CN103892829B - Eye movement signal identification system based on common spatial mode and identification method thereof - Google Patents
Eye movement signal identification system based on common spatial mode and identification method thereof Download PDFInfo
- Publication number
- CN103892829B CN103892829B CN201410156043.9A CN201410156043A CN103892829B CN 103892829 B CN103892829 B CN 103892829B CN 201410156043 A CN201410156043 A CN 201410156043A CN 103892829 B CN103892829 B CN 103892829B
- Authority
- CN
- China
- Prior art keywords
- eye
- signal
- moves
- matrix
- pan
- 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.)
- Expired - Fee Related
Links
- 238000000034 method Methods 0.000 title claims abstract description 52
- 230000004424 eye movement Effects 0.000 title claims abstract description 19
- 238000012549 training Methods 0.000 claims abstract description 29
- 238000001914 filtration Methods 0.000 claims abstract description 19
- 238000012360 testing method Methods 0.000 claims abstract description 17
- 238000007781 pre-processing Methods 0.000 claims abstract description 11
- 210000001508 eye Anatomy 0.000 claims description 125
- 239000011159 matrix material Substances 0.000 claims description 48
- 238000012706 support-vector machine Methods 0.000 claims description 20
- 230000008569 process Effects 0.000 claims description 17
- 210000005252 bulbus oculi Anatomy 0.000 claims description 6
- 238000000605 extraction Methods 0.000 claims description 6
- 230000009466 transformation Effects 0.000 claims description 6
- 238000012935 Averaging Methods 0.000 claims description 3
- 229910021607 Silver chloride Inorganic materials 0.000 claims description 3
- 238000006243 chemical reaction Methods 0.000 claims description 3
- 238000000354 decomposition reaction Methods 0.000 claims description 3
- 238000010606 normalization Methods 0.000 claims description 3
- HKZLPVFGJNLROG-UHFFFAOYSA-M silver monochloride Chemical compound [Cl-].[Ag+] HKZLPVFGJNLROG-UHFFFAOYSA-M 0.000 claims description 3
- 210000000481 breast Anatomy 0.000 claims description 2
- 210000001061 forehead Anatomy 0.000 claims description 2
- 230000000638 stimulation Effects 0.000 claims description 2
- 230000008901 benefit Effects 0.000 abstract description 3
- 230000000875 corresponding effect Effects 0.000 description 5
- 230000008859 change Effects 0.000 description 4
- 230000002596 correlated effect Effects 0.000 description 4
- 238000010586 diagram Methods 0.000 description 4
- 230000005611 electricity Effects 0.000 description 4
- 238000002474 experimental method Methods 0.000 description 4
- 238000010408 sweeping Methods 0.000 description 4
- 210000004087 cornea Anatomy 0.000 description 3
- 210000001525 retina Anatomy 0.000 description 3
- 230000009471 action Effects 0.000 description 2
- 210000003484 anatomy Anatomy 0.000 description 2
- 230000003542 behavioural effect Effects 0.000 description 2
- 210000004556 brain Anatomy 0.000 description 2
- 238000010276 construction Methods 0.000 description 2
- 230000003993 interaction Effects 0.000 description 2
- 230000002452 interceptive effect Effects 0.000 description 2
- 238000011160 research Methods 0.000 description 2
- 210000003583 retinal pigment epithelium Anatomy 0.000 description 2
- 235000007926 Craterellus fallax Nutrition 0.000 description 1
- 240000007175 Datura inoxia Species 0.000 description 1
- 238000013473 artificial intelligence Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000015572 biosynthetic process Effects 0.000 description 1
- 230000004397 blinking Effects 0.000 description 1
- 210000003986 cell retinal photoreceptor Anatomy 0.000 description 1
- 239000012141 concentrate Substances 0.000 description 1
- 230000007812 deficiency Effects 0.000 description 1
- 238000001514 detection method Methods 0.000 description 1
- 238000003745 diagnosis Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000004070 electrodeposition Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 230000007613 environmental effect Effects 0.000 description 1
- 230000006870 function Effects 0.000 description 1
- 239000000203 mixture Substances 0.000 description 1
- 238000012544 monitoring process Methods 0.000 description 1
- 238000003909 pattern recognition Methods 0.000 description 1
- 230000008447 perception Effects 0.000 description 1
- 108091008695 photoreceptors Proteins 0.000 description 1
- 230000004044 response Effects 0.000 description 1
- 230000002123 temporal effect Effects 0.000 description 1
- 230000002087 whitening effect Effects 0.000 description 1
Landscapes
- Eye Examination Apparatus (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses an eye movement signal identification system and an eye movement signal identification method based on a common space mode. The identification method comprises the following steps: acquiring eye movement data based on an electrooculogram and preprocessing the eye movement data; dividing all the preprocessed data into two parts of training data and testing data; performing spatial filtering on the training data by adopting a CSP algorithm, and inputting a filtered result serving as a characteristic parameter into an SVM classifier to perform SVM model training; and (3) similarly extracting the characteristics of the test data by using a CSP algorithm, and sending the result into a trained SVM classifier for recognition to finally obtain the recognition result of the eye movement. The eye movement signal identification system and the identification method thereof have the advantages of high eye movement signal identification accuracy, strong eye movement signal expansion classification capability, large application potential and the like.
Description
Technical field
The present invention relates to a kind of eye based on common space pattern (CommonSpatialPattern, CSP) and move signal recognition system and recognition methods thereof.
Background technology
Human bodys' response (HumanActivityRecognition, HAR) refers to synthetically to be analyzed being observed the individual information such as type of action, behavioral pattern and identifying, and is described by modes such as natural languages by recognition result.Because HAR system can active perception user view, therefore be with a wide range of applications in fields such as intelligent video monitoring, medical diagnosis, motion analysis and man-machine interactions, and become in artificial intelligence and area of pattern recognition an emerging study hotspot.Present stage, the acquisition of human body behavioural information mainly adopts contactless environmental sensor and wearable human body information sensor two kinds of methods.Wherein, effectively can make up the deficiency of traditional HAR system based on wearable biopotential sensor, progressively become HAR system and realizing the new research direction of of field.And in human biological signal, electro-ocular signal (EOG) compare other bioelectrical signals such as brain electricity, electrocardio have amplitude large, be easy to detect, the carry information more feature such as horn of plenty, therefore, use electro-ocular signal to carry out HAR and identify the advantage had not available for other bioelectrical signals.Meanwhile, in our daily life, almost done every thing feelings all be unable to do without eyes, and therefore, the information that the electro-ocular signal produced due to eye motion comprises will provide very valuable clue for HAR system.
So-called electro-ocular signal refers to the potential difference existed between the electromotive force that people produces due to the motion of eyes and human skin's electrode potential.Further medical research shows, this electromotive force relation is caused by the electric potential difference between the cornea of eyes and retina (as shown in Fig. 1 left side).This electromotive force is initiated by retinal pigment epithelium and photoreceptor cell, its positive pole is positioned at photoreceptor end, negative pole is positioned at retinal pigment epithelium end, the electric current produced has flowed to cornea end from retina end, thus formation amplitude is about 0.4mV ~ 10mV, cornea is positive pole, retina is the electromotive force of negative pole, and we claim this electromotive force to be electro-ocular signal.When people's ocular movement, continuous change can be there is in the amplitude of electro-ocular signal along with the motion of eyeball, the electromotive force of this change is plotted to the curve that time shaft then can form by us, and this curve is just referred to as electro-oculogram, and Fig. 2 shows one section and contains the EOG collection of illustrative plates that eye moves signal.
In the HAR based on EOG identifies, the identification of pan signal is a most key step, and in order to realize identifying it, some eye movement characteristics parameters are suggested in succession.Comprise 1) utilize the visible angle of original EOG signal to carry out end-point detection and identification that eye moves signal; 2) EOG signal intensity corresponding when utilizing Rotation of eyeball faster feature extraction eye moves the characteristic parameter of signal; 3) eye is used to move the statistics of signal and temporal signatures carries out the method such as identifying.But above-mentioned way mainly concentrates on the analysis of time-domain characteristic to EOG signal, obviously, under some noise circumstances, (such as: the movement of Hz noise, electrode position, the distortion etc. of channel) will be difficult to the characteristic describing primary signal exactly, and in true use procedure, this noise jamming is again often inevitable, and the pan signal characteristic extracting methods therefore based on time-domain analysis presents poor robustness; On the other hand, EOG signal moves information (comprise significantly sweep, micro-ly to sweep, micro-jumping etc.) to obtain abundanter eye in gatherer process, and multiple leading usually can be used to carry out data acquisition.At present, multi-lead eye is moved to the process of signal, the way that everybody commonly uses is the position according to leading, targetedly each leads is separated and analyze, obviously, this way only considers the change of single lead signals and ignores the related information between leading, and is unfavorable for that eye moves the lifting of signal identification rate.
Summary of the invention
The present invention is for avoiding the weak point that exists in above-mentioned prior art, providing the eye based on common space pattern that a kind of discriminator accuracy is high, extended capability is strong, application potential is large to move signal recognition method.
The present invention be technical solution problem by the following technical solutions.
Eye based on common space pattern moves a signal recognition system, and its construction features is, comprises eye and moves signal pre-processing module, spatial filter training module and eye and move Signal analysis module;
Described eye moves signal pre-processing module, is used for carrying out bandpass filtering to all training and test data and removing averaging operation, in order to reduce different noise signal original multi-lead eye is moved to the interference of signal, thus improves recognition correct rate;
Described spatial filter training module, moves signal space wave filter for the many classification eye built based on common space pattern by the method for Joint Diagonalization of Matrix;
Described eye moves Signal analysis module, for original multi-lead eye being moved result that signal obtains after above-mentioned spatial filter filtering as characteristic parameter, and use support vector machine (SupportVectorMachine, SVM) to carry out identification that eye moves signal.
The construction features that a kind of eye based on common space pattern of the present invention moves signal recognition system and recognition methods thereof is also:
Described a kind of eye based on common space pattern moves the recognition methods of signal recognition system, is characterized in comprising the steps:
Step 1: gather the eye movement data based on electro-oculogram and pretreatment is carried out to eye movement data;
Step 2: pretreated all data are divided into training data and test data two parts; Described training data is that original multi-lead eye moves signal, adopts CSP algorithm to move signal to original multi-lead eye and carries out airspace filter, calculates the spatial filter W of left pan, right pan, upper pan and lower pan four class
l, W
r, W
uand W
d;
Step 3: use the W in above-mentioned steps 2
l, W
r, W
uand W
dfour spatial filters carry out airspace filter to training data, and result after filtering are input in SVM classifier as characteristic parameter and train;
Step 4: to test data, the same W used in above-mentioned steps 2
l, W
r, W
uand W
dfour spatial filters carry out airspace filter, feature extraction, are then sent in the SVM classifier trained and identify, finally obtain the recognition result that eye moves signal.
In described step 1 to eye move signal carry out preprocessing process comprise bandpass filtering step with go average step.
Be 0.3Hz ~ 12.5Hz for carrying out the cut-off frequency of the band filter of bandpass filtering step.
Adopt in described step 2 CSP algorithm to original multi-lead eye move signal carry out airspace filter time, spatial filter W
l, W
r, W
uand W
dcomputational process be:
Assuming that X
l, X
r, X
uwith X
drepresent that the original multi-lead eye under left pan, right pan, upper pan and lower pan 4 kinds of tasks moves signal, X respectively
l, X
r, X
uwith X
dspace covariance matrix after normalization is
The covariance matrix of the many groups training data under same eye movement mode is averaged, obtains average covariance matrices
with
calculate the average covariance matrices sum that all eyes move signal
then Eigenvalues Decomposition is carried out to C, namely
In formula (2), U
0representation feature vector,
representing matrix U
0transposed matrix, Σ is the diagonal matrix of eigenvalue, then prewhitening transformation matrix P can be expressed as
Utilize prewhitening transformation matrix P to above-mentioned 4 classification average covariance matrices
with
convert, can obtain
In formula (4), S
l, S
r, S
uand S
dbe respectively average covariance matrices
with
carry out the matrix after prewhitening conversion, to S
l, S
r, S
uand S
dafter carrying out Joint Diagonalization of Matrix, an orthogonal matrix U and approximate diagonal battle array Σ can be found
l, Σ
r, Σ
uwith Σ
d, and meet following relation
US
lU
T＝Σ
lUS
uU
T＝Σ
u
(5)
US
rU
T＝Σ
rUS
dU
T＝Σ
d
Σ
l+Σ
r+Σ
u+Σ
d＝I(6)
In formula (5) and formula (6), I is unit matrix; Eye is moved to many classification problems of signal, diagonal matrix Σ
l, Σ
r, Σ
uand Σ
dcan carry out eigenvalue selection according to following formula, suppose that classification number to be sorted is N, diagonal matrix is Σ, then
score(Σ)＝max(Σ,(1-Σ)/(1-Σ+(N-1)
2Σ))(7)
Calculate Σ respectively
l, Σ
r, Σ
uand Σ
dthe eigenvalue that middle score is the highest, and take out the highest eigenvalue of this score row vector U corresponding in orthogonal matrix U
l, U
r, U
uand U
d, then calculate corresponding spatial filter by following formula (8), namely
In formula (8), W
l, W
r, W
uand W
drepresent the spatial filter of left pan, right pan, upper pan and lower pan four class respectively.
In step 3, move in Signal analysis process eye, the many classification task correlated source signal (left and right, upper and lower pan four class eye moves the relevant source signal of task) obtained after airspace filter obtains by following formula (9)
Z
l＝W
lXZ
r＝W
rXZ
u＝W
uXZ
d＝W
dX(9)
In formula (9), X represents that original multi-lead eye moves signal, Z
l, Z
r, Z
u, Z
drepresent the result of left and right, upper and lower pan signal after airspace filter (i.e. many classification task correlated source signal) respectively, and the characteristic parameter that this result moves signal as eye is sent into SVM classifier, in order to train SVM model or to identify test data.
Compared with the prior art, beneficial effect of the present invention is embodied in:
A kind of eye based on common space pattern of the present invention moves signal recognition system and recognition methods thereof, has the feature of the following aspects.
1, the present invention moves signal to eye and has higher recognition correct rate.
Present invention employs common space pattern (CSP) algorithm, this algorithm builds the spatial filter relevant to certain particular task according to the label data marked in advance, while extracting the component relevant to this particular task, effectively can suppress to move with eye the interference of component that signal has nothing to do and noise, carry out many experiments to several experimenter, its result shows that Mean accurate rate of recognition of the present invention all reaches more than 98%.
2, the present invention has stronger eye and moves signal extension classification capacity.
The present invention have employed the method for approximate matrix Joint diagonalization when realizing multiclass CSP classification, carry out approximately joint diagonalization, to obtain the orthogonal whitening matrix of original each observation signal by multiclass covariance matrix.The method not only can be applied to the classification that the four class eyes mentioned in the present invention move signal (sweep, sweep, upwards sweep and sweep) left to the right downwards, when eye move signal type increase time, such as: the signal of blinking of different number of times, the eye of different pan angle move signal etc., the method does not need to carry out the classification that too much change just can realize more dynamic signals, has stronger expansion classification capacity.
3, the present invention has huge application potential.
The dynamic signal of eye is as a kind of principal mode of man-machine interaction's method, there is more and more important effect in our daily life, especially have a bodily deformity concerning some and can not paleocinetic crowd, the man-machine interactive system based on eye electricity can effectively help them to promote quality of life; On the other hand, be inconvenient to make bimanual occasion for some, such as: under mine, in spacecraft, this system can effectively improve user work quality.In the man-machine interactive system based on eye electricity, the recognition accuracy of the dynamic signal of eye is particularly important, higher accuracy just more easily allows user accept, and main purpose of the present invention is exactly the precise classification in order to realize moving eye signal, and the method can be implemented in line function, therefore the present invention has huge application potential.
A kind of eye based on common space pattern of the present invention moves signal recognition system and recognition methods thereof, have eye is moved that Signal analysis accuracy is higher, eye moves the advantages such as signal extension classification capacity is comparatively strong, application potential is large.
Accompanying drawing explanation
Fig. 1 is human eye internal anatomy.
Fig. 2 is that in the present invention, eye moves signal generating principle figure (electro-oculogram during pan of left and right).
Fig. 3 is algorithm basic flow sheet of the present invention.
Fig. 4 is that eye of the present invention moves distribution of electrodes figure in signal acquisition process.
Fig. 5 is the relative position schematic diagram that eye of the present invention moves observed object and experimenter in signal acquisition process.
Fig. 6 is that the present invention moves single experiment normal form schematic diagram.
Fig. 7 is the original electro-ocular signal oscillogram of passage of the present invention.
Fig. 8 is CSP spatial filter coefficient schematic diagram of the present invention.
Below by way of detailed description of the invention, and the invention will be further described by reference to the accompanying drawings.
Detailed description of the invention
See Fig. 1 ~ 8, a kind of eye based on common space pattern moves signal recognition system, and it comprises eye and moves signal pre-processing module, spatial filter training module and eye and move Signal analysis module;
Described eye moves signal pre-processing module, is used for carrying out bandpass filtering to all training and test data and removing averaging operation, in order to reduce different noise signal original multi-lead eye is moved to the interference of signal, thus improves recognition correct rate;
Described spatial filter training module, moves signal space wave filter for the many classification eye built based on common space pattern by the method for Joint Diagonalization of Matrix;
Described eye moves Signal analysis module, for original multi-lead eye being moved result that signal obtains after above-mentioned spatial filter filtering as characteristic parameter, and use support vector machine (SupportVectorMachine, SVM) to carry out identification that eye moves signal.
Described a kind of eye based on common space pattern moves the recognition methods of signal recognition system, and it comprises the steps:
Step 1: gather the eye movement data based on electro-oculogram and pretreatment is carried out to eye movement data;
Step 2: pretreated all data are divided into training data and test data two parts; Described training data is that original multi-lead eye moves signal, adopts CSP algorithm to move signal to original multi-lead eye and carries out airspace filter, calculates the spatial filter W of left pan, right pan, upper pan and lower pan four class
l, W
r, W
uand W
d;
Step 3: use the W in above-mentioned steps 2
l, W
r, W
uand W
dfour spatial filters carry out airspace filter to training data, and result after filtering are input in SVM classifier as characteristic parameter and train;
Step 4: to test data, the same W used in above-mentioned steps 2
l, W
r, W
uand W
dfour spatial filters carry out airspace filter, feature extraction, are then sent in the SVM classifier trained and identify, finally obtain the recognition result that eye moves signal.
In described step 1 to eye move signal carry out preprocessing process comprise bandpass filtering step with go average step.
Described bandpass filtering step comprises the interference of the signal such as baseline drift, myoelectricity EMG, electrocardio ECG, brain electricity EEG with going the object of average step to be used to remove.
Be 0.3Hz ~ 12.5Hz for carrying out the cut-off frequency of the band filter of bandpass filtering step.
Adopt in described step 2 CSP algorithm to original multi-lead eye move signal carry out airspace filter time, spatial filter W
l, W
r, W
uand W
dcomputational process be:
Assuming that X
l, X
r, X
uwith X
drepresent that the original multi-lead eye under left pan, right pan, upper pan and lower pan 4 kinds of tasks moves signal, X respectively
l, X
r, X
uwith X
dspace covariance matrix after normalization is
The covariance matrix of the many groups training data under same eye movement mode is averaged, obtains average covariance matrices
with
calculate the average covariance matrices sum that all eyes move signal
then Eigenvalues Decomposition is carried out to C, namely
In formula (2), U
0representation feature vector,
representing matrix U
0transposed matrix, Σ is the diagonal matrix of eigenvalue, then prewhitening transformation matrix P can be expressed as
Utilize prewhitening transformation matrix P to above-mentioned 4 classification average covariance matrices
with
convert, can obtain
In formula (4), S
l, S
r, S
uand S
dbe respectively average covariance matrices
with
carry out the matrix after prewhitening conversion, to S
l, S
r, S
uand S
dafter carrying out Joint Diagonalization of Matrix, an orthogonal matrix U and approximate diagonal battle array Σ can be found
l, Σ
r, Σ
uwith Σ
d, and meet following relation
US
lU
T＝Σ
lUS
uU
T＝Σ
u
(5)
US
rU
T＝Σ
rUS
dU
T＝Σ
d
Σ
l+Σ
r+Σ
u+Σ
d＝I(6)
In formula (5) and formula (6), I is unit matrix; Eye is moved to many classification problems of signal, diagonal matrix Σ
l, Σ
r, Σ
uand Σ
dcan carry out eigenvalue selection according to following formula, suppose that classification number to be sorted is N, diagonal matrix is Σ, then
score(Σ)＝max(Σ,(1-Σ)/(1-Σ+(N-1)
2Σ))(7)
Calculate Σ respectively
l, Σ
r, Σ
uand Σ
dthe eigenvalue that middle score is the highest, and take out the highest eigenvalue of this score row vector U corresponding in orthogonal matrix U
l, U
r, U
uand U
d, then calculate corresponding spatial filter by following formula (8), namely
In formula (8), W
l, W
r, W
uand W
drepresent the spatial filter of left pan, right pan, upper pan and lower pan four class respectively.
In step 3, move in Signal analysis process eye, the many classification task correlated source signal (left and right, upper and lower pan four class eye moves the relevant source signal of task) obtained after airspace filter obtains by following formula (9)
Z
l＝W
lXZ
r＝W
rXZ
u＝W
uXZ
d＝W
dX(9)
In formula (9), X represents that original multi-lead eye moves signal, Z
l, Z
r, Z
u, Z
drepresent the result of left and right, upper and lower pan signal after airspace filter (i.e. many classification task correlated source signal) respectively, and the characteristic parameter that this result moves signal as eye is sent into SVM classifier, in order to train SVM model or to identify test data.
Fig. 1 is human eye internal anatomy.
See Fig. 2, describing the generating principle that eye in the present embodiment moves signal, is the electro-oculogram during pan of left and right.
See Fig. 3, the method in the present embodiment that describes is moved signal pre-processing module, spatial filter training module and eye primarily of eye and is moved Signal analysis module three part composition.Method utilizes support vector machine (SVM) to classify, to realize identification eye being moved to signal after adopting common space pattern (CSP) method to carry out feature extraction.Method mainly comprises following step: 1) gather EOG data and carry out pretreatment; 2) pretreated all data are divided into training data and test data two parts.For training data, adopt CSP algorithm to carry out airspace filter, calculate the spatial filter W of left pan, right pan, upper pan and lower pan four class
l, W
r, W
uand W
d; 3) use above-mentioned four wave filter to carry out airspace filter to training data, and result after filtering is input in SVM classifier as characteristic parameter carries out SVM model training; 4) use four spatial filters to carry out filtering, feature extraction equally to test data, be then sent in the SVM classifier trained and identify, obtain final recognition correct rate, and it is analyzed, the performance of appraisal procedure.
See Fig. 4, describe eye in the present embodiment and move distribution of electrodes in signal acquisition process.The collection of electro-ocular signal uses Ag/AgCl electrode.In order to the eye obtaining left and right, the upper and lower four direction of experimenter moves information and more spatial positional information, we employ 9 electrodes altogether, wherein, electrode V1 and electrode V2 to be placed on the left of experimenter 3cm and lower 3cm place on (or right side) eyeball, moves signal in order to gather vertical eye; Electrode H1 and electrode H2 to be placed on the left of experimenter's left eye 3cm place on the right side of 3cm and right eye respectively, moves signal in order to gather level eye; Electrode A 1 and electrode A 2 are placed in forehead position, in order to strengthen spatial information; Reference electrode C1 and C2 is positioned over the convex place of breast, the left and right sides respectively, and ground electrode G is positioned at center, the crown.
See Fig. 5, describe the relative position that eye in the present embodiment moves observed object and experimenter in signal acquisition process.In specific implementation process, experimenter is sitting on an armchair, and, about 2 meters, its front is provided with the observed object of upper and lower, left and right four direction respectively and distance experimenter's optic centre point (O) is all 1.5 meters.
See Fig. 6, describe the detailed process of single experiment normal form in the present embodiment.When testing beginning, first on screen, there is " beginning " character, and along with the sonic stimulation that a 20ms grows, after 1 second kind, experimenter can see red arrow prompting (being respectively: to upward arrow, downwards arrow, left arrow and arrow to the right) on screen, it is 3 seconds that arrow continues time of occurrence on screen, within this time, requirement of experiment experimenter rotates eyeball to arrow direction indication after seeing arrow, after seeing observation station, rotate back into central point, experimenter can not blink in this course.Afterwards, have 2 seconds takes a break, experimenter can blink, loosen.
See Fig. 7, describe in the present embodiment the original waveform figure that the eye using said method to collect moves signal.This waveform contains experimenter and upwards sweeps, sweeps downwards, sweep left and sweep 4 actions to the right.
See Fig. 8, describe by the spatial filter coefficient schematic diagram that CSP algorithm calculates in the present embodiment, its abscissa represents the position that filter coefficient is residing in this wave filter, and vertical coordinate represents the size of coefficient amplitude.Can find out, when experimenter upwards sweeps, in all filter coefficients, the 6th value is maximum; During downward pan, the 5th value is maximum; In like manner, when sweeping left, the 1st value is maximum; When sweeping to the right, the 2nd value is maximum.Thus, we can presumably (V1), under (V2), left (H2), right (H1) 4 electrodes correspond respectively to wave filter the 6th, 5,1, the coefficient of 2 positions.The above results describes 4 class eyes that this spatial filter can make us classify and moves signal and reach difference maximization.
Can find out, when experimenter sweeps left, in all filter coefficients, the 1st value is maximum in all leading; When sweeping to the right, the 2nd value is maximum, and in like manner, when upwards sweeping, the 6th value is maximum; During downward pan, the 5th value is maximum, thus, we can presumably (V1), under (V2), left (H2), right (H1) 4 electrodes correspond respectively to wave filter the 6th, 5,1, the coefficient of 2 positions.Therefore, the 4 class eyes that this spatial filter can make us classify move signal and reach difference maximization.
Claims (6)
1. one kind is moved signal recognition system based on the eye of common space pattern, it is characterized in that, comprise electro-ocular signal acquisition electrode, multiple armchairs, multiple observed object, screen, eye move signal pre-processing module, spatial filter training module, eye move Signal analysis module;
Electro-ocular signal acquisition electrode uses Ag/AgCl electrode;
The eye using 9 described Ag/AgCl electrodes to obtain left and right, the upper and lower four direction of experimenter moves information and spatial positional information, wherein: electrode V1 and electrode V2 is placed on the left of experimenter or 3cm and lower 3cm place on the eyeball of right side, moves signal in order to gather vertical eye; Electrode H1 and electrode H2 to be placed on the left of experimenter's left eye 3cm place on the right side of 3cm and right eye respectively, moves signal in order to gather level eye; Electrode A 1 and electrode A 2 are placed in experimenter's forehead position, in order to strengthen spatial information; Reference electrode C1 and C2 is positioned over the convex place of breast, the experimenter left and right sides respectively; Ground electrode G is positioned at center, the experimenter crown;
Multiple experimenter is respectively sitting on an armchair, and, about 2 meters, its front is provided with the observed object of upper and lower, left and right four direction respectively, and observed object distance experimenter's optic centre point O is all 1.5 meters;
Screen is for display " beginning " character and red prompting arrow: " beginning " character is along with the sonic stimulation of a 20ms length, then experimenter can see the redness prompting arrow shown successively on screen, red arrow is respectively to upward arrow, downwards arrow, left arrow and arrow to the right, it is 3 seconds that arrow continues time of occurrence on screen, within this time, experimenter rotates eyeball to arrow direction indication after seeing arrow, rotates back into the central point of screen after seeing observation station;
Described eye moves signal pre-processing module, is used for carrying out bandpass filtering to all training and test data and removing averaging operation, in order to reduce different noise signal original multi-lead eye is moved to the interference of signal, thus improves recognition correct rate;
Described spatial filter training module, moves signal space wave filter for the many classification eye built based on common space pattern by the method for Joint Diagonalization of Matrix;
Described eye moves Signal analysis module, for original multi-lead eye being moved result that signal obtains after above-mentioned spatial filter filtering as characteristic parameter, and use support vector machine to carry out identification that eye moves signal.
2. move a recognition methods for signal recognition system based on the eye of common space pattern, employ a kind of eye based on common space pattern according to claim 1 and move signal recognition system, it is characterized in that, comprise the steps:
Step 1: gather the eye movement data based on electro-oculogram and pretreatment is carried out to eye movement data;
Step 2: pretreated all data are divided into training data and test data two parts; Described training data is that original multi-lead eye moves signal, adopts CSP algorithm to move signal to original multi-lead eye and carries out airspace filter, calculates the spatial filter W of left pan, right pan, upper pan and lower pan four class
l, W
r, W
uand W
d;
Step 3: use the W in above-mentioned steps 2
l, W
r, W
uand W
dfour spatial filters carry out airspace filter to training data, and result after filtering are input in SVM classifier as characteristic parameter and train;
Step 4: to test data, the same W used in above-mentioned steps 2
l, W
r, W
uand W
dfour spatial filters carry out airspace filter, feature extraction, are then sent in the SVM classifier trained and identify, finally obtain the recognition result that eye moves signal.
3. a kind of eye based on common space pattern according to claim 2 moves the recognition methods of signal recognition system, it is characterized in that, in described step 1 to eye move signal carry out preprocessing process comprise bandpass filtering step with go average step.
4. a kind of eye based on common space pattern according to claim 3 moves the recognition methods of signal recognition system, it is characterized in that, is 0.3Hz ~ 12.5Hz for carrying out the cut-off frequency of the band filter of bandpass filtering step.
5. a kind of eye based on common space pattern according to claim 2 moves the recognition methods of signal recognition system, it is characterized in that, adopt in described step 2 CSP algorithm to original multi-lead eye move signal carry out airspace filter time, spatial filter W
l, W
r, W
uand W
dcomputational process be:
Assuming that X
l, X
r, X
uwith X
drepresent that the original multi-lead eye under left pan, right pan, upper pan and lower pan 4 kinds of tasks moves signal, X respectively
l, X
r, X
uwith X
dspace covariance matrix after normalization is
The covariance matrix of the many groups training data under same eye movement mode is averaged, obtains average covariance matrices
with
calculate the average covariance matrices sum that all eyes move signal
then Eigenvalues Decomposition is carried out to C, namely
In formula (2), U
0representation feature vector,
representing matrix U
0transposed matrix, Σ is the diagonal matrix of eigenvalue, then prewhitening transformation matrix P can be expressed as
Utilize prewhitening transformation matrix P to above-mentioned 4 classification average covariance matrices
with
convert, can obtain
In formula (4), S
l, S
r, S
uand S
dbe respectively average covariance matrices
with
carry out the matrix after prewhitening conversion, to S
l, S
r, S
uand S
dafter carrying out Joint Diagonalization of Matrix, an orthogonal matrix U and approximate diagonal battle array Σ can be found
l, Σ
r, Σ
uwith Σ
d, and meet following relation:
Σ
l+Σ
r+Σ
u+Σ
d＝I(6)；
In formula (5) and formula (6), I is unit matrix; Eye is moved to many classification problems of signal, diagonal matrix Σ
l, Σ
r, Σ
uand Σ
dcan carry out eigenvalue selection according to following formula, suppose that classification number to be sorted is N, diagonal matrix is Σ, then
score(Σ)＝max(Σ,(1-Σ)/(1-Σ+(N-1)
2Σ))(7)；
Calculate Σ respectively
l, Σ
r, Σ
uand Σ
dthe eigenvalue that middle score is the highest, and take out the highest eigenvalue of this score row vector U corresponding in orthogonal matrix U
l, U
r, U
uand U
d, then calculate corresponding spatial filter by following formula (8), namely
In formula (8), W
l, W
r, W
uand W
drepresent the spatial filter of left pan, right pan, upper pan and lower pan four class respectively.
6. a kind of eye based on common space pattern according to claim 5 moves the recognition methods of signal recognition system, it is characterized in that, in step 3, move in Signal analysis process eye, the source signal that left and right, the upper and lower pan four class eye obtained after airspace filter moves task relevant obtains by following formula (9)
Z
l＝W
lXZ
r＝W
rXZ
u＝W
uXZ
d＝W
dX(9)；
In formula (9), X represents that original multi-lead eye moves signal, Z
l, Z
r, Z
u, Z
drepresent the result of left and right, upper and lower pan signal after airspace filter respectively, and the characteristic parameter that this result moves signal as eye is sent into SVM classifier, in order to train SVM model or to identify test data.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410156043.9A CN103892829B (en) | 2014-04-17 | 2014-04-17 | Eye movement signal identification system based on common spatial mode and identification method thereof |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410156043.9A CN103892829B (en) | 2014-04-17 | 2014-04-17 | Eye movement signal identification system based on common spatial mode and identification method thereof |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN103892829A CN103892829A (en) | 2014-07-02 |
| CN103892829B true CN103892829B (en) | 2016-04-27 |
Family
ID=50984645
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201410156043.9A Expired - Fee Related CN103892829B (en) | 2014-04-17 | 2014-04-17 | Eye movement signal identification system based on common spatial mode and identification method thereof |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN103892829B (en) |
Families Citing this family (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105640500A (en) * | 2015-12-21 | 2016-06-08 | 安徽大学 | Independent component analysis-based saccade signal feature extraction method and recognition method |
| CN105447475A (en) * | 2015-12-21 | 2016-03-30 | 安徽大学 | Independent component analysis-based sweep signal sample optimization method |
| EP3417775B1 (en) * | 2017-06-22 | 2020-08-19 | Oticon A/s | A system for capturing electrooculography signals |
| CN107480716B (en) * | 2017-08-15 | 2021-01-29 | 安徽大学 | Method and system for identifying saccade signal by combining EOG and video |
| CN107348958B (en) * | 2017-08-15 | 2019-12-24 | 安徽大学 | Robust glance EOG signal identification method and system |
| CN108491792B (en) * | 2018-03-21 | 2022-07-12 | 安徽大学 | Office scene human-computer interaction behavior recognition method based on electro-oculogram signals |
| CN109144238B (en) * | 2018-05-14 | 2021-09-28 | 孙佳楠 | Human-computer interaction system based on electro-oculogram coding and interaction method thereof |
| CN108693217B (en) * | 2018-08-13 | 2020-09-08 | 上海市宝山区中西医结合医院 | Clinical eye resting potential measuring system |
| CN110298303B (en) * | 2019-06-27 | 2022-03-25 | 西北工业大学 | Crowd identification method based on long-time memory network glance path learning |
| CN112036229B (en) * | 2020-06-24 | 2024-04-19 | 宿州小马电子商务有限公司 | Intelligent bassinet electroencephalogram signal channel configuration method with demand sensing function |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101219048A (en) * | 2008-01-25 | 2008-07-16 | 北京工业大学 | Method for extracting brain electrical character of imagine movement of single side podosoma |
| CN101339413A (en) * | 2008-08-07 | 2009-01-07 | 北京师范大学 | Switching control method based on brain electric activity human face recognition specific wave |
| CN101599127A (en) * | 2009-06-26 | 2009-12-09 | 安徽大学 | The feature extraction of electro-ocular signal and recognition methods |
| CN101897640A (en) * | 2010-08-10 | 2010-12-01 | 北京师范大学 | Novel movement imagery electroencephalogram control-based intelligent wheelchair system |
| CN103699216A (en) * | 2013-11-18 | 2014-04-02 | 南昌大学 | Email communication system and method based on motor imagery and visual attention mixed brain-computer interface |
-
2014
- 2014-04-17 CN CN201410156043.9A patent/CN103892829B/en not_active Expired - Fee Related
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101219048A (en) * | 2008-01-25 | 2008-07-16 | 北京工业大学 | Method for extracting brain electrical character of imagine movement of single side podosoma |
| CN101339413A (en) * | 2008-08-07 | 2009-01-07 | 北京师范大学 | Switching control method based on brain electric activity human face recognition specific wave |
| CN101599127A (en) * | 2009-06-26 | 2009-12-09 | 安徽大学 | The feature extraction of electro-ocular signal and recognition methods |
| CN101897640A (en) * | 2010-08-10 | 2010-12-01 | 北京师范大学 | Novel movement imagery electroencephalogram control-based intelligent wheelchair system |
| CN103699216A (en) * | 2013-11-18 | 2014-04-02 | 南昌大学 | Email communication system and method based on motor imagery and visual attention mixed brain-computer interface |
Also Published As
| Publication number | Publication date |
|---|---|
| CN103892829A (en) | 2014-07-02 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN103892829B (en) | Eye movement signal identification system based on common spatial mode and identification method thereof | |
| Lee et al. | A brain-wave-actuated small robot car using ensemble empirical mode decomposition-based approach | |
| CN101711709B (en) | Method for controlling electrically powered artificial hands by utilizing electro-coulogram and electroencephalogram information | |
| CN102793540B (en) | Method for optimizing audio-visual cognitive event-related potential experimental paradigm | |
| CN102184019B (en) | Method for audio-visual combined stimulation of brain-computer interface based on covert attention | |
| CN105640500A (en) | Independent component analysis-based saccade signal feature extraction method and recognition method | |
| CN103793058A (en) | Method and device for classifying active brain-computer interaction system motor imagery tasks | |
| CN105447475A (en) | Independent component analysis-based sweep signal sample optimization method | |
| Singla et al. | Comparison of ssvep signal classification techniques using svm and ann models for bci applications | |
| CN105942974A (en) | Sleep analysis method and system based on low frequency electroencephalogram | |
| CN103610447A (en) | Mental workload online detection method based on forehead electroencephalogram signals | |
| CN112732090B (en) | Muscle cooperation-based user-independent real-time gesture recognition method | |
| US20220265218A1 (en) | Real-time evaluation method and evaluation system for group emotion homogeneity | |
| CN107480716A (en) | Method and system for identifying saccade signal by combining EOG and video | |
| CN106484106A (en) | The non-attention event related potential brain-machine interface method of visual acuity automatic identification | |
| CN106491129A (en) | A kind of Human bodys' response system and method based on EOG | |
| CN107411738A (en) | A kind of mood based on resting electroencephalogramidentification similitude is across individual discrimination method | |
| CN109144238A (en) | A kind of man-machine interactive system and its exchange method based on eye electricity coding | |
| CN107480635B (en) | Glance signal identification method and system based on bimodal classification model fusion | |
| Daud et al. | Time frequency analysis of electrooculograph (EOG) signal of eye movement potentials based on wavelet energy distribution | |
| Singla et al. | BCI based wheelchair control using steady state visual evoked potentials and support vector machines | |
| CN107348958A (en) | Robust glance EOG signal identification method and system | |
| CN204759349U (en) | Aircraft controlling means based on stable state vision evoked potential | |
| Lv et al. | An ICA-based spatial filtering approach to saccadic EOG signal recognition | |
| CN201308487Y (en) | Amblyopia detecting system based on P-VEP |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C14 | Grant of patent or utility model | ||
| GR01 | Patent grant | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20160427 |
|
| CF01 | Termination of patent right due to non-payment of annual fee |