CN112998646A - Multi-channel surface electromyography action intention recognition method without feature description - Google Patents
Multi-channel surface electromyography action intention recognition method without feature description Download PDFInfo
- Publication number
- CN112998646A CN112998646A CN201911324955.1A CN201911324955A CN112998646A CN 112998646 A CN112998646 A CN 112998646A CN 201911324955 A CN201911324955 A CN 201911324955A CN 112998646 A CN112998646 A CN 112998646A
- Authority
- CN
- China
- Prior art keywords
- matrix
- action
- data
- channel
- action intention
- 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
Images
Classifications
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
-
- 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
Landscapes
- Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Engineering & Computer Science (AREA)
- Surgery (AREA)
- General Health & Medical Sciences (AREA)
- Biophysics (AREA)
- Biomedical Technology (AREA)
- Heart & Thoracic Surgery (AREA)
- Medical Informatics (AREA)
- Molecular Biology (AREA)
- Physics & Mathematics (AREA)
- Animal Behavior & Ethology (AREA)
- Pathology (AREA)
- Public Health (AREA)
- Veterinary Medicine (AREA)
- Artificial Intelligence (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Physiology (AREA)
- Psychiatry (AREA)
- Signal Processing (AREA)
- Measurement And Recording Of Electrical Phenomena And Electrical Characteristics Of The Living Body (AREA)
Abstract
The invention discloses a multi-channel surface electromyography action intention recognition method without feature description, which comprises the following steps: 1. and calculating a multidimensional vector carrying P pieces of surface electromyography data with known action intentions, and determining a domain value of each action to serve as a standard for screening the action intentions of the surface electromyography data. 2. The obtained surface electromyogram data needing action intention identification is compared and multiplied with a matrix H consisting of P multidimensional vectors of known actions to obtain a contrast value matrix Vj. 3. And performing matrix subtraction operation on the screening matrix F and the comparison value matrix to obtain a deviation value matrix, extracting the maximum value element in the deviation value matrix and determining the standard action intention corresponding to the maximum value as the action intention represented by the surface electromyogram data. The invention utilizes the multi-channel surface electromyography without feature description to identify the action intention, can effectively avoid the problem that the identification result is influenced by inaccurate feature description of the surface electromyography, thereby being capable of identifying the action intentionThe accuracy and the reliability of action intention recognition are improved. The channel position of the invention can not be required accurately, and the action intention can be effectively identified; the recognition algorithm of the invention can effectively recognize the intention of each limb action of the human body.
Description
Technical Field
The invention relates to the technical field of multi-channel surface electromyogram signal action intention recognition, in particular to a recognition technology which does not need to describe the characteristics of surface electromyograms of limb movement.
Background
The limb movement of the human body is realized by single joint movement or multi-joint movement, and the joint movement comprises sliding movement, flexion and extension, horizontal flexion and extension, folding and unfolding, convolution, rotation and the like; these complex movements are often accomplished by the coordinated participation of multiple muscles. Therefore, the common meaning of the multi-channel electromyographic signals is accurately identified, and the intention of the human limb actions can be expressed more completely.
The current electromyographic signal action recognition technology needs to accurately describe the surface electromyographic characteristics obtained at a specific position of a human body corresponding to a certain known movement intention. It is stated above that the joint movement of the limbs of the human body is often accomplished by the cooperation of a plurality of muscles, and not only the muscle contraction has the meaning of kinematics, but also the relaxed muscle has the meaning of kinematics. Focusing only on the myoelectric signal of muscle contraction does not fully express the intention of exercise.
Therefore, there is a need for a method of identifying an intention commonly expressed by the surface electromyography of muscle groups involved in joint movement, without the need for specific characterization of the surface electromyography of limb movements.
Disclosure of Invention
The invention aims to solve the problems in the prior art and provides a multi-channel surface electromyographic action intention recognition method without characteristic description, which utilizes a covariance matrix method, a Gauss-Jordan elimination method and a linear discriminant method to recognize the multi-channel surface electromyographic action intention without characteristic description, can compare multi-channel surface electromyographic data and recognize action intention under various conditions, and avoids the situation that muscles related to the same receive the same surface electromyographic data (such as a contraction electromyographic signal or a relaxation electromyographic signal) when the joints of the limbs of a human body do different actions to influence the accuracy and reliability of action intention recognition, thereby improving the accuracy and reliability of action intention recognition.
In order to achieve the purpose, the invention provides a multi-channel surface electromyogram action intention recognition method without feature description, which utilizes a covariance matrix method and a linear discriminant analysis method to recognize action intentions expressed by effective multi-channel surface electromyograms. The method for recognizing the electromyographic action intention of the multi-channel surface without the characteristic description is carried out as follows.
Aiming at the multi-channel surface electromyography action intention recognition method without the characteristic description, the step 1 is carried out as follows.
1.1, expressing the recorded P effective surface electromyographic data of the known human body limb actions in a matrix form as shown in the following.
In the formula
The elements represent channels, each column represents surface electromyography data within the same channel, each of each rowAn element represents a channel.
Step 1.2, the active surface electromyography data for each action consists of f channels (f =0,1, …, n-1), each channel containing g different attribute values (g =0,1, …, m-1).
Step 1.3, when the joints of the limbs of the human body move, f channels are located at muscle groups related to the limbs, the positions of the channels are not required to be accurate, and k groups of f channel effective surface electromyographic data form a k row and (f x g) column matrix (k =0,1, …, i-1).
Step 1.4, according to the covariance matrix formula
Calculating a covariance matrix of the effective surface electromyography data of the P actions; and the covariance matrices of the valid surface electromyography data of the P actions are accumulated.
The addition of the covariance matrix for calculating the effective surface electromyography data for P actions described in step 1.4 is performed as follows.
Step 1.4.1, set up the column vector
K sets of surface electromyography data representing g different attribute values for each channel over a finite time, such that each action contains k sets of surface electromyography data for g different attribute values for each channel over a finite time
。
Step 1.4.2, use
Calculating the average value of the surface electromyogram data of each action k group to obtain the average value of the surface electromyogram data as
，
Represents the average value of the surface electromyogram data of the (n-1) th channel,
containing the mean of g different attribute values, i.e.
。
Step 1.4.3, for the k groups of surface electromyography data, use
Calculating the difference value between the k groups of surface electromyographic data and the average value of the corresponding channel inner surface electromyographic data, wherein the formula of the calculation result is
。
Step 1.4.4, transpose operation is carried out on the matrix formed by the difference data to obtain a matrix after the difference data is transposed
I.e. by
。
Step 1.4.5, using the formula
The covariance matrix of the k sets of surface electromyography data for each action is calculated as follows.
And step 1.4.6, calculating the accumulated covariance matrix S by using matrix accumulation operation, wherein a calculation formula is as follows.
Step 1.5, calculating the inverse S of the covariance matrix after the covariance matrix accumulation of the effective surface electromyographic data of the P actions according to a Gauss-Jordan elimination method-1The formula is as follows.
Wherein E represents an identity matrix.
And step 1.6, calculating a one-dimensional screening matrix F consisting of the screening threshold values of the P action intents by using a linear discriminant analysis method.
Aiming at the matrix F formed by the calculation and screening threshold values in the step 1.6, the method is carried out as follows.
Step 1.6.1, a matrix H consisting of P multidimensional vectors is calculated according to the following formula.
Using the inverse S of the covariance matrix-1Multiply the average value of the attribute values in each channel in each action to obtain a multidimensional vector expression of
(ii) a The matrix H formed by the multidimensional vectors has the expression as follows.
In the formula S-1The inverse of the cumulative covariance matrix is represented.
And 1.6.2, calculating the screening threshold values of the P actions according to the following formula and expressing the screening threshold values in a matrix form.
In the formula (I), the compound is shown in the specification,
each channel containing g different attributesThe average values of the values form a one-dimensional matrix, and H represents a matrix formed by multi-dimensional vectors.
The invention relates to a multi-channel surface electromyography action intention recognition method without feature description, wherein the step 2 is carried out as follows.
And 2.1, setting Y to represent each group of effective surface electromyographic data to be identified by the action intention.
Step 2.2, calculating a matrix V formed by contrast values of the effective surface electromyography data according to the following formulaj。
Wherein H represents a matrix of multidimensional vectors.
The multi-channel surface electromyography action intention recognition method without the characteristic description is characterized in that the step 3 is carried out as follows.
Step 3.1, calculate the matrix V composed of the contrast values according to the following formulajAnd the deviation value of the screening domain value matrix F.
And 3.2, extracting the maximum value of the elements in the deviation matrix according to the following formula.
And 3.3, judging which action intention is specifically expressed by the electromyographic data according to the standard action intention corresponding to the maximum value.
The invention has the beneficial effects.
The invention utilizes the multi-channel surface electromyography without feature description to identify the action intention, can effectively avoid the problem that the identification result is influenced due to inaccurate feature description of the surface electromyography, and can improve the accuracy and reliability of the action intention identification.
The method comprises the steps of obtaining multi-dimensional vectors and screening domain values of the electromyographic data of the known P action intentions through mathematical operations such as covariance matrix operation, a Gauss-Jordan elimination method, linear discriminant analysis and the like; comparing and multiplying myoelectric data to be identified by action intentions with multidimensional vectors of known action intentions to obtain a comparison value; the action intention is judged by using the deviation value of the comparison value and the screening threshold value, so that the accuracy and reliability of action intention identification can be improved.
The features and advantages of the present invention will be described in detail by embodiments in conjunction with the accompanying drawings.
Drawings
FIG. 1 is an overall workflow diagram of the multi-channel surface electromyographic action intention recognition method without feature description of the invention.
FIG. 2 is 46 groups of valid surface electromyography data for action 0 out of 6 known actions according to the multi-channel surface electromyography action intention recognition method without feature description of the invention.
FIG. 3 is 46 sets of valid surface electromyography data for action 1 out of 6 known actions according to the multi-channel surface electromyography action intent recognition method without characterization of the present invention.
FIG. 4 is 46 sets of valid surface electromyography data for action 2 out of 6 known actions according to the multi-channel surface electromyography action intent recognition method without characterization of the present invention.
FIG. 5 is 46 sets of valid surface electromyography data for action 3 of the known 6 actions according to the multi-channel surface electromyography action intent recognition method without characterization of the present invention.
FIG. 6 is 46 sets of valid surface electromyography data for action 4 out of 6 known actions according to the multi-channel surface electromyography action intent recognition method without characterization of the present invention.
FIG. 7 is 46 sets of valid surface electromyography data for action 5 out of 6 known actions according to the multi-channel surface electromyography action intent recognition method without characterization of the present invention.
FIG. 8 is the average of 46 sets of valid surface electromyography data for each of the known 6 actions according to the multi-channel surface electromyography action intent recognition method without characterization of the present invention.
FIG. 9 is a covariance matrix obtained by adding 6 action intention covariance matrices according to the multi-channel surface electromyography action intention recognition method without feature description of the present invention.
FIG. 10 is the inverse of the cumulative covariance matrix as described in the featureless multi-channel surface electromyographic action intent recognition method of the present invention.
FIG. 11 is a multi-dimensional vector of valid electromyographic data for each of the known 6 actions according to the multi-channel surface electromyographic action intention recognition method without characterization of the present invention.
FIG. 12 is a domain value of valid electromyography data for each action intention of known 6 actions according to the multi-channel surface electromyography action intention recognition method without feature description of the present invention.
FIG. 13 is a set of effective electromyographic data to be recognized by action intention according to the multi-channel surface electromyographic action intention recognition method without feature description of the invention.
FIG. 14 is a difference value between a contrast value of effective electromyographic data to be recognized for action intention according to the multi-channel surface electromyographic action intention recognition method without feature description of the present invention and a threshold value in FIG. 9.
FIG. 15 is the recognition result of the action intention expressed by effective electromyography data by the multi-channel surface electromyography action intention recognition method without feature description of the invention.
FIG. 16 is the action intention recognition of the surface electromyogram data of the sample 1 by the multi-channel surface electromyogram action intention recognition method without the characteristic description of the invention.
FIG. 17 is the action intention recognition of the surface electromyogram data of the sample 2 by the multi-channel surface electromyogram action intention recognition method without the characteristic description of the invention.
Detailed Description
In the embodiment, a multi-channel surface electromyogram action intention recognition method without feature description is disclosed, wherein an overall flow working diagram is shown in fig. 1, 1. a multi-dimensional vector carrying P pieces of surface electromyogram data with known action intentions is calculated, and a domain value of each action is determined, so that the multi-dimensional vector is used as a standard for carrying out action intention screening on the surface electromyogram data. 2. For the acquired surface electromyogram data needing action intention identification and P known actionsMatrix V composed of contrast values is obtained by carrying out comparison multiplication on matrix H composed of multidimensional vectorsj. 3. And performing matrix subtraction operation on the screening matrix F and the comparison value matrix to obtain a deviation value matrix, extracting the maximum value element in the deviation value matrix and determining the standard action intention corresponding to the maximum value as the action intention represented by the surface electromyogram data.
The procedure for step 1 was as follows.
1.1, expressing the recorded P effective surface electromyographic data of the known human body limb actions in a matrix form as shown in the following.
In the formula
The elements represent channels, each column represents surface electromyography data in the same channel, and each element in each row represents a channel; the specific implementation process records 6 different actions of the palm, wherein the 6 different actions are action 0, action 1, action 2, action 3, action 4 and action 5.
Step 1.2, the effective surface electromyography data of each action is composed of f channels (f =0,1, …, n-1), each channel containing g different attribute values (g =0,1, …, m-1); in the specific implementation process, 6 channels of 6 different actions of the palm are selected, and each channel contains surface electromyogram data of 4 different attribute values.
Step 1.3, when the joints of the limbs of the human body move, f channels are located at muscle groups related to the limbs, the positions of the channels are not required to be accurate, and the effective surface electromyographic data of k groups of the f channels is expressed in a mode of k rows and (f x g) column matrixes (k =0,1, …, i-1)
(P =0,1, …, r-1), wherein
. In the specific implementation process, 46 groups of effective surface electromyography data of each action of 6 actions are taken, and 46 groups of surface electromyography data, each of which comprises 4 different attribute values, of 6 channels of each action in the 6 actions are obtained, and as shown in fig. 2 to 7, the data sequentially represent action 0, action 1, action 2, action 3, action 4 and action 5.
Step 1.4, according to the covariance matrix formula
Calculating a covariance matrix of the effective surface electromyography data of the P actions; and the covariance matrices of the valid surface electromyography data of the P actions are accumulated.
The procedure for step 1.4 is as follows.
Step 1.4.1, set up the column vector
K sets of surface electromyography data representing g different attribute values for each channel over a finite time, such that each action contains k sets of surface electromyography data for g different attribute values for each channel over a finite time
. In the specific implementation process, the column vector of 6 channels
，
，
，
，
，
；
Representing that each channel contains 4 different attribute values.
Step 1.4.2, use
Calculating the average value of the surface electromyogram data of each action k group to obtain the average value of the surface electromyogram data as
，
Represents the average value of the surface electromyogram data of the (n-1) th channel,
containing the mean of g different attribute values, i.e.
. In the specific implementation process, the average value of 46 groups of surface electromyography data
I.e. each of the 6 channels contains an average of 4 different attribute values of.
The mean data for each of the 6 channels containing 4 different attribute values is shown in fig. 8.
Step 1.4.3, for the k groups of surface electromyography data, use
Calculating the electromyographic data of k groups of surfaces and the electromyographic data of the surfaces of corresponding channelsAccording to the difference value of the average values, the calculation result is formulated as follows.
In a specific implementation process, a matrix composed of difference values of the myoelectric data of 46 groups of surface myoelectric data containing 4 different attribute values in each channel and the mean value of the myoelectric data in the corresponding channel in 6 channels in a limited time is as follows.
Step 1.4.4, transpose operation is carried out on the matrix formed by the difference data to obtain a matrix after the difference data is transposed
Is as follows.
In a specific implementation process, the result of transposing the difference data matrix in step 1.5 is as follows.
Step 1.4.5, using the formula
The covariance matrix of the k sets of surface electromyography data for each action is calculated as follows.
In the implementation, the covariance matrix of the 46 sets of surface electromyography data for each of the 6 actions is as follows.
Each action is a 24 row and 24 column covariance matrix of 46 sets of surface electromyography data for a finite time of 6 channels, each channel containing 4 different attribute values.
Step 1.4.6, the covariance matrices of the P actions are accumulated using the following formula to obtain a new covariance matrix S.
In a specific implementation process, the 6 covariance matrices obtained in step 1.4.5 are accumulated to obtain a new covariance matrix S as shown in fig. 9.
Step 1.5, calculating the inverse S of the covariance matrix after the covariance matrix accumulation of the effective surface electromyographic data of the P actions according to a Gauss-Jordan elimination method-1The formula is as follows.
Wherein E represents an identity matrix.
In the specific implementation process, the inverse matrix S of the covariance matrix is obtained after Gauss-Jordan elimination-1As shown in fig. 10.
And step 1.6, calculating a one-dimensional screening matrix F consisting of the screening threshold values of the P action intents by using a linear discriminant analysis method. The procedure for step 1.6 is as follows.
Step 1.6.1, a matrix H consisting of P multidimensional vectors is calculated according to the following formula.
Using the inverse S of the covariance matrix-1And performing multiplication operation on the average value of the attribute values in each channel in each action to obtain a multidimensional vector expression.
The matrix H formed by the multidimensional vectors has the expression as follows.
In the formula S-1The inverse of the cumulative covariance matrix is represented.
In the implementation, 24-dimensional vectors of 6 actions form a matrix with 24 rows and 6 columns, and the data is shown in fig. 11.
And 1.6.2, calculating the screening threshold values of the P actions according to the following formula and expressing the screening threshold values in a matrix form.
In the formula (I), the compound is shown in the specification,
representing a one-dimensional matrix formed by the average values of different types of attribute values in each channel of each action, wherein H represents a matrix formed by multi-dimensional vectors; in the specific implementation process, the domain values of the 6 actions are represented in a matrix form as shown in fig. 12 by multiplying the average value of the attribute values in each channel in each action in the 6 actions by a matrix composed of 6 multidimensional vectors, and the domain values are used as the basis for identifying the following action intentions.
The procedure for step 2 was as follows.
And 2.1, setting Y to represent each group of effective surface electromyographic data to be identified by the action intention. In the specific implementation process, 10 groups of effective surface electromyography data to be identified by action intentions are taken, wherein 1 group of surface electromyography data is taken as an action intention identification example, and 1 group of effective electromyography data to be identified by action intentions is shown in fig. 13.
Step 2.2, calculating a matrix V consisting of contrast values by carrying out contrast multiplication on a matrix H consisting of multidimensional vectors of the effective surface electromyographic data and the known surface electromyographic data according to the following formulaj。
Wherein H represents a matrix composed of multidimensional vectors; in a specific implementation process, a matrix with 1 row and 24 columns formed by effective surface electromyography data to be identified by action intentions in each group is multiplied by a matrix with 24 rows and 6 columns formed by multidimensional vectors of effective surface electromyography data with 6 known action intentions, and a contrast value matrix with 1 row and 6 columns is obtained as a result.
The procedure for step 3 was as follows.
Step 3.1, calculate the contrast value matrix V according to the following formulajThe deviation value of the filtered domain value matrix F is obtained to obtain a new 1-row 6-column matrix Lj。
In a specific implementation process, the 1-row 6-column contrast matrix and the 1-row 6-column domain value matrix are subjected to subtraction operation, and a 1-row 6-column deviation matrix is obtained as shown in fig. 14.
And 3.2, extracting the maximum value of the elements in the deviation matrix according to the following formula.
And 3.3, judging which action intention is specifically expressed by the surface electromyography data according to the standard action intention corresponding to the maximum value. In the specific implementation process, the maximum value element of the 6 elements of the matrix is extracted, and it is determined to which of the known 6 actions the maximum value element belongs, and the result is shown in fig. 15.
The action intention recognition of multiple groups of surface myoelectricity is carried out according to the action intention recognition example implementation process, and 2 groups of surface myoelectricity data action intention recognition are taken as samples, as shown in sample 1 in fig. 16 and sample 2 in fig. 17.
The above embodiments are illustrative of the present invention, and are not intended to limit the present invention, and any simple modifications of the present invention are within the scope of the present invention.
Claims (4)
1. A multi-channel surface electromyography action intention recognition method without feature description is characterized by comprising the following steps:
step 1, expressing surface electromyogram data carrying P known action intentions in a matrix manner
(P =0,1, …, r-1); then, obtaining a matrix H consisting of P multidimensional vectors through mathematical operations such as covariance matrix operation, a Gauss-Jordan elimination method, linear discriminant analysis and the like, and calculating a one-dimensional screening matrix F consisting of P domain values according to the matrix H to serve as a standard for screening action intentions of the electromyographic signals;
step 2, representing the acquired surface electromyographic data needing action intention identification in a one-dimensional matrix form, and then performing comparison multiplication operation with a matrix H consisting of P multidimensional vectors in the step 1 to obtain a one-dimensional matrix V consisting of P contrast valuesj；
Step 3, taking the one-dimensional screening matrix F in the step 1 as a screening standard, and carrying out comparison value matrix V calculated in the step 2jAnd performing difference operation with the screening standard, and taking the known action intention corresponding to the maximum value to judge which action intention the surface electromyogram data specifically expresses.
2. The featureless multi-channel surface electromyographic action intention recognition method of claim 1, wherein the step 1 is performed according to the following steps:
step 1.1, the recorded effective surface electromyography data of the known human body limb actions is expressed in a matrix form:
in the formula
The elements represent channels, each column represents surface electromyography data in the same channel, and each element in each row represents a channel;
step 1.2, the effective surface electromyography data is composed of f channels (f =0,1, …, n-1), and each channel contains g different attribute values (g =0,1, …, m-1);
step 1.3, when the joints of the limbs of the human body move, f channels are located at muscle groups related to the limbs, the positions of the channels are not required to be accurate, and k groups of f channel effective surface electromyographic data form a matrix (k =0,1, …, i-1) with k rows and (f × g) columns;
step 1.4, according to the covariance matrix formula
Calculating covariance matrixes of the effective surface electromyographic data of the P actions, and accumulating the covariance matrixes of the effective surface electromyographic data of the P actions, wherein an accumulation formula (1) is as follows:
step 1.5, calculating the inverse S of the covariance matrix after the covariance matrix accumulation of the effective surface electromyographic data of the P actions according to a Gauss-Jordan elimination method-1The calculation formula (2) is as follows:
wherein E represents an identity matrix;
step 1.6, calculating a one-dimensional screening matrix F consisting of screening threshold values of P action intents by using a linear discriminant analysis method, and performing the following steps:
step 1.6.1, calculating a matrix H consisting of P multidimensional vectors according to the following formula (3):
the multidimensional vector expression is:
the matrix H formed by the multidimensional vectors has the expression:
in the formula S-1Representing the inverse of the cumulative covariance matrix;
step 1.6.2 the screening threshold values for P actions are calculated and expressed in matrix form according to the following formula (4):
in the formula (I), the compound is shown in the specification,
and H represents a matrix formed by multidimensional vectors.
3. The featureless multi-channel surface electromyographic action intention recognition method of claim 1, the step 2 performed by:
step 2.1, setting Y to represent each group of effective surface electromyographic data to be identified by the action intention;
step 2.2, calculating a matrix V formed by contrast values of the effective surface electromyography data according to the following formula (5)j：
Where H represents a matrix of multidimensional vectors.
4. The featureless multi-channel surface electromyographic action intention recognition method as claimed in claim 1, wherein step 3 is performed by:
step 3.1, calculate the contrast value matrix V according to the following formula (6)jDeviation values from the filtered domain value matrix F:
step 3.2, extracting the maximum value of the elements in the deviation matrix according to the following formula (7)
And 3.3, judging which action intention is specifically expressed by the surface electromyography data according to the standard action intention corresponding to the maximum value.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201911324955.1A CN112998646B (en) | 2019-12-20 | 2019-12-20 | Multi-channel surface electromyography action intention recognition method without feature description |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201911324955.1A CN112998646B (en) | 2019-12-20 | 2019-12-20 | Multi-channel surface electromyography action intention recognition method without feature description |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN112998646A true CN112998646A (en) | 2021-06-22 |
| CN112998646B CN112998646B (en) | 2022-03-25 |
Family
ID=76382609
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201911324955.1A Active CN112998646B (en) | 2019-12-20 | 2019-12-20 | Multi-channel surface electromyography action intention recognition method without feature description |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112998646B (en) |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2014050590A (en) * | 2012-09-07 | 2014-03-20 | National Institute Of Advanced Industrial & Technology | Action recognition device for hand, artificial hand operation device, and composite sensor |
| WO2018050191A1 (en) * | 2016-09-14 | 2018-03-22 | Aalborg Universitet | A human intention detection system for motion assistance |
| CN108309290A (en) * | 2018-02-24 | 2018-07-24 | 华南理工大学 | The automatic removal method of Muscle artifacts in single channel EEG signals |
| CN110537913A (en) * | 2019-08-28 | 2019-12-06 | 华南理工大学 | LDA algorithm-based channel selection method for surface electromyographic signals |
-
2019
- 2019-12-20 CN CN201911324955.1A patent/CN112998646B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2014050590A (en) * | 2012-09-07 | 2014-03-20 | National Institute Of Advanced Industrial & Technology | Action recognition device for hand, artificial hand operation device, and composite sensor |
| WO2018050191A1 (en) * | 2016-09-14 | 2018-03-22 | Aalborg Universitet | A human intention detection system for motion assistance |
| CN108309290A (en) * | 2018-02-24 | 2018-07-24 | 华南理工大学 | The automatic removal method of Muscle artifacts in single channel EEG signals |
| CN110537913A (en) * | 2019-08-28 | 2019-12-06 | 华南理工大学 | LDA algorithm-based channel selection method for surface electromyographic signals |
Non-Patent Citations (2)
| Title |
|---|
| 丁其川等: "基于肌电信号容错分类的手部动作识别", 《机器人》 * |
| 李向新等: "一种下肢运动意图识别算法性能实时测评系统", 《仪器仪表学报》 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN112998646B (en) | 2022-03-25 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108714026B (en) | Fine-grained electrocardiosignal classification method based on deep convolutional neural network and online decision fusion | |
| Khushaba et al. | A framework of temporal-spatial descriptors-based feature extraction for improved myoelectric pattern recognition | |
| Sapsanis et al. | EMG based classification of basic hand movements based on time-frequency features | |
| Huang et al. | Surface EMG decoding for hand gestures based on spectrogram and CNN-LSTM | |
| CN109598219B (en) | Adaptive electrode registration method for robust electromyography control | |
| CN108154085B (en) | Method and device for identity recognition based on electrocardiogram data and electronic equipment | |
| Kang et al. | A method of denoising multi-channel EEG signals fast based on PCA and DEBSS algorithm | |
| CN103092971B (en) | A kind of sorting technique for brain-computer interface | |
| CN112426160A (en) | Electrocardiosignal type identification method and device | |
| CN113128552A (en) | Electroencephalogram emotion recognition method based on depth separable causal graph convolution network | |
| CN110638444A (en) | Cortical muscle coupling analysis method based on MEMD-rTVgPDC | |
| CN111631710B (en) | Method for eliminating myoelectric artifacts in state-related dynamic electroencephalogram signals | |
| Adochiei et al. | ECG waves and features extraction using Wavelet Multi-Resolution Analysis | |
| CN108647584A (en) | Cardiac arrhythmia method for identifying and classifying based on rarefaction representation and neural network | |
| CN112998646B (en) | Multi-channel surface electromyography action intention recognition method without feature description | |
| Azami et al. | Automatic signal segmentation based on singular spectrum analysis and imperialist competitive algorithm | |
| Serbest et al. | Classification of cardiac arrhythmias with artificial neural networks according to gender differences | |
| Miah et al. | Motor imagery classification using subband tangent space mapping | |
| CN112205990A (en) | Wrist angle prediction method and device based on sEMG under different loads | |
| Ebrahimzadeh et al. | An efficient technique for classification of electrocardiogram signals | |
| Zandi et al. | Quantitative evaluation of a wavelet-based method in ventricular late potential detection | |
| Nguyen et al. | A novel surface electromyographic gesture recognition using discrete cosine transform-based attention network | |
| Du et al. | The effect of combining stationary wavelet transform and independent component analysis in the multichannel SEMGs hand motion identification system | |
| CN106901731B (en) | Electromyographic signal feature extraction method fusing wavelet packet and bispectrum analysis | |
| Naik et al. | Subtle electromyographic pattern recognition for finger movements: A pilot study using BSS techniques |
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 | ||
| TA01 | Transfer of patent application right | ||
| TA01 | Transfer of patent application right |
Effective date of registration: 20210720 Address after: Room 405, building a, the third branch of leibaizhongcheng Life Science Park, 22 Jinxiu East Road, Jinsha community, Kengzi street, Pingshan District, Shenzhen, Guangdong 518000 Applicant after: Shenzhen wolikang biomedical Co.,Ltd. Address before: 450000 3rd floor, building 1, Hengfeng science and technology innovation center, Ruyun Road, Zhengzhou Airport Economic Comprehensive Experimental Zone, Zhengzhou City, Henan Province Applicant before: Henan Dushu Digital Technology Co.,Ltd. |
|
| GR01 | Patent grant | ||
| GR01 | Patent grant |