CN110059564A - Feature extracting method based on power spectral density and cross-correlation entropy-spectrum density fusion - Google Patents
Feature extracting method based on power spectral density and cross-correlation entropy-spectrum density fusion Download PDFInfo
- Publication number
- CN110059564A CN110059564A CN201910208851.8A CN201910208851A CN110059564A CN 110059564 A CN110059564 A CN 110059564A CN 201910208851 A CN201910208851 A CN 201910208851A CN 110059564 A CN110059564 A CN 110059564A
- Authority
- CN
- China
- Prior art keywords
- cross
- sample
- correlation entropy
- power spectral
- density
- 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
- G06F18/25—Fusion techniques
- G06F18/253—Fusion techniques of extracted features
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/40—Extraction of image or video features
- G06V10/44—Local feature extraction by analysis of parts of the pattern, e.g. by detecting edges, contours, loops, corners, strokes or intersections; Connectivity analysis, e.g. of connected components
- G06V10/443—Local feature extraction by analysis of parts of the pattern, e.g. by detecting edges, contours, loops, corners, strokes or intersections; Connectivity analysis, e.g. of connected components by matching or filtering
-
- 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
- G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
- G06F2218/08—Feature extraction
Abstract
The invention discloses a kind of feature extracting method based on power spectral density and cross-correlation entropy-spectrum density fusion, cross-correlation entropy-spectrum density is applied into EEG Processing for the first time and extracts frequency domain character, and cross-correlation entropy-spectrum density is merged to obtain a kind of new feature with power spectral density, compared with traditional power spectral density and cross-correlation entropy-spectrum density, fused feature extracting method can not only extract the frequency information in signal well, moreover it is possible to inhibit the influence of noise.The present invention is more suitable for signal-to-noise ratio height and the low signal of signal-to-noise ratio compared to power spectral density and cross-correlation entropy-spectrum density respectively, new feature is applicable not only to both the above scene, and the also various interference signals for the low signal-to-noise ratio of EEG signals and comprising unknown characteristics in environment provide a kind of frequency domain character extracting method of good performance.Therefore the feature extracting method based on power spectral density and cross-correlation entropy-spectrum density fusion is more easily promoted and is used in the utilization of practical brain-computer interface.
Description
[technical field]
The invention belongs to signal processing technology field, it is related to the spy based on power spectral density and cross-correlation entropy-spectrum density fusion
Levy extracting method.
[background technique]
Brain-computer interactive system can be defined as one include many external accessories system, the system can by with
The idea at family controls, i.e., nervous system is wherein directly allowed to exchange with the external world using brain-computer interface.Brain-computer interface can turn brain signal
It is changed to control instruction, helps people not pass through itself muscle and is directly interacted with the external world.Due to the brain electricity of non-intrusion type
(electroencephalography) signal is convenient with acquisition, equipment price is relatively cheap, equipment is relatively portable, non-intruding
Formula brain wave acquisition is a kind of Signal Collection Technology being widely used in brain-machine interaction.Brain based on non-intrusion type EEG signals
The fields such as machine interfacing has been employed for medical, military, education, endowment helps the disabled, entertains.
Due to including various noises, such as electricity physiological signal in EEG signals: myoelectricity, eye electricity, electrocardio, additionally it contained
Various noises in environment, as the power frequency component of 50Hz interferes.In order to extracted from original signal effectively can dtex sign, grind
The persons of studying carefully develop various algorithms.The common attribute extracted from EEG signals has: temporal signatures, frequency domain character, time-frequency characteristics,
But majority robust features Study on Extraction Method concentrates on temporal signatures at present, such as the robust algorithm of common space mode.For frequency
The Robustness Study of characteristic of field extracting method is less, mainly extracts frequency domain character using traditional power spectral density.
Widely used power spectral density be easy when extracting feature it is affected by noise, mainly due to calculate power
When spectrum density, square of signal is directlyed adopt, the negative effect of outlier affected by noise in eeg data can be amplified, thus asked
The feature obtained also will receive the influence of noise.Therefore, the frequency domain character extraction algorithm for being highly desirable exploitation robust is made an uproar to mitigate
The influence of sound.
In the research of cross-correlation function, researcher defines broad sense cross-correlation function, and broad sense cross-correlation function is transported
It uses in power spectral density calculating, obtains cross-correlation entropy-spectrum density, and obtained in life detection and heart rate detection by Successful utilization
Good effect, key are to inhibit the negative effect of noise.In view of in the low signal-to-noise ratio and environment of EEG signals not
Cross-correlation entropy-spectrum density, is applied to extract the feature of EEG signals by the various interference signals for knowing characteristic, and merges traditional function
Rate spectrum density feature obtains having both the new brain electrical feature extracting method of the two characteristic.Compared to two kinds of feature extractions of exclusive use
Method, this method can inhibit the influence of noise under the premise of guaranteeing feature extraction performance.
[summary of the invention]
It is an object of the invention to overcome the above-mentioned prior art, provide based on power spectral density and cross-correlation entropy-spectrum
The feature extracting method of density fusion, this method merge cross-correlation entropy-spectrum density with traditional power spectral density to obtain new spy
Sign, using cross-correlation entropy to big noise have the characteristics that robustness this so that new algorithm is more robust to noise, to make to extract
Feature discriminability it is stronger.
In order to achieve the above objectives, the present invention is achieved by the following scheme:
Feature extracting method based on power spectral density and cross-correlation entropy-spectrum density fusion, comprising the following steps:
Step 1: after obtaining original motion imagination EEG signals, each channel being filtered, data are obtained after filtering
Collection:
{{X1, y1, { X2, y2..., { XN, yN}}
Wherein, N is Mental imagery experiment sample number, each sample includes a data matrix Xi∈Rc×nIt is worth with one
For -1 or 1 label yi, left and right is respectively corresponded, c is number of channels, and n is the sampling number of each section of Mental imagery sample;
Step 2: for each channel of each sample, calculating power spectral density p using formula (1)PSD:
Wherein, vector x is the data that a channel records in a sample, is XiA line, n is the length of vector, function
Rate spectrum density pPSDIt is the row vector that a length is M, power spectral density is calculated separately to each channel of a sample, is obtained
To the power spectral density feature vector of a sample:
After the completion of calculating all samples, N number of feature vector will be obtained:
Then, for each channel of each sample, cross-correlation entropy-spectrum density p is calculated using formula (2)CSD:
Wherein, cross-correlation entropy-spectrum density pCSDThe row vector that size is L, L be by based on Yule-Walker from
It returns spectrum analysis to determine, vcExpression center cross-correlation entropy is calculated by formula (3):
Wherein,The mean value of data cross-correlation entropy, by cross-correlation entropy subtract mean value can effectively eliminate it is straight in data
Flow component, data cross-correlation entropy v (m) and its mean valueIt is calculated by formula (4) and formula (5):
Wherein, x is the data in a channel in sample, is the vector that a length is n, k () is gaussian kernel function:
Wherein, core width cs need to be selected by cross validation, calculates each channel of a certain sample by formula (2)-(6)
Cross-correlation entropy-spectrum density pCSD, it will be able to obtain the cross-correlation entropy-spectrum density feature vector an of sample data matrix:
Then two feature vectors of same sample are merged to obtain N number of feature vector pPSD CSD:
Initial sample set is converted to new sample set:
Wherein, the subscript of feature vector indicates the serial number of sample;
Step 3: respectively under classification method arest neighbors and support vector machines, doing cross validation with new sample set and find often
The optimal core width cs of kind classification methodo, then using the feature for using optimal core width to be calculated as mode input, training
Obtain corresponding disaggregated model.
A further improvement of the present invention lies in that:
In step 1, Signal Pretreatment uses 8-45Hz bandpass filter.
In step 2, core width cs are a free parameters, by training dataset cross validation obtain optimal value.
The feature vector process of the energy spectral density and cross-correlation entropy-spectrum density fusion that extract a certain sample X is as follows:
1) according to existing training sample { { X1, y1, { X2, y2..., { XN, yN, it is up to standard with nicety of grading,
Optimal core width cs are chosen using cross validationo;
2) the corresponding power spectral density feature vector of each sample is calculated according to formula (1):
3) the cross-correlation entropy-spectrum density feature vector of each sample is calculated according to formula (2)-(6):
4) two feature vectors are merged to obtain new feature vector:
Compared with prior art, the invention has the following advantages:
Center width cs of the present invention are capable of the robustness of control algolithm, and when core width is smaller, robustness is preferable, for not
Same problem, degree affected by noise is different, and corresponding core is of different size, selects a suitable core width, may be implemented
Extraordinary robustness, and be the value that can determine core width by training set.
[Detailed description of the invention]
Fig. 1 is the specific implementation flow chart of the method for the present invention;
Fig. 2 is the concrete application flow chart of the method for the present invention;
Fig. 3 is modulation simulation electricity physiological signal;
Fig. 4 is the modulation simulation electricity physiological signal that noise is added;
Fig. 5 is power spectral density processing Noise and the comparison diagram without noise signal;Wherein, (a) be not under plus noise
The power spectral density of noise (b) is added in power spectral density;
Fig. 6 is cross-correlation entropy-spectrum density processing Noise and the comparison diagram without noise signal;Wherein, (a) not plus noise
Under cross-correlation entropy-spectrum density, (b) be added noise cross-correlation entropy-spectrum density.
[specific embodiment]
In order to enable those skilled in the art to better understand the solution of the present invention, below in conjunction in the embodiment of the present invention
Attached drawing, technical scheme in the embodiment of the invention is clearly and completely described, it is clear that described embodiment is only
The embodiment of a part of the invention, the embodiment being not all of, and it is not intended to limit range disclosed by the invention.In addition, with
In lower explanation, descriptions of well-known structures and technologies are omitted, obscures concept disclosed by the invention to avoid unnecessary.It is based on
Embodiment in the present invention, it is obtained by those of ordinary skill in the art without making creative efforts every other
Embodiment should fall within the scope of the present invention.
The various structural schematic diagrams for disclosing embodiment according to the present invention are shown in the attached drawings.These figures are not in proportion
It draws, wherein some details are magnified for the purpose of clear expression, and some details may be omitted.As shown in the figure
The shape in various regions, layer and relative size, the positional relationship between them out is merely exemplary, in practice may be due to
Manufacturing tolerance or technical restriction and be deviated, and those skilled in the art may be additionally designed as required have not
Similar shape, size, the regions/layers of relative position.
In context disclosed by the invention, when one layer/element is referred to as located at another layer/element "upper", the layer/element
Can may exist intermediate layer/element on another layer/element or between them.In addition, if in a kind of court
One layer/element is located at another layer/element "upper" in, then when turn towards when, the layer/element can be located at another layer/
Element "lower".
It should be noted that description and claims of this specification and term " first " in above-mentioned attached drawing, "
Two " etc. be to be used to distinguish similar objects, without being used to describe a particular order or precedence order.It should be understood that using in this way
Data be interchangeable under appropriate circumstances, so that the embodiment of the present invention described herein can be in addition to illustrating here or retouching
Sequence other than those of stating is implemented.In addition, term " includes " and " having " and their any deformation, it is intended that covering
Non-exclusive includes, for example, the process, method, system, product or equipment for containing a series of steps or units are not necessarily limited to clearly
Those of list to Chu step or unit, but may include be not clearly listed or for these process, methods, product or
The intrinsic other step or units of equipment.
The invention will be described in further detail with reference to the accompanying drawing:
Referring to Fig. 1 and Fig. 2, the present invention is based on the feature extracting method of power spectral density and cross-correlation entropy-spectrum density fusion,
Implement in the decoding of Mental imagery brain signal, bandpass filtering and feature extraction in data prediction mainly to data are false
Data { { X is recorded equipped with Mental imagery after bandpass filtering1, y1, { X2, y2..., { XN, yN, N is sample number, each sample
Include a data matrix Xi∈Rc×nThe label y for being -1 or 1 with a valuei, respectively correspond and make left and right, c is number of channels, n
It is the sampling number of each Mental imagery sample data section.Needle is for each sample, the power spectral density in a channel is from phase
The Fourier transformation of relationship number:
Vector x is the data that a channel records in a sample data in formula, is XiA line, n is the length of vector,
Power spectral density pPSDIt is the row vector that a length is M.Power spectrum is calculated separately to each channel of some sample
Degree, the power spectral density feature vector of an available sample
After the completion of calculating all samples, N number of power spectral density feature vector will be obtained
Then, for each channel of each sample, cross-correlation entropy-spectrum density is calculated using following formula:
P in formulaCSDIt is the row vector that a size is L, L is determined by the Autoregressive Spectrum Analysis based on Yule-Walker,
vcIt is center cross-correlation entropy, can be calculated by following formula:
Wherein,It is the mean value of data cross-correlation entropy, cross-correlation entropy, which is subtracted mean value, can effectively eliminate direct current in data
Component, data cross-correlation entropy v (m) and its mean value in formulaCalculation formula:
With it is as before, x is the value in a channel in a sample in formula (4) and (5), be a length be n
Vector, k () are gaussian kernel functions,
Core width cs in formula need to be selected by cross validation.
The cross-correlation entropy-spectrum density p in each channel of a certain sample is calculated by above formula (2)-(6)CSD, can be obtained
The cross-correlation entropy-spectrum density feature vector of one sample data matrix
Gaussian kernel expression formula includes in (6)If some in x (l) and x (l-m) is affected by noise,
Then for the value of the exponential expression with regard to small, the influence to cross-correlation entropy-spectrum density is small, to achieve the effect that inhibit noise.
Gaussian kernel function is available by Taylor expansion
All even order information is contained in formula, compared with power spectral density only includes secondary rank information, is more suitable
Non-Gaussian signal.Except this, power spectral density is the linear operation to signal, and cross-correlation entropy-spectrum density is to the non-linear of signal
Operation.
The calculating process of power spectral density and cross-correlation entropy-spectrum density is described above, it is contemplated that power spectral density energy
The effective frequency information in data is extracted, but the influence of noise cannot be inhibited very well, by cross-correlation entropy-spectrum density and power spectrum
Density blends, i.e., directly obtains two combination of eigenvectorsThis new feature vector is being protected
It demonstrate,proves in the effective situation of information extracted, the influence of noise can be inhibited.
After the completion of calculating all samples, N number of feature vector will be obtainedIt then will be same
Two feature vectors of one sample merge to obtain N number of new feature vectorOriginal sample set is converted to newly
Sample setIt is different from front, feature herein
The subscript of vector indicates the serial number of sample.The process of fusion is as shown in Figure 1.
Classification method arest neighbors and support vector machines are used respectively, are done cross validation with new sample set and are found every kind of classification
The optimal core width cs of methodo.Then using the feature for using optimal core width to be calculated as mode input, training is obtained pair
The disaggregated model answered.
For freshly harvested data, the optimal core width cs determined are usedoDirectly carry out the spy that feature extraction is merged
Levy vector pPSD CSD;Then identification classification is carried out to new sample using trained disaggregated model.Concrete processing procedure is still
So be divided into three steps: data prediction, special medical treatment extract and classification.Partial noise is removed using 8-45Hz bandpass filtering first,
Then using the optimal core width cs being determined aboveoCalculate the feature based on power spectral density and cross-correlation entropy-spectrum density fusion
Vector pPSD CSD, feature vector is inputted into trained disaggregated model, obtains corresponding classification results.Classification results are just
It can be used as the measurement of feature extracting method performance quality whether really.
Embodiment
The feature vector process of the power spectral density and cross-correlation entropy-spectrum density fusion of extracting a certain sample X is as follows:
1) according to existing training sample { { X1, y1, { X2, y2..., { XN, yN, it is up to standard with nicety of grading,
Optimal core width cs are chosen using cross validationo;
2) the power spectral density feature vector of each sample is calculated according to formula (1):
3) the cross-correlation entropy-spectrum density feature vector of each sample is calculated according to formula (2)-(6):
4) two feature vectors are merged to obtain new feature vector:
Simulation analysis
In order to show the advantage of the feature extracting method based on power spectral density and cross-correlation entropy-spectrum density fusion, carry out
Two groups of experiments, first group of emulation experiment are used to verify the robustness of cross-correlation entropy-spectrum density, and second group of experiment is in real EEG electricity number
It is a kind of outstanding algorithm to verify new feature extracting method in actual use according to upper progress.
Electricity physiological signal, the formula that analog signal generates are simulated in first group of experiment using modulated signal are as follows:
X (l)=(1+sin (2 π fcl))sin(2πfml)
Frequency of carrier signal f in formulac=4Hz, frequency modulating signal fm=2Hz.
Under 10Hz sample frequency, 1000 sample points are generated with above-mentioned formula, obtain sample data such as Fig. 3, for verifying
Its robustness generates 100 noises with the Levy alpha-stable distribution that parameter is [α β δ γ]=[1.8 03 0]
Point is added on 1000 sample points at random, and 1000 sample data such as Fig. 4 after noise is added then use power spectrum respectively
Degree and cross-correlation entropy-spectrum density carry out feature extractions to two groups of data, and obtained result is respectively such as Fig. 5 and Fig. 6, wherein Fig. 5 (a)
The power spectral density of noise is added in the not power spectral density under plus noise, Fig. 5 (b);Fig. 6 (a) not cross-correlation entropy under plus noise
The cross-correlation entropy-spectrum density of noise is added in spectrum density, Fig. 6 (b).Comparison diagram 5 and Fig. 6, power spectral density cannot mention after noise is added
The frequency of modulated signal is taken out, and cross-correlation entropy-spectrum density remains to extract the frequency of modulated signal.
Second group of experiment carries out in the Section II b data of brain-machine interaction match IV data set, data downloading and details
Referring to website:http://www.bbci.de/competition/iv/.The data set includes 5 groups of data, is had for three groups after selection
The data of visual feedback use the method for three kinds of feature extractions in above-mentioned data respectively, mutual light entropy and power spectral density
Use it is identical as the process for using of new method, in power spectral density do not include free parameter, remove free parameter determine step
Suddenly.Result such as Tables 1 and 2 in 9 subjects, respectively corresponds two different classification methods, as a result takes 10 Monte Carlos
Mean value.As can be seen from the table, in 9 subjects, most of subject is melted using power spectral density and cross-correlation entropy-spectrum density
After nicety of grading is higher than power spectral density and cross-correlation entropy-spectrum Density extraction feature is used respectively after the feature extracting method of conjunction
Nicety of grading, and average value is also above the classification essence after using power spectral density and cross-correlation entropy-spectrum Density extraction feature respectively
Degree.
1 data of table not plus noise, when classifier is arest neighbors, nicety of grading
Note: black matrix indicates that three kinds of feature extracting method intermediate values are maximum
2 data of table not plus noise, when classifier is support vector machines, nicety of grading
Note: black matrix indicates that three kinds of feature extracting method intermediate values are maximum
Then to further look at robustness of the feature extracting method on EEG signals, 5% data in each channel
The alpha-stable partition noise that parameter is [α β δ γ]=[1.8 0.1 0.1 0] is added on point at random.Again by three kinds
Feature extracting method is used for the data, and calculating process and not Noise phase are same.Obtain the classification average value such as table of 9 subjects
3, result is 10 Monte Carlo mean values.The spy of power spectral density and cross-correlation entropy-spectrum density fusion is used as can be seen from the table
Nicety of grading is still best after levying extracting method, and the accuracy decline of cross-correlation entropy-spectrum density is minimum, as a result illustrates actual
New feature extracting method still has robustness on eeg data.
3 Noise Data of table, the average value of nicety of grading
In conclusion the feature extracting method accuracy based on power spectral density and cross-correlation entropy-spectrum density fusion is good, Shandong
Stick is high, has high application value.
The above content is merely illustrative of the invention's technical idea, and this does not limit the scope of protection of the present invention, all to press
According to technical idea proposed by the present invention, any changes made on the basis of the technical scheme each falls within claims of the present invention
Protection scope within.
Claims (4)
1. the feature extracting method based on power spectral density and cross-correlation entropy-spectrum density fusion, which is characterized in that including following step
It is rapid:
Step 1: after obtaining original motion imagination EEG signals, each channel is filtered, data set is obtained after filtering:
{{X1, y1, { X2, y2..., { XN, yN}}
Wherein, N is Mental imagery experiment sample number, each sample includes a data matrix Xi∈Rc×nIt is -1 with a value
Or 1 label yi, left and right is respectively corresponded, c is number of channels, and n is the sampling number of each section of Mental imagery sample;
Step 2: for each channel of each sample, calculating power spectral density p using formula (1)PSD:
Wherein, vector x is the data that a channel records in a sample, is XiA line, n is the length of vector, power spectrum
Spend pPSDIt is the row vector that a length is M;Power spectral density is calculated separately to each channel of a sample, obtains one
The power spectral density feature vector of sample:
After the completion of calculating all samples, N number of feature vector will be obtained:
Then, for each channel of each sample, cross-correlation entropy-spectrum density p is calculated using formula (2)CSD:
Wherein, cross-correlation entropy-spectrum density pCSDIt is the row vector that a length is L, L is by the autoregression based on Yule-Walker
Spectrum analysis decision, vcExpression center cross-correlation entropy is calculated by formula (3):
Wherein,It is the mean value of data cross-correlation entropy, cross-correlation entropy, which is subtracted mean value, can effectively eliminate direct current in data point
Amount, data cross-correlation entropy v (m) and its mean valueIt is calculated by formula (4) and formula (5):
Wherein, x is the data in a channel in sample, is the vector that a length is n, k () is gaussian kernel function:
Wherein, core width cs need to be selected by cross validation, calculates the mutual of each channel of sample by formula (2)-(6)
Close entropy-spectrum density pCSD, it will be able to obtain the cross-correlation entropy-spectrum density feature vector an of sample data matrix:
Then two feature vectors of same sample are merged to obtain N number of feature vector pPSDCSD:
Initial sample set is converted to new sample set:
Wherein, the subscript of feature vector indicates the serial number of sample;
Step 3: respectively under classification method arest neighbors and support vector machines, new sample set being done into cross validation and finds every kind
The optimal core width cs of classification methodo, then corresponded to the feature for using optimal core width to be calculated as input training
Disaggregated model.
2. the feature extracting method according to claim 1 based on power spectral density and cross-correlation entropy-spectrum density fusion,
It is characterized in that, in step 1, Signal Pretreatment uses 8-45Hz bandpass filter.
3. the feature extracting method according to claim 1 based on power spectral density and cross-correlation entropy-spectrum density fusion,
Be characterized in that, in step 2, core width cs are a free parameters, by training dataset cross validation obtain optimal value.
4. the feature extracting method according to claim 1 based on power spectral density and cross-correlation entropy-spectrum density fusion,
It is characterized in that, the process for extracting the feature vector of a certain sample X is as follows:
1) according to existing training sample set { { X1, y1, { X2, y2..., { XN, yN, it is up to standard with nicety of grading, makes
Optimal core width cs are chosen with cross validationo;
2) the corresponding power spectral density feature vector of each sample is calculated according to formula (1):
3) the cross-correlation entropy-spectrum density feature vector of each sample is calculated according to formula (2)-(6):
4) obtain two feature vectors are merged to obtain new feature vector:
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910208851.8A CN110059564B (en) | 2019-03-19 | 2019-03-19 | Feature extraction method based on power spectral density and cross-correlation entropy spectral density fusion |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910208851.8A CN110059564B (en) | 2019-03-19 | 2019-03-19 | Feature extraction method based on power spectral density and cross-correlation entropy spectral density fusion |
Publications (2)
Publication Number | Publication Date |
---|---|
CN110059564A true CN110059564A (en) | 2019-07-26 |
CN110059564B CN110059564B (en) | 2021-01-19 |
Family
ID=67317223
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201910208851.8A Active CN110059564B (en) | 2019-03-19 | 2019-03-19 | Feature extraction method based on power spectral density and cross-correlation entropy spectral density fusion |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN110059564B (en) |
Cited By (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111053553A (en) * | 2019-12-09 | 2020-04-24 | 清华大学 | Brain-computer interface lead selection method based on event-related potential distribution |
CN111310571A (en) * | 2020-01-17 | 2020-06-19 | 中国科学院长春光学精密机械与物理研究所 | Hyperspectral image classification method and device based on spatial-spectral-dimensional filtering |
CN111626093A (en) * | 2020-03-27 | 2020-09-04 | 国网江西省电力有限公司电力科学研究院 | Electric transmission line related bird species identification method based on sound power spectral density |
CN112190261A (en) * | 2020-09-16 | 2021-01-08 | 电子科技大学 | Autism electroencephalogram signal classification device based on resting brain network |
CN115219199A (en) * | 2022-07-14 | 2022-10-21 | 天津职业技术师范大学 | Bearing weak fault extraction method based on depth-related entropy spectral density |
Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20100241020A1 (en) * | 2009-02-05 | 2010-09-23 | Yissum Research Development Company Of The Hebrew University Of Jerusalem | Real-time methods and systems for mapping a target region in the brain during surgery |
CN103425249A (en) * | 2013-09-06 | 2013-12-04 | 西安电子科技大学 | Electroencephalogram signal classifying and recognizing method based on regularized CSP and regularized SRC and electroencephalogram signal remote control system |
US20170325702A1 (en) * | 2016-05-11 | 2017-11-16 | Oulun Yliopisto | Apparatus and method for electroencephalographic measurement |
CN107832709A (en) * | 2017-11-10 | 2018-03-23 | 辽宁师范大学 | The brain electricity emotion identification method of feature extraction is carried out based on experience wavelet transformation |
CN108888264A (en) * | 2018-05-03 | 2018-11-27 | 南京邮电大学 | EMD and CSP merges power spectral density brain electrical feature extracting method |
-
2019
- 2019-03-19 CN CN201910208851.8A patent/CN110059564B/en active Active
Patent Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20100241020A1 (en) * | 2009-02-05 | 2010-09-23 | Yissum Research Development Company Of The Hebrew University Of Jerusalem | Real-time methods and systems for mapping a target region in the brain during surgery |
CN103425249A (en) * | 2013-09-06 | 2013-12-04 | 西安电子科技大学 | Electroencephalogram signal classifying and recognizing method based on regularized CSP and regularized SRC and electroencephalogram signal remote control system |
US20170325702A1 (en) * | 2016-05-11 | 2017-11-16 | Oulun Yliopisto | Apparatus and method for electroencephalographic measurement |
CN107832709A (en) * | 2017-11-10 | 2018-03-23 | 辽宁师范大学 | The brain electricity emotion identification method of feature extraction is carried out based on experience wavelet transformation |
CN108888264A (en) * | 2018-05-03 | 2018-11-27 | 南京邮电大学 | EMD and CSP merges power spectral density brain electrical feature extracting method |
Non-Patent Citations (2)
Title |
---|
ROBINSON N等: "EEG-based classification of fast and slow hand movements using wavelet-CSP algorithm", 《IEEE TRANSACTIONS ON BIOMEDICAL ENGINEERING》 * |
李钢等: "基于时域滤波互谱密度的扩频信号测向", 《专题技术与工程应用》 * |
Cited By (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111053553A (en) * | 2019-12-09 | 2020-04-24 | 清华大学 | Brain-computer interface lead selection method based on event-related potential distribution |
CN111310571A (en) * | 2020-01-17 | 2020-06-19 | 中国科学院长春光学精密机械与物理研究所 | Hyperspectral image classification method and device based on spatial-spectral-dimensional filtering |
CN111626093A (en) * | 2020-03-27 | 2020-09-04 | 国网江西省电力有限公司电力科学研究院 | Electric transmission line related bird species identification method based on sound power spectral density |
CN111626093B (en) * | 2020-03-27 | 2023-12-26 | 国网江西省电力有限公司电力科学研究院 | Method for identifying related bird species of power transmission line based on sound power spectral density |
CN112190261A (en) * | 2020-09-16 | 2021-01-08 | 电子科技大学 | Autism electroencephalogram signal classification device based on resting brain network |
CN115219199A (en) * | 2022-07-14 | 2022-10-21 | 天津职业技术师范大学 | Bearing weak fault extraction method based on depth-related entropy spectral density |
Also Published As
Publication number | Publication date |
---|---|
CN110059564B (en) | 2021-01-19 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN110059564A (en) | Feature extracting method based on power spectral density and cross-correlation entropy-spectrum density fusion | |
Yuan et al. | Fingerprint liveness detection using an improved CNN with image scale equalization | |
Billinger et al. | Single-trial connectivity estimation for classification of motor imagery data | |
Xu et al. | High accuracy classification of EEG signal | |
CN110991406B (en) | RSVP electroencephalogram characteristic-based small target detection method and system | |
CN109497990B (en) | Electrocardiosignal identity recognition method and system based on canonical correlation analysis | |
Mert | ECG feature extraction based on the bandwidth properties of variational mode decomposition | |
Falzon et al. | Complex-valued spatial filters for SSVEP-based BCIs with phase coding | |
CN103425249A (en) | Electroencephalogram signal classifying and recognizing method based on regularized CSP and regularized SRC and electroencephalogram signal remote control system | |
CN104536573B (en) | A kind of brain-machine interface method stimulated based on high frequency flicker emotion | |
CN108021892A (en) | A kind of human face in-vivo detection method based on extremely short video | |
CN105212949A (en) | A kind of method using skin pricktest signal to carry out culture experience emotion recognition | |
CN108042132A (en) | Brain electrical feature extracting method based on DWT and EMD fusions CSP | |
Gao et al. | Multi-ganglion ANN based feature learning with application to P300-BCI signal classification | |
CN107661099A (en) | A kind of brain electricity analytical method for detecting the horizontal lower motion state of different Mental Workloads | |
CN110287918A (en) | Vivo identification method and Related product | |
CN111000555A (en) | Training data generation method, automatic recognition model modeling method and automatic recognition method for epilepsia electroencephalogram signals | |
CN111580643A (en) | Brain-computer interface method based on steady-state asymmetric visual evoked potential | |
Jianfeng et al. | Multi-feature authentication system based on event evoked electroencephalogram | |
Malekmohammadi et al. | An efficient hardware implementation for a motor imagery brain computer interface system | |
Mingai et al. | A novel EEG feature extraction method based on OEMD and CSP algorithm | |
CN101703405A (en) | Wavelet analysis and Welch algorithm personal identification method | |
Abbas et al. | FBCSP-based multi-class motor imagery classification using BP and TDP features | |
CN106897656A (en) | Brain-computer interface training method and device based on Mental imagery | |
Chen et al. | An adaptive feature extraction method for motor-imagery BCI systems |
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 |