有限公司 Conformal sensor systems for sensing and analysis
CN105824418A (en) * 2016-03-17 2016-08-03 天津大学 Brain-computer interface communication system based on asymmetric visual evoked potential

Non-Patent Citations (4)

* Cited by examiner, † Cited by third party
Title
MASAKI NAKANISHI ET AL: "Enhancing Detection of SSVEPs for a High-Speed Brain Speller Using Task-Related Component Analysis", 《IEEE TRANSACTIONS ON BIOMEDICAL ENGINEERING》 *
MINPENG XU ET AL: "A Brain–Computer Interface Based on Miniature-Event-Related Potentials Induced by Very Small Lateral Visual Stimuli", 《IEEE TRANSACTIONS ON BIOMEDICAL ENGINEERING》 *
江京 等: "整合贝叶斯动态停止策略对SSVEP-BCIs的性能提升研究", 《仪器仪表学报》 *
王春慧 等: "基于动态自适应策略的SSVEP快速目标选择方法", 《清华大学学报（自然科学版）》 *

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112861629A (en) * 2021-01-07 2021-05-28 天津大学 Multi-window distinguishing typical pattern matching method and brain-computer interface application

Also Published As

Publication number Publication date
CN109214325B (en) 2022-04-29

Similar Documents

Publication Publication Date Title
Zerafa et al. To train or not to train? A survey on training of feature extraction methods for SSVEP-based BCIs
CN109271887A (en) A kind of composite space filtering and template matching method for the identification of brain power mode
CN104586387B (en) Method for extracting and fusing time, frequency and space domain multi-parameter electroencephalogram characters
Sun et al. A review of adaptive feature extraction and classification methods for EEG-based brain-computer interfaces
Bascil et al. Spectral feature extraction of EEG signals and pattern recognition during mental tasks of 2-D cursor movements for BCI using SVM and ANN
Zhang et al. Bayesian learning for spatial filtering in an EEG-based brain–computer interface
CN109299751B (en) EMD data enhancement-based SSVEP electroencephalogram classification method of convolutional neural model
Cheng et al. A motor imagery EEG feature extraction method based on energy principal component analysis and deep belief networks
Singh et al. Small sample motor imagery classification using regularized Riemannian features
WO2020042511A1 (en) Motion potential brain-machine interface encoding and decoding method based on spatial filtering and template matching
Chen et al. Multiattention adaptation network for motor imagery recognition
CN111310656A (en) Single motor imagery electroencephalogram signal identification method based on multi-linear principal component analysis
Lv et al. Common spatial pattern and particle swarm optimization for channel selection in BCI
Zhang et al. Transfer learning algorithm design for feature transfer problem in motor imagery brain-computer interface
Zou et al. An inter-subject model to reduce the calibration time for motion imagination-based brain-computer interface
Tang et al. Research on extraction and classification of EEG features for multi-class motor imagery
CN109214325A (en) A kind of movement related potential detection method based on space filtering and stencil matching
Abdulla et al. A review study for electrocardiogram signal classification
Liu et al. Identification of anisomerous motor imagery EEG signals based on complex algorithms
CN107961005A (en) The feature extracting method of few passage brain-computer interface EEG signal
Zhang et al. Low-rank linear dynamical systems for motor imagery EEG
Zhang et al. An ECoG-based binary classification of BCI using optimized extreme learning machine
CN114587384A (en) Motor imagery electroencephalogram signal feature extraction method combining low-rank representation and manifold learning
Li et al. Classification of imaginary movements in ECoG
CN114358090A (en) Motor imagery electroencephalogram signal classification method based on PSD and CSP

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
GR01 Patent grant
GR01 Patent grant