CN105718953A - Single-time P300 detection method based on matrix grey modeling - Google Patents
Single-time P300 detection method based on matrix grey modeling Download PDFInfo
- Publication number
- CN105718953A CN105718953A CN201610045616.XA CN201610045616A CN105718953A CN 105718953 A CN105718953 A CN 105718953A CN 201610045616 A CN201610045616 A CN 201610045616A CN 105718953 A CN105718953 A CN 105718953A
- Authority
- CN
- China
- Prior art keywords
- matrix
- feature
- data
- time
- detection method
- 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
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
- G06F18/2411—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on the proximity to a decision surface, e.g. support vector machines
-
- 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
Landscapes
- Engineering & Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Theoretical Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Artificial Intelligence (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Investigating Or Analysing Biological Materials (AREA)
- Measurement And Recording Of Electrical Phenomena And Electrical Characteristics Of The Living Body (AREA)
Abstract
The invention provides a single-time P300 detection method based on matrix grey modeling, belongs to the field of cognitive neuroscience and relates to a feature extraction and recognition detection method of an event related potential P300, in particular to a single-time P300 detection method based on matrix grey modeling in a grey theory.The method comprises the steps that 1, a original acquired electroencephalogram signal is preprocessed; 2, electrocardiographic lead combinations are selected, namely four electrocardiographic leads with most obvious top occipital region waveform differences are selected as optimal electrodes according to oscillograms of target stimulus and non-target stimulus of training set data; 3, segmented matrix grey modeling is conducted on the data of the four electrocardiographic leads, and model parameters are extracted to serve as feature vectors; 4, a Fisher ratio value method is utilized to perform optimal feature selection, and meanwhile the purpose of decreasing the number of feature vector dimensions is achieved; 5, a support vector machine classifier is utilized to classify feature vectors, and single-time P300 detection recognition is achieved.Experimental data tests show that the method can improve single-time P300 detection recognition rate, and the correct recognition rate can be further improved during less-time superposition.
Description
Technical field
The invention belongs to cognitive neuroscience field, relate to feature extraction and the recognition detection method of a kind of event-related potential N400, be specifically related to a kind of single P300 detection method based on the matrix Motar transport in gray theory.
Background technology
P300 (also referred to as P3b) is the endogenous component that can reflect higher cognitive processing procedure, it is when tested identification " target stimulation ", the positive wave in maximum late period of 300ms it is about in the incubation period that scalp recorded, can by vision, audition and body-sensing Induced by Stimulation.
P300 current potential is widely used in brain-computer interface system (BrainComputerInterface, BCI) and psychology is detected a lie in research.Traditional P300 recognition methods is the method adopting time-domain coherence average, can effectively reduce background random noise by superposed average.The method can identify P300 well, but there is obvious defect: multiple averaging easily makes tested fatigue of feeling, and have ignored the difference between different examination time ERP, it is most important that can not meet the requirement of real-time, be unfavorable for the real world applications of P300.
Researcheres propose the feature extraction of few time/single P300 of accomplished in many ways, such as the recognizer based on wavelet transformation, the recognizer based on independent component analysis, recognizer etc. based on principal component analysis.But, these methods all have some limitations, and the detection accuracy of single-trial extraction P300 is all not as high.
Gray system theory (GreySystemTheory) is taught by Chinese scholar Deng Julong and is first proposed in the world in nineteen eighty-two.It is particularly suitable for the process of signal (such as non-stationary, non-gaussian distribution, nonwhite noise) to atypia rule, compared with some other methods processing data by statistical law and priori rule, gray method desired data amount is little, does not need the obvious advantages such as priori.And be the further expansion to tradition grey modeling towards the Motar transport of matrix, point sequence modeling is extended to face by it, i.e. matrix sequence, not only contains temporal information, and is also the modeling to spatial information simultaneously.Motar transport towards matrix is a kind of space-time analysis method.
The present invention proposes and relates to a kind of single P300 feature extraction based on matrix Motar transport and recognition methods.
Summary of the invention
(1) technical problem to be solved
Low for solving single P300 extracting method accuracy rate in prior art, the defect that practicality is not high, the present invention provides a kind of P300 feature extracting method based on matrix Motar transport, P300 signal is modeled by the method by matrix Motar transport, using model parameter as feature, the effect making P300 Classification and Identification is promoted preferably, stimulates in number of repetition situation reducing, improves the detection discrimination of P300 signal.
(2) technical scheme
The present invention realizes by the following technical solutions.Overall procedure is as shown in Figure 1:
1, the EEG signal of acquired original is carried out pretreatment, first carry out the bandpass filtering of 1-15Hz, then carry out segmentation according to data label;Then, the Epoch after segmentation is carried out artefact rejecting;Finally, post-stimulatory data, as baseline, are carried out Baseline wander by the meansigma methods of the data of 100ms before stimulation;
2, Conduction choice, by observing the training data gathered, draws the oscillogram that target stimulus stimulates with non-targeted, selects the most obvious 4 of different wave shape to lead as optimum electrode;
3, choose 4 pretreated Epoch that lead are carried out segmentation, set up the matrix ash model of every little segment data, and extract the grey actuating quantity of each segmented model as characteristic parameter, all segmented model parameters of each Epoch couple the characteristic vector as this Epoch, as shown in Figure 2;
4, feature selection, being coupled by all characteristic parameters of the Epoch obtained through said method is a characteristic vector, then carries out the calculating of Fisher ratio, is ranked up according to the size of rate value, the sequence number that after being sorted, each feature is corresponding.Optimal characteristics space is selected finally according to feature ordering;
5, svm classifier identification, selects support vector machine as grader, Selection of kernel function Radial basis kernel function, the characteristic vector obtained through above-mentioned steps is input to SVM, it is achieved the discriminator to P300.
(3) beneficial effect
The present invention proposes a kind of single P300 feature extracting method based on matrix Motar transport, and the method can extract the characteristic feature amount of P300 well.By combining the feature selection based on Fisher ratio and support vector machine (SupportVectorMachine, SVM) grader, characteristic vector is realized classification, be effectively increased the single P300 accuracy extracted and identify.Showing through experimental data actual measurement, the method all achieves higher recognition correct rate in the measured data of most of objects, has the advantage that the fast precision of on-line operation speed is high, enables the invention to be suitable to practical application.
Accompanying drawing explanation
Fig. 1 is context of methods general flow chart
Fig. 2 is that Epoch data carry out DTMGM modeling procedure figure
Fig. 3 is the characteristic parameter distribution after target stimulus and non-targeted stimulate lower data segment to carry out matrix Motar transport
The Fig. 4 P300 recognition correct rate when single-trial extraction and few examination time superposition
Specific embodiments
Below in conjunction with the drawings and specific embodiments, the present invention will be further described.
The present invention provides a kind of P300 feature extracting method based on matrix Motar transport, in conjunction with Fisher Ratio Features system of selection and support vector machine (SupportVectorMachine, SVM) for the feature extraction of P300 signal and identification, stimulate corresponding EEG data section extraction model parameter as characteristic vector to each, improve recognition accuracy when single-trial extraction P300, and then character transmission rate and the P300 recognition efficiency in application of detecting a lie of the BCI system based on P300 can be improved.
The key step of the present invention is referring to Fig. 1.The method comprises the following steps:
Step S1: to EEG data through pretreatment, including filtering, artefact is rejected and baseline calibration.First the initial data gathered carries out the bandpass filtering of 1-15Hz, and wave filter selects Butterworth filter;Then carry out segmentation according to data label, choose the data of 0.7s after 0.1s before stimulating extremely stimulates, be segmented into the Epoch of 0.8s;Then, the Epoch after segmentation being carried out artefact rejecting, by maximum amplitude and minimum amplitude difference, the data segment more than 140 μ v is labeled as noise artefact, rejects, and is not involved in follow-up data and processes.Finally, post-stimulatory data, as baseline, are carried out Baseline wander by the meansigma methods of the data of 100ms before stimulation;
Step S2: Conduction choice, P300 composition is obvious in occipital region, top, therefore chooses the part in this region and leads and be analyzed, to select the best combination of leading to classify in order to subsequent treatment.By observing the training data gathered, draw the oscillogram of target and non-targeted, and select the most obvious 4 of different wave shape to lead as optimum electrode.
Step S3: pretreated Epoch is carried out k segmentation and carries out matrix Motar transport, and extract the grey actuating quantity of each segmented model as characteristic parameter.All segmented model parameters of each Epoch couple the characteristic vector as this Epoch, as shown in Figure 2;
Step S4: feature selection, is spliced into 2k dimensional feature vector by the characteristic vector of obtain through said method 2 k dimensions, then carries out the calculating of Fisher ratio, be ranked up according to the size of rate value, the sequence number that after being sorted, each feature is corresponding.Optimal characteristics space is selected finally according to feature ordering;
Step S5: Classification and Identification, selects support vector machine as grader, Selection of kernel function Radial basis kernel function, the characteristic vector obtained through above-mentioned steps is input to SVM, it is achieved the discriminator to P300.
Further, in an embodiment of the present invention (sample rate is Fs) step S3 includes step in detail below:
Step S31: each sample includes 4 data leading 0.8s, it is possible to being expressed as the matrix form of 4* (Fs*0.8), spatially leading 4 carries out the matrixing of 2*2, so finally being write as the three dimensional form of 2*2* (Fs*0.8);
Step S32: upper for time Fs*0.8 sample point carries out segmentation, and every n point is divided into a segment, it is possible to be divided intoSection, every one piece of data matrix form is 2*2*n;
Step S33: these 2*2*n data are carried out DTMGM (1,1+2) modeling.Idiographic flow is:
(1) x=(x (1), x (2) ..., x (n)), wherein x (i) represents the matrix of a 2*2, i ∈ 1,2 ..., n}, x(0)For the diagonal matrix transform sequence of matrix sequence x, x(1)=MAGOx(0), then:
x(1)=(x(1)(1),x(1)(2),...,x(1)(n))(1)
Wherein
(2) diagonal transformation matrix sequence ash model (DiagonalTransformationMatrixSequenceGreyModel, DTMGM) is set up:
x(0)(i)+z(1)(i)=diag (a1,a2)(2)
Wherein, z(1)(i)=0.5x(1)(i)+0.5x(1)(i-1), z(1)(i)∈z(1)=(z(1)(2),...,z(1)(n))。
(3) parameter vector p is calculated according to following formuladm:
pdm=[a, a1,a2]T=(BTB)-1yn(3)
Wherein,
(4) by ash actuating quantity M=[a1,a2] as the characteristic parameter of this segment data.
Step S34: to k segmentation respectively according to the given Method Modeling of S33 extracting parameter feature, each Epoch may finally extract k*2 characteristic parameter.
In an embodiment of the present invention, sample rate Fs is 200Hz, when n takes 8, and k=20.The characteristic parameter extracted after context of methods is distributed as shown in Figure 3.
The method fully applies the advantage of the few data modeling of Motar transport, its feature can will be effectively strengthened after P300 segmentation modeling, extraction model parameter, as characteristic vector, adopts Fisher ratio to be input to support vector machine after carrying out feature space dimensionality reduction and carries out Classification and Identification, and discrimination effectively improves.
Characteristic vector being put in SVM classifier, with RBF for kernel function, adopt leaving-one method to carry out cross validation, all tested average accuracy can reach 91.43%, shows the superiority of method given by the present invention.When, after few examination time superposition, Classification and Identification rate can further improve, as shown in Figure 4.
Claims (2)
1. the single P300 detection method based on matrix Motar transport, it is characterised in that said method comprising the steps of:
Step S1: to EEG data through pretreatment, including filtering, segmentation, artefact is rejected and baseline calibration;
Step S2: by observing the training data gathered, draws the oscillogram of target and non-targeted, and selects the most obvious 4 of different wave shape to lead as optimum electrode;
Step S3: pretreated each sample is carried out k segmentation and carries out matrix Motar transport, and extract the grey actuating quantity of each segmented model as characteristic parameter;
Step S4: the characteristic vector of 2 the k dimensions obtained through said method being coupled is 2k dimensional feature vector, then carries out the calculating of Fisher ratio, is ranked up according to the size of rate value, the sequence number that after being sorted, each feature is corresponding.Optimal characteristics space is selected finally according to feature ordering;
Step S5: Classification and Identification, selects support vector machine as grader, Selection of kernel function Radial basis kernel function, the characteristic vector obtained through above-mentioned steps is input to SVM, it is achieved the discriminator to single P300.
2. based on the single P300 detection method based on matrix Motar transport described in claim 1, it is characterised in that described step 3 comprises the following steps:
Step S31: each examination time sample includes 4 data leading 0.8s, the matrix form (sample rate is Fs) of 4* (Fs*0.8) can be expressed as, spatially 4 are led and carry out the matrixing of 2*2, so finally being write as the three dimensional form of 2*2* (Fs*0.8);
Step S32: upper for time Fs*0.8 sample point carries out segmentation, and every n point is divided into a segment, it is possible to be divided intoSection, every one piece of data matrix form is 2*2*n;
Step S33: these 2*2*n data are carried out DTMGM (1,1+2) modeling extraction model ash actuating quantity parameter;
Step S34: to k segmentation respectively according to the given Method Modeling of S33 extracting parameter feature, each Epoch may finally extract k*2 characteristic parameter.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201610045616.XA CN105718953B (en) | 2016-01-25 | 2016-01-25 | A kind of single P300 detection method based on matrix Motar transport |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201610045616.XA CN105718953B (en) | 2016-01-25 | 2016-01-25 | A kind of single P300 detection method based on matrix Motar transport |
Publications (2)
Publication Number | Publication Date |
---|---|
CN105718953A true CN105718953A (en) | 2016-06-29 |
CN105718953B CN105718953B (en) | 2019-07-05 |
Family
ID=56154004
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201610045616.XA Active CN105718953B (en) | 2016-01-25 | 2016-01-25 | A kind of single P300 detection method based on matrix Motar transport |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN105718953B (en) |
Cited By (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN106713787A (en) * | 2016-11-02 | 2017-05-24 | 天津大学 | Evaluation method for watching comfort level caused by rolling subtitles of different speed based on EEG |
CN107180227A (en) * | 2017-05-02 | 2017-09-19 | 西北工业大学 | Cloud prediction and forecasting procedure based on DTMGM+ gray models |
CN110192875A (en) * | 2019-05-24 | 2019-09-03 | 中南民族大学 | Based on the lie detecting method for more leading EEG signals shape factor |
TWI734222B (en) * | 2019-10-17 | 2021-07-21 | 孫光天 | A system that combines brain waves and artificial intelligence to diagnose dyslexia |
Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20090156956A1 (en) * | 2007-11-13 | 2009-06-18 | Michael Milgramm | Method of Determining Whether a Test Subject is a Specific Individual |
CN102521206A (en) * | 2011-12-16 | 2012-06-27 | 天津大学 | Lead optimization method for SVM-RFE (support vector machine-recursive feature elimination) based on ensemble learning thought |
CN103150023A (en) * | 2013-04-01 | 2013-06-12 | 北京理工大学 | System and method for cursor control based on brain-computer interface |
US8699767B1 (en) * | 2006-10-06 | 2014-04-15 | Hrl Laboratories, Llc | System for optimal rapid serial visual presentation (RSVP) from user-specific neural brain signals |
CN104899573A (en) * | 2015-06-18 | 2015-09-09 | 福州大学 | P300 feature extraction method based on wavelet transformation and Fisher criterion |
CN105068644A (en) * | 2015-07-24 | 2015-11-18 | 山东大学 | Method for detecting P300 electroencephalogram based on convolutional neural network |
-
2016
- 2016-01-25 CN CN201610045616.XA patent/CN105718953B/en active Active
Patent Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US8699767B1 (en) * | 2006-10-06 | 2014-04-15 | Hrl Laboratories, Llc | System for optimal rapid serial visual presentation (RSVP) from user-specific neural brain signals |
US20090156956A1 (en) * | 2007-11-13 | 2009-06-18 | Michael Milgramm | Method of Determining Whether a Test Subject is a Specific Individual |
CN102521206A (en) * | 2011-12-16 | 2012-06-27 | 天津大学 | Lead optimization method for SVM-RFE (support vector machine-recursive feature elimination) based on ensemble learning thought |
CN103150023A (en) * | 2013-04-01 | 2013-06-12 | 北京理工大学 | System and method for cursor control based on brain-computer interface |
CN104899573A (en) * | 2015-06-18 | 2015-09-09 | 福州大学 | P300 feature extraction method based on wavelet transformation and Fisher criterion |
CN105068644A (en) * | 2015-07-24 | 2015-11-18 | 山东大学 | Method for detecting P300 electroencephalogram based on convolutional neural network |
Non-Patent Citations (2)
Title |
---|
SONGYUN XIE等: "Single Channel Single Trial P300 Detection Using Extreme Learning Machine Compared with BPNN and SVM", 《2014 INTERNATIONAL JOINT CONFERENCE ON NEURAL NETWORK》 * |
黄璐等: "单导联脑电信号P300分类研究", 《仪器仪表学报》 * |
Cited By (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN106713787A (en) * | 2016-11-02 | 2017-05-24 | 天津大学 | Evaluation method for watching comfort level caused by rolling subtitles of different speed based on EEG |
CN107180227A (en) * | 2017-05-02 | 2017-09-19 | 西北工业大学 | Cloud prediction and forecasting procedure based on DTMGM+ gray models |
CN107180227B (en) * | 2017-05-02 | 2021-02-19 | 西北工业大学 | Cloud prediction and forecast method based on DTMMG + gray model |
CN110192875A (en) * | 2019-05-24 | 2019-09-03 | 中南民族大学 | Based on the lie detecting method for more leading EEG signals shape factor |
TWI734222B (en) * | 2019-10-17 | 2021-07-21 | 孫光天 | A system that combines brain waves and artificial intelligence to diagnose dyslexia |
Also Published As
Publication number | Publication date |
---|---|
CN105718953B (en) | 2019-07-05 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Haberman et al. | The visual system discounts emotional deviants when extracting average expression | |
CN108960182B (en) | P300 event related potential classification identification method based on deep learning | |
CN103714281B (en) | A kind of personal identification method based on electrocardiosignal | |
CN105718953A (en) | Single-time P300 detection method based on matrix grey modeling | |
CN109344816A (en) | A method of based on brain electricity real-time detection face action | |
CN110037693A (en) | A kind of mood classification method based on facial expression and EEG | |
EP3449409B1 (en) | Biometric method and device for identifying a person through an electrocardiogram (ecg) waveform | |
CN108256579A (en) | A kind of multi-modal sense of national identity quantization measuring method based on priori | |
CN107480716A (en) | Method and system for identifying saccade signal by combining EOG and video | |
Guan et al. | Low-dimensional denoising embedding transformer for ECG classification | |
CN111616702A (en) | Lie detection analysis system based on cognitive load enhancement | |
CN108920699B (en) | Target identification feedback system and method based on N2pc | |
CN111067513A (en) | Sleep quality detection key brain area judgment method based on characteristic weight self-learning | |
CN111053552A (en) | QRS wave detection method based on deep learning | |
CN104793743B (en) | A kind of virtual social system and its control method | |
Wang et al. | Warp-averaging event-related potentials | |
Buvaneswari et al. | A review of EEG based human facial expression recognition systems in cognitive sciences | |
Huang et al. | Automatic artifact removal in EEG using independent component analysis and one-class classification strategy | |
CN111281382A (en) | Feature extraction and classification method based on electroencephalogram signals | |
CN113057654B (en) | Memory load detection and extraction system and method based on frequency coupling neural network model | |
CN105787459A (en) | ERP signal classification method based on optimal score sparse determination | |
Visconti et al. | Automatic recognition of error potentials in a P300-based brain-computer interface | |
Yu et al. | Improving distinguishability of photoplethysmography in emotion recognition using deep convolutional generative adversarial networks | |
Szymanska et al. | Source location as a feature for the classification of multi-sensor extracellular action potentials | |
Ahmed et al. | Emotion recognition based on correlation between left and right frontal EEG assymetry |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |