CN109620215A - A kind of extracting method of brain electrical feature - Google Patents
A kind of extracting method of brain electrical feature Download PDFInfo
- Publication number
- CN109620215A CN109620215A CN201811533519.0A CN201811533519A CN109620215A CN 109620215 A CN109620215 A CN 109620215A CN 201811533519 A CN201811533519 A CN 201811533519A CN 109620215 A CN109620215 A CN 109620215A
- Authority
- CN
- China
- Prior art keywords
- feature
- brain electrical
- wave
- extracting method
- energy ratio
- 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
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/24—Detecting, measuring or recording bioelectric or biomagnetic signals of the body or parts thereof
- A61B5/316—Modalities, i.e. specific diagnostic methods
- A61B5/369—Electroencephalography [EEG]
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/24—Detecting, measuring or recording bioelectric or biomagnetic signals of the body or parts thereof
- A61B5/316—Modalities, i.e. specific diagnostic methods
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/72—Signal processing specially adapted for physiological signals or for diagnostic purposes
- A61B5/7235—Details of waveform analysis
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/72—Signal processing specially adapted for physiological signals or for diagnostic purposes
- A61B5/7235—Details of waveform analysis
- A61B5/7264—Classification of physiological signals or data, e.g. using neural networks, statistical classifiers, expert systems or fuzzy systems
Landscapes
- Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Surgery (AREA)
- Public Health (AREA)
- Pathology (AREA)
- Veterinary Medicine (AREA)
- Biomedical Technology (AREA)
- Heart & Thoracic Surgery (AREA)
- Medical Informatics (AREA)
- Molecular Biology (AREA)
- Biophysics (AREA)
- Animal Behavior & Ethology (AREA)
- General Health & Medical Sciences (AREA)
- Psychiatry (AREA)
- Artificial Intelligence (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Physiology (AREA)
- Signal Processing (AREA)
- Psychology (AREA)
- Evolutionary Computation (AREA)
- Fuzzy Systems (AREA)
- Mathematical Physics (AREA)
- Measurement And Recording Of Electrical Phenomena And Electrical Characteristics Of The Living Body (AREA)
Abstract
A kind of extracting method of brain electrical feature, the initial data of brain wave is got using brain electrical detection device, carries out signal processing to initial data, extracted feature is trained, and obtains the feature that temporal signatures are strong, frequency domain character and temporal signatures are stable.On the one hand, it solves in real-time processing, the EEG signals strong for temporal signatures extract difficult problem, on the other hand, the problem of the stability of feature is effectively strengthened from frequency domain and time domain angle, improves waveform severe jamming.
Description
Technical field
The invention belongs to test and analyze technical field more particularly to a kind of brain wave to extract creation and its extraction side of model
Method.
Background technique
With the development of science and technology, the identification of image, voice and text has very big progress.Along with the curiosity of people
The heart, the research of brain wave also cause always the concern of many scholars, and at the same time, EEG signals are a kind of time-varying, background
The very strong nonstationary random signal of noise, so that electroencephalogramsignal signal analyzing becomes a kind of attractive but has quite hardly possible again
The research topic of degree.
Summary of the invention
The present invention is directed to get the initial data of brain wave by using brain wave acquisition equipment, letter is carried out to initial data
Number processing, extracted feature is trained, and obtains temporal signatures strong, frequency domain character and the stable spy of temporal signatures
Sign.
To achieve the above object, the present invention adopts the following technical scheme that.
A kind of extracting method of brain electrical feature, includes the following steps:
1) EEG signals under different conditions are acquired；
2) it treats identification signal and carries out sub-frame processing, and signal to be identified is divided into the obvious part S1 of feature and feature is unknown
Aobvious part S2；
3) the brain electrical feature information in S1 and S2 is extracted respectively；
4) brain electrical feature is carried out different features to combine to form different feature vectors, statistical is carried out to feature vector
Analysis obtains the feature vector that fast convergence rate, accuracy rate are high, computation complexity is small；
Correspondingly, state includes blink state, state of gritting one's teeth, quiet resting state, reading state in step 1)；
Correspondingly, retaining the data segment that Time Domain Amplitude is greater than 55% in -400uv to the section 400uv accounting in step 1)；
Correspondingly, the Time Domain Amplitude after obtaining sub-frame processing is greater than in -400uv to the section 400uv accounting in step 2)
The short-time energy of 55% data segment, and be normalized, it will be greater than 2 times of average value of feature as the obvious part of feature
S1, remaining is characterized unobvious part S2；
Correspondingly, in step 3), the brain electrical feature of extraction include the energy ratio of δ wave, the energy ratio of θ wave, α wave energy
Than, the energy ratio of β wave, the energy ratio of γ wave, (α+θ)/β, α/β, (α+θ)/(alpha+beta), θ/β, the energy ratio of 0-40Hz, 100-
The energy ratio of 200Hz, average energy, zero-crossing rate；
Correspondingly, feature vector is handled by C language to solve real time problems in step 4)；
Correspondingly, the characteristic value of feature vector is recorded in step 4), and it is for statistical analysis, it determines characteristic threshold value, realizes special
The classification of sign.
The invention has the benefit that on the one hand, solving in real-time processing, the brain telecommunications strong for temporal signatures
On the other hand number extracting difficult problem effectively strengthens the stability of feature from frequency domain and time domain angle, improves wave
The problem of shape severe jamming.
Detailed description of the invention
Fig. 1 is that EEG signals extract flow diagram in one embodiment of the invention.
Specific embodiment
As shown in Figure 1, in one embodiment of the invention, the extracting method of brain wave includes the following steps:
1) EEG signals for acquiring blink state, state of gritting one's teeth, quiet resting state, state of reading a book, retain Time Domain Amplitude
It is greater than 55% data segment in -400uv to the section 400uv accounting；
2) sub-frame processing is carried out in the data segment of -400uv to the section 400uv accounting greater than 55% to Time Domain Amplitude, obtained
The short-time energy energy of a frame voice signal (short-time energy be) after sub-frame processing, and it is normalized that (normalization is a kind of
Simplify the mode calculated, will have the expression formula of dimension, by transformation, turn to nondimensional expression formula, become scalar) processing, it will
For feature of 2 times greater than average value as the obvious part S1 of feature, remaining is characterized unobvious part S2；
3) extract respectively the energy ratio of δ wave in S1 and S2, the energy ratio of θ wave, the energy ratio of α wave, the energy ratio of β wave,
The energy ratio of γ wave, (α+θ)/β, α/β, (α+θ)/(alpha+beta), θ/β, the energy ratio of 0-40Hz, the energy ratio of 100-200Hz,
Average energy, zero-crossing rate, wherein δ wave is the brain wave of 0-4Hz, and θ wave is the brain wave of 4-8Hz, the brain electricity that α wave is 8-12Hz
Wave, β wave are the brain wave of 12-30Hz, and γ wave is the brain wave of 30-100Hz；
4) brain electrical feature is carried out different features to combine to form different feature vectors, statistical is carried out to feature vector
Analysis, obtains the feature vector that fast convergence rate, accuracy rate are high, computation complexity is small, and feature vector is handled by C language to solve
Real time problems record the characteristic value of feature vector, for statistical analysis, determine characteristic threshold value, realize the classification of feature.
Its specific application scenarios is as follows:
1) using brain electrical detection device, (brain electrical detection device is the culture and outside in people, animal brain or brain cell
Equipment room establishes the equipment being directly connected to, such as the Mindwave product of Neursky brand) it acquires and blinks, grits one's teeth, reading a book, stopping
Eeg data under breath state carries out the observation of human eye to eeg data, due to the EEG signals Time Domain Amplitude of this equipment acquisition
Under normal circumstances in -400uv between 400uv, statistical result is confirmed, in one section of EEG signals, 45% or more signal
Time Domain Amplitude is greater than 400uv or is less than 400uv, then it is assumed that the segment signal category interference signal removes interference signal, only retains and mention
Time Domain Amplitude is taken to be greater than 55% data segment in -400uv to the section 400uv accounting, which is valid data section；
2) valid data section is put into different files with the file format of csv, and imported into Matlab, saved
For the data of " mat " format；
3) sub-frame processing is carried out to valid data section, the short-time energy of the valid data section after obtaining sub-frame processing, and it is right
Valid data section is normalized, and will be greater than 2 times of average value of feature as the obvious part S1 of feature, remaining is characterized not
Obvious part S2；
4) energy ratio of the δ wave in extraction S1 and S2, the energy ratio of θ wave, the energy ratio of α wave, the energy ratio γ wave of β wave
Energy ratio, (α+θ)/β, α/β, (α+θ)/(alpha+beta), θ/β, while the energy ratio of 0-40Hz is introduced, the energy ratio of 100-200Hz,
Average energy, zero-crossing rate are as supplementary features, five features of random combine, as a feature vector；
5) using feature vector as the input of classifier, in the present embodiment, the classifier of use includes SVM classifier, ANN
Neural network and boost decision tree, wherein the type of SVM classifier is C-SVC, and kernel function type selects RBF function, in kernel function
Gamma be set as 0.1, the coef0 in kernel function is set as 0.1, and the degree of kernel function is set as 0.1, and the number of iterations is set
It is set to 2000 times, each learning rate is 0.001；Input feature value is 13 in ANN neural network, and hidden layer is set as 10
A neuron, exporting is 4, and wherein the activation primitive of hidden layer is ReLU function, and the activation primitive of last output layer is softmax
Function, the number of iterations are 2000 times, and the error rate of setting is 1e-4, and learning rate is set as 0.01；The stage of boost is 20, if
The each layer of minimum detection rate set is 0.995, and maximum false detection rate is 0.5；By training, fast convergence rate, accuracy rate are obtained
Feature vector high, computation complexity is small；
6) lock-in feature vector, the feature extraction for doing real-time are tested, record character numerical value, after statistical analysis, are determined special
Threshold value is levied, and on the basis of this threshold value, realizes the classification of feature.
In addition, it should be understood that although this specification is described in terms of embodiments, but not each embodiment is only wrapped
Containing an independent technical solution, this description of the specification is merely for the sake of clarity, and those skilled in the art should
It considers the specification as a whole, the technical solutions in the various embodiments may also be suitably combined, forms those skilled in the art
The other embodiments being understood that.
Claims (8)
1. a kind of extracting method of brain electrical feature, it is characterised in that: include the following steps,
1) EEG signals under different conditions are acquired；
2) it treats identification signal and carries out sub-frame processing, and signal to be identified is divided into the obvious part S1 of feature and the unobvious portion of feature
Divide S2；
3) the brain electrical feature information in S1 and S2 is extracted respectively；
4) brain electrical feature different features is carried out to combine to form different feature vectors, it is for statistical analysis to feature vector,
Obtain the feature vector that fast convergence rate, accuracy rate are high, computation complexity is small.
2. a kind of extracting method of brain electrical feature according to claim 1, it is characterised in that: in step 1), state includes
Blink state, state of gritting one's teeth, quiet resting state, reading state.
3. a kind of extracting method of brain electrical feature according to claim 1, it is characterised in that: in step 1), retain time domain
Amplitude is greater than 55% data segment in -400uv to the section 400uv accounting.
4. a kind of extracting method of brain electrical feature according to claim 1, it is characterised in that: in step 2), obtain framing
Treated Time Domain Amplitude and carries out normalizing in the short-time energy of the data segment of -400uv to the section 400uv accounting greater than 55%
Change processing, will be greater than 2 times of average value of feature as the obvious part S1 of feature, remaining is characterized unobvious part S2.
5. a kind of extracting method of brain electrical feature according to claim 1, it is characterised in that: in step 3), the brain of extraction
Electrical feature include the energy ratio of δ wave, the energy ratio of θ wave, the energy ratio of α wave, the energy ratio of β wave, the energy ratio of γ wave, (α+θ)/
β, α/β, (α+θ)/(alpha+beta), θ/β, the energy ratio of 0-40Hz, the energy ratio of 100-200Hz, average energy, zero-crossing rate.
6. a kind of extracting method of brain electrical feature according to claim 1, it is characterised in that: in step 4), feature vector
It is handled by C language to solve real time problems.
7. a kind of extracting method of brain electrical feature according to claim 1, it is characterised in that: in step 4), record feature
The characteristic value of vector, it is for statistical analysis, it determines characteristic threshold value, realizes the classification of feature.
8. a kind of computer readable storage medium is stored thereon with computer program, realization when described program is executed by processor
Such as the step of any one of claim 1-7 the method.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811533519.0A CN109620215A (en) | 2018-12-14 | 2018-12-14 | A kind of extracting method of brain electrical feature |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811533519.0A CN109620215A (en) | 2018-12-14 | 2018-12-14 | A kind of extracting method of brain electrical feature |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN109620215A true CN109620215A (en) | 2019-04-16 |
Family
ID=66074073
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201811533519.0A Pending CN109620215A (en) | 2018-12-14 | 2018-12-14 | A kind of extracting method of brain electrical feature |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109620215A (en) |
Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2011088227A1 (en) * | 2010-01-13 | 2011-07-21 | Regents Of The University Of Minnesota | Imaging epilepsy sources from electrophysiological measurements |
| CN102715903A (en) * | 2012-07-09 | 2012-10-10 | 天津市人民医院 | Method for extracting electroencephalogram characteristic based on quantitative electroencephalogram |
| CN102940490A (en) * | 2012-10-19 | 2013-02-27 | 西安电子科技大学 | Method for extracting motor imagery electroencephalogram signal feature based on non-linear dynamics |
| CN106137187A (en) * | 2016-07-15 | 2016-11-23 | 广州视源电子科技股份有限公司 | Electroencephalogram state detection method and device |
| CN106200975A (en) * | 2016-07-15 | 2016-12-07 | 广州视源电子科技股份有限公司 | Biofeedback relaxation method and device |
| CN107714038A (en) * | 2017-10-12 | 2018-02-23 | 北京翼石科技有限公司 | The feature extracting method and device of a kind of EEG signals |
| CN108143409A (en) * | 2016-12-06 | 2018-06-12 | 中国移动通信有限公司研究院 | Sleep stage method and device by stages |
| CN108960299A (en) * | 2018-06-15 | 2018-12-07 | 东华大学 | A kind of recognition methods of multiclass Mental imagery EEG signals |
-
2018
- 2018-12-14 CN CN201811533519.0A patent/CN109620215A/en active Pending
Patent Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2011088227A1 (en) * | 2010-01-13 | 2011-07-21 | Regents Of The University Of Minnesota | Imaging epilepsy sources from electrophysiological measurements |
| CN102715903A (en) * | 2012-07-09 | 2012-10-10 | 天津市人民医院 | Method for extracting electroencephalogram characteristic based on quantitative electroencephalogram |
| CN102940490A (en) * | 2012-10-19 | 2013-02-27 | 西安电子科技大学 | Method for extracting motor imagery electroencephalogram signal feature based on non-linear dynamics |
| CN106137187A (en) * | 2016-07-15 | 2016-11-23 | 广州视源电子科技股份有限公司 | Electroencephalogram state detection method and device |
| CN106200975A (en) * | 2016-07-15 | 2016-12-07 | 广州视源电子科技股份有限公司 | Biofeedback relaxation method and device |
| CN108143409A (en) * | 2016-12-06 | 2018-06-12 | 中国移动通信有限公司研究院 | Sleep stage method and device by stages |
| CN107714038A (en) * | 2017-10-12 | 2018-02-23 | 北京翼石科技有限公司 | The feature extracting method and device of a kind of EEG signals |
| CN108960299A (en) * | 2018-06-15 | 2018-12-07 | 东华大学 | A kind of recognition methods of multiclass Mental imagery EEG signals |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108388348B (en) | Myoelectric signal gesture recognition method based on deep learning and attention mechanism | |
| CN110797021B (en) | Hybrid speech recognition network training method, hybrid speech recognition device and storage medium | |
| CN110786850B (en) | Electrocardiosignal identity recognition method and system based on multi-feature sparse representation | |
| CN105023573B (en) | It is detected using speech syllable/vowel/phone boundary of auditory attention clue | |
| CN110353702A (en) | A kind of emotion identification method and system based on shallow-layer convolutional neural networks | |
| CN106951753B (en) | Electrocardiosignal authentication method and device | |
| CN107007278A (en) | Sleep mode automatically based on multi-parameter Fusion Features method by stages | |
| CN109344816A (en) | A method of based on brain electricity real-time detection face action | |
| CN108256579A (en) | A kind of multi-modal sense of national identity quantization measuring method based on priori | |
| CN112545532B (en) | Data enhancement method and system for electroencephalogram signal classification and identification | |
| US10885361B2 (en) | Biometric method and device for identifying a person through an electrocardiogram (ECG) waveform | |
| Douglas et al. | Single trial decoding of belief decision making from EEG and fMRI data using independent components features | |
| CN107391994A (en) | A kind of Windows login authentication system methods based on heart sound certification | |
| CN109512441A (en) | Emotion identification method and device based on multiple information | |
| CN109620260A (en) | Psychological condition recognition methods, equipment and storage medium | |
| CN115969392A (en) | Cross-period brainprint recognition method based on tensor frequency space attention domain adaptive network | |
| KR102174232B1 (en) | Bio-signal class classification apparatus and method thereof | |
| CN116738295B (en) | sEMG signal classification method, system, electronic device and storage medium | |
| Liu et al. | Speech emotion recognition based on meta-transfer learning with domain adaption | |
| US20230315203A1 (en) | Brain-Computer Interface Decoding Method and Apparatus Based on Point-Position Equivalent Augmentation | |
| Zhang et al. | Environmental sound recognition using double-level energy detection | |
| CN109620215A (en) | A kind of extracting method of brain electrical feature | |
| CN116531001A (en) | Method and device for generating multi-listener electroencephalogram signals and identifying emotion of cross-listener | |
| Tang et al. | Eye movement prediction based on adaptive BP neural network | |
| Kim et al. | Development of person-independent emotion recognition system based on multiple physiological signals |
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 | ||
| WD01 | Invention patent application deemed withdrawn after publication | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20190416 |