CN103345641A - Hand electromyographic signal motion recognition method based on wavelet entropy and support vector machine - Google Patents
Hand electromyographic signal motion recognition method based on wavelet entropy and support vector machine Download PDFInfo
- Publication number
- CN103345641A CN103345641A CN2013103000988A CN201310300098A CN103345641A CN 103345641 A CN103345641 A CN 103345641A CN 2013103000988 A CN2013103000988 A CN 2013103000988A CN 201310300098 A CN201310300098 A CN 201310300098A CN 103345641 A CN103345641 A CN 103345641A
- Authority
- CN
- China
- Prior art keywords
- wavelet
- wavelet packet
- electromyographic signal
- support vector
- vector machine
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Abstract
The invention provides a hand electromyographic signal motion recognition method based on wavelet entropy and a support vector machine. According to the hand electromyographic signal motion recognition method based on the wavelet entropy and the support vector machine, wavelet packet decomposition is carried out on a surface electromyographic signal, the wavelet packet coefficient matrix of each node is extracted, and the wavelet entropy is calculated by means of the energy of each sub-band; a feature vector is constructed with the wavelet entropy of the electromyographic signal used as the feature and then is input into a classifier of the support vector machine, the value of a parameter is gradually increased at fixed intervals to search an SVM classifier parameter value which enables the mode recognition rate to be highest, and a plurality of hand motions are classified. Due to the fact that wavelet packet conversion is an effective method used for analyzing signals with unstable characteristics and is capable of analyzing the signals accurately in different frequency ranges, providing rich mode information and inputting the feature vector extracted by three channels of sEMG signals into the support vector machine, six kinds of motion modes, namely the wrist stretching mode, the wrist bending mode, the fist stretching mode, the fist clenching mode, the outward turning mode and the inward turning mode, can be effectively recognized, and a recognition rate which is higher than that achieved with a traditional neural network is obtained.
Description
Technical field
The invention belongs to area of pattern recognition, relate to a kind of hand multi-locomotion mode recognition methods based on surface electromyogram signal.
Background technology
Surface electromyogram signal (Surface electromyography, sEMG) note by the surface myoelectric pickoff electrode from the human skeletal muscle surface, the bioelectrical signals movable relevant with neuromuscular, wherein containing the information that much is associated with limb motion, different limb actions has different contraction of muscle patterns, the electromyographic signal feature is also incited somebody to action difference to some extent, by just can distinguish the different patterns of limbs to the analysis of these features, therefore, it is not only extensively applied to clinical diagnosis, fields such as sports medical science also become the desirable control signal of physical disabilities' artificial limb.Along with the research to the electromyographic signal mechanism of production, researchers find that sEMG has non-periodic, non-stationary, chaotic characteristic such as non-linear, use non-linear index to come the pattern of identification maneuver surface electromyogram signal also to obtain deep research in recent years, for example Zou Xiao sun, Lei Min etc. combine largest Lyapunov exponent and multiscale analysis method, utilize then support vector machine identified preferably the human body forearm in turn over, turn up, clench fist, open up fist, on cut with incision six classes and move; The Arjunan S P of Australia RMIT, Naik G R etc. extract the FRACTAL DIMENSION feature of the electromyographic signal of all kinds of actions of hand, and each action of combination supporting vector machine identification hand has then obtained high recognition.
Because surface electromyogram signal is a kind of physiological signal with non-stationary property in itself, wavelet package transforms is that a kind of effective ways and wavelet package transforms frequency range when difference of analyzing the non-stationary property signal all can accurately be portrayed signal, the characteristics of enriching pattern information are provided, again because the uncertainty of non-stationary signal, spectral power distribution situation and the characterization system complexity of Wavelet Packet Entropy energy reflected signal in each frequency band, so the present invention analyzes the surface electromyogram signal with non-stationary property with Wavelet Packet Entropy, find out the stable characteristics value, and with the carry out pattern classification of support vector machine to surface electromyogram signal, realized the pattern-recognition of 6 kinds of different actions of hand preferably.
Summary of the invention
Realize the multi-locomotion mode identification of hand for utilize less electromyographic signal power-collecting electrode as far as possible, at characteristics such as the non-linear non-stationaries of surface electromyogram signal (sEMG), the present invention proposes a kind of hand electromyographic signal action identification method based on Wavelet Packet Entropy and support vector machine.Gather corresponding surface electromyogram signal from the related muscles group, utilize wavelet packet coefficient energy distribution analytical table facial muscle electric signal characteristic, the combining information entropy is analyzed its uncertainty and complicacy, earlier surface electromyogram signal is carried out WAVELET PACKET DECOMPOSITION, extract the wavelet packet coefficient matrix of each node, calculate Wavelet Packet Entropy by each sub belt energy.Wavelet Packet Entropy with electromyographic signal is feature construction proper vector input support vector machine classifier, and a plurality of actions of hand are classified.
In order to realize above purpose, the inventive method mainly may further comprise the steps:
Step (1). obtain human upper limb electromyographic signal sample data, specifically: at first pick up human upper limb electromyographic signal sEMG by the electromyographic signal collection instrument, adopt small echo spatial domain correlation filtering method that the electromyographic signal that contains interference noise is carried out de-noising again.
Step (2). the sEMG actuating signal that step (1) is obtained is carried out WAVELET PACKET DECOMPOSITION, specifically: adopt the sym8 wavelet function, sEMG is carried out 3 layers of WAVELET PACKET DECOMPOSITION, owing to kept the quadrature resolution characteristic in the multiresolution analysis, two frequency bands after each node decomposes are not overlapping mutually, and bandwidth reduces by half.
Step (3). extract the wavelet packet coefficient of each node, calculate Wavelet Packet Entropy by each sub belt energy.Specific as follows:
The wavelet packet node energy can effectively be represented the energy of signal, the definition wavelet-packet energy
Be WAVELET PACKET DECOMPOSITION coefficient component
In energy.
Wherein,
jBe the decomposition number of plies,
iBe the subband sequence.
The signal gross energy can be decomposed into different frequency bands wavelet packet component sum, namely
The normalization of wavelet-packet energy is expressed as
Obviously,
Very responsive to energy change, it has reflected the distribution situation of wavelet-packet energy and the energy relativeness of each subband sequence signal.
According to the definition of Shannon entropy, the definition Wavelet Packet Entropy is
For the signal (as a simple signal) of a rule, because all energy only concentrate in the frequency band, so, its relative wavelet-packet energy is very little.On the other hand, if a very complicated signal (as a random signal), energy will be distributed in each frequency band, and its relative wavelet-packet energy can be more close, thereby make Wavelet Packet Entropy reach maximum.
Step (4). import support vector machine classifier with the Wavelet Packet Entropy that step (3) is tried to achieve as proper vector, obtain recognition result.
Support vector machine (Support vector machines, SVM) be based on Statistical Learning Theory and structural risk minimization, its basic thought is that the sample of the input space is mapped to high-dimensional feature space by nonlinear transformation, asks for the optimal classification face that the sample linearity is separated then in feature space.Algorithm uses the capacity of class interval control line inquiry learning machine, thereby makes the structure risk minimum, also makes it have stronger generalization ability under limited sample.According to structural risk minimization, in order to minimize the upper bound of expected risk, SVM is by the structure of optimum lineoid, minimizes VC (Vapnik and Chervonnenkis) fiducial range under the condition of fixing learning machine empiric risk.Here, the construction problem of optimum lineoid comes down to find the solution under the constraint condition quadratic programming problem, to obtain an optimal classification function:
Here
Be threshold value,
Be i training sample,
Be test sample book,
Be total number of training,
Be Lagrangian coefficient,
For satisfying the kernel function of Mercer condition.Different Kernel Function Transformation uses kernel function to avoid directly calculating in the feature space of higher-dimension to different feature spaces, and generally using more kernel function has following three kinds: polynomial kernel function, radially basic kernel function and neural network kernel function.
The thought of SVM is the classification at two class problems, for the multiclass problem, must re-construct the svm classifier device and find the solution, and mainly contains two kinds of methods at present.A kind of is that the multiclass algorithm that proposed in 1998 with Weston is representative, and this algorithm is on the basis of classical SVM theory, re-constructs many-valued disaggregated model and realizes many-valued classification.The objective function that this algorithm is selected is very complicated, realizes difficulty, and computation complexity is also very high, thereby less use.Another kind of building method is to realize structure to many-valued sorter by making up a plurality of two-value sub-classifiers, and this method has two kinds of branching algorithms, i.e. " one-to-many " and " one to one " algorithm." one-to-many " algorithm is proposed by Vapnik, its basic thought be for
Class problem structure
Individual two class sorters, the
Individual SVM is with
Training sample in the class is as positive training sample, and with other sample as negative training sample, last output is that two class sorters are output as that maximum class." one to one " algorithm be
The all possible two class sorters of structure only exist at every turn in the class sample
Train on the two class samples in the class, the result constructs altogether
Individual sorter makes up these two classes sorters and uses the ballot method, and who gets the most votes's class is the class under the new point.The present invention selects for use on the basis of " one to one " algorithm, and the between class distance in the utilization cluster analysis and the method for binary tree are constructed the multicategory classification device.
The present invention compares with existing many hand electromyographic signal action identification methods, has following characteristics:
Wavelet packet analysis is a kind of signal to be carried out more careful analytical approach, it is divided frequency band at many levels, there is not the HFS of segmentation further to decompose to wavelet transformation, and can be according to analyzed feature, select frequency band adaptively, make it to be complementary with signal band, thereby improve the Signal Processing ability.Wavelet transformation can become signal decomposition the weighted sum of different scale base small echo, is a kind of effective ways of analyzing non-stationary signal.But not enough is that it is lower in the frequency resolution of high band, causes in some applications, can not satisfy actual requirement.And wavelet package transforms also decomposes with two filter-dividers HFS, has improved high frequency resolution.Wavelet package transforms is that a kind of effective ways and wavelet package transforms frequency range when difference of analyzing the non-stationary property signal all can accurately be portrayed signal, the characteristics of enriching pattern information are provided, again because the uncertainty of non-stationary signal, Wavelet Packet Entropy can reflected signal spectral power distribution situation and the characterization system complexity in each frequency band, the uniformity coefficient that can each frequency band energy of reflected signal distributes.The present invention carries out feature extraction with Wavelet Packet Entropy to surface electromyogram signal, and progressively increase the value of parameter with fixed intervals, search for the svm classifier device parameter value that makes the pattern-recognition rate the highest, the proper vector of three road sEMG signal extractions input support vector machine, can effectively identify stretch wrist, bend wrist, the exhibition fist, clench fist, outward turning, 6 kinds of patterns of inward turning, obtained the discrimination higher than traditional neural network.
Description of drawings
Fig. 1 is implementing procedure figure of the present invention;
The feature distribution plan of the electromyographic signal Wavelet Packet Entropy that Fig. 2 obtains for the present invention;
Embodiment
Below in conjunction with accompanying drawing embodiments of the invention are elaborated: present embodiment is being to implement under the prerequisite with the technical solution of the present invention, has provided detailed embodiment and concrete operating process.
As shown in Figure 1, present embodiment comprises the steps:
Step 1, specifically: at first pick up the human upper limb electromyographic signal by the electromyographic signal collection instrument, adopt small echo spatial domain correlation filtering method that the electromyographic signal that contains interference noise is carried out de-noising again.
(1) the mt400 electromyographic signal collection instrument of employing U.S. Noraxon company, earlier with alcohol obliterating decontamination on experimenter's musculus extensor carpi ulnaris and musculus flexor carpi ulnaris respectively, to strengthen the pick-up capability of signal, then disposable electromyographic electrode is affixed on experimenter's forearm musculus extensor carpi ulnaris, musculus flexor carpi ulnaris and extensor muscle of fingers place, is used for the collection surface electromyographic signal.The output of electromyographic signal collection instrument (frequency acquisition 1000Hz) is connected to by bluetooth USB on the computing machine that the myoelectricity process software is housed, gathers that wrist protractor, side wrist are bent, turn on the extensor muscle of fingers place wrist, turn over, open up fist under the wrist, clench fist, each 60 groups of the surface electromyogram signals of 6 kinds of forearm actions such as wrist outward turning, wrist inward turning.
(2) adopt spatial domain correlation filtering method to carry out denoising Processing.The original electromyographic signal of gathering is carried out 5 layers of wavelet decomposition, and basic small echo is selected Bi-orthogonal Spline Wavelet Transformation bior1.5 for use.In the correlation filtering de-noising of spatial domain, need to set a certain noise threshold, get the wrist extensor electric signal the wavelet transformation high frequency coefficient before
(in the present embodiment
Get 200) individual point, the high frequency coefficient of electromyographic signal when these that get do not move corresponding to hand is estimated the noise energy threshold value of electromyographic signal with the variance of these points.
Step 2, the sEMG actuating signal that step 1 is obtained is carried out WAVELET PACKET DECOMPOSITION, specifically be to adopt the sym8 wavelet function, sEMG is carried out 3 layers of WAVELET PACKET DECOMPOSITION, owing to kept the quadrature resolution characteristic in the multiresolution analysis, two frequency bands after each node decomposes are not overlapping mutually, and bandwidth reduces by half.Frequency band range through each node correspondence after three layers of decomposition of wavelet packet is as shown in table 1.
The corresponding frequency band range of the 3rd layer of each node of table 1 WAVELET PACKET DECOMPOSITION
| Node | Frequency range (Hz) | Node | Frequency range (Hz) |
| 5－67 |
 |
253－315 | |
| 67－129 |
 |
315－377 | |
 |
129－191 |
 |
377－439 |
 |
191－253 |
 |
439－500 |
Step 3 is extracted the wavelet packet coefficient of each node, calculates Wavelet Packet Entropy by each sub belt energy.
The wrist protractor, the wrist musculus flexor that Figure 2 shows that the experimenter are being stretched wrist, are bending wrist, are being clenched fist, opening up the Wavelet Packet Entropy eigenwert distribution plan that each the 30 groups of sample data under four kinds of patterns of fist are drawn by Matlab.By can finding out clearly among the figure, by the two dimensional character vector that the wavelet packet entropy of wrist protractor, wrist musculus flexor constitutes, can well distinguish and stretch wrist, bend wrist, clench fist, open up 4 kinds of patterns such as fist
Stretch wrist, bend wrist, the exhibition fist, clench fist, outward turning, inward turning each move corresponding 60 group of 3 dimensional feature vector, therefrom appoint and get 20 stack features vectors and train as training set input SVM, import support vector machine as test set for all the other 40 groups.
For support vector machine classifier, the present invention has adopted 3 kinds of different kernel functions, i.e. linear nuclear, rbf nuclear and poly nuclear.With three kinds of different IPs functions same eigenwert is tested respectively, the result shows that the classifying quality that adopts rbf to examine (radially base nuclear) is best.Therefore the present invention adopts rbf kernel function design svm classifier device, also relates to the selection of parameter here, comprises that control divides the adjustable parameter of the degree of sample punishment to mistake
, and the undetermined parameter in the rbf kernel function
, the selection of these two parameters directly influences classifying quality.In order effectively to determine the value of these two parameters, the present invention progressively increases the value of parameter with fixed intervals, searches for and makes the highest parameter value of pattern-recognition rate.
Table 2 has been listed 40 groups of test sample books at the recognition result of SVM pattern classifier, and average recognition rate reaches more than 89.5%, and recognition effect is more satisfactory.
Table 2 svm classifier device recognition result
| Action classification | Correct identification | Mistake identification | Discrimination |
| Stretch wrist | 37 | 3 | 92.5% |
| Bend wrist | 39 | 1 | 97.5% |
| The exhibition fist | 36 | 4 | 90% |
| Clench fist | 38 | 2 | 95% |
| Outward turning | 32 | 8 | 80% |
| Inward turning | 33 | 7 | 82.5% |
Claims (2)
1. based on the hand electromyographic signal action identification method of Wavelet Packet Entropy and support vector machine, it is characterized in that this method comprises the steps:
Step (1). obtain human upper limb electromyographic signal sample data, specifically: at first pick up the human upper limb surface electromyogram signal by the electromyographic signal collection instrument, adopt small echo spatial domain correlation filtering method that the surface electromyogram signal that contains interference noise is carried out de-noising again;
Step (2). the surface electromyogram signal that step (1) is obtained carries out WAVELET PACKET DECOMPOSITION, specifically: adopt the sym8 wavelet function, surface electromyogram signal is carried out three layers of WAVELET PACKET DECOMPOSITION, owing to kept the quadrature resolution characteristic in the multiresolution analysis, two frequency bands after each node decomposes are not overlapping mutually, and bandwidth reduces by half;
Step (3). extract the wavelet packet coefficient of each node, calculate Wavelet Packet Entropy by each sub belt energy, specifically:
The definition Wavelet Packet Entropy is
, wherein
For the normalization of wavelet-packet energy is represented, namely
, in the formula
Be WAVELET PACKET DECOMPOSITION coefficient component
In energy
,
Be gross energy
,
jBe the decomposition number of plies,
iBe the subband sequence;
Step (4). import support vector machine classifier with the Wavelet Packet Entropy that step (3) is tried to achieve as proper vector, obtain recognition result.
2. the hand electromyographic signal action identification method based on Wavelet Packet Entropy and support vector machine according to claim 1, it is characterized in that: the between class distance in the utilization cluster analysis and the method for binary tree are constructed the described sorter of multiclass, progressively increase the value of parameter with fixed intervals, search for and make the highest parameter value of pattern-recognition rate.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN2013103000988A CN103345641A (en) | 2013-07-16 | 2013-07-16 | Hand electromyographic signal motion recognition method based on wavelet entropy and support vector machine |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN2013103000988A CN103345641A (en) | 2013-07-16 | 2013-07-16 | Hand electromyographic signal motion recognition method based on wavelet entropy and support vector machine |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN103345641A true CN103345641A (en) | 2013-10-09 |
Family
ID=49280436
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN2013103000988A Pending CN103345641A (en) | 2013-07-16 | 2013-07-16 | Hand electromyographic signal motion recognition method based on wavelet entropy and support vector machine |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN103345641A (en) |
Cited By (14)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105326501A (en) * | 2015-12-10 | 2016-02-17 | 宁波工程学院 | Muscle disease monitoring method based on sEMG |
| CN105426842A (en) * | 2015-11-19 | 2016-03-23 | 浙江大学 | Support vector machine based surface electromyogram signal multi-hand action identification method |
| CN106019090A (en) * | 2016-05-11 | 2016-10-12 | 西安西热节能技术有限公司 | Partial discharge electromagnetic wave signal energy feature extraction method |
| CN106293057A (en) * | 2016-07-20 | 2017-01-04 | 西安中科比奇创新科技有限责任公司 | Gesture identification method based on BP neutral net |
| CN106527716A (en) * | 2016-11-09 | 2017-03-22 | 努比亚技术有限公司 | Wearable equipment based on electromyographic signals and interactive method between wearable equipment and terminal |
| CN107811635A (en) * | 2016-09-12 | 2018-03-20 | 深圳先进技术研究院 | A kind of health status sorting technique and device based on physiology signal |
| CN107918162A (en) * | 2017-11-16 | 2018-04-17 | 昆山光创信息技术有限公司 | A kind of intelligent identification Method of main line of communication surrounding enviroment disturbing signal |
| CN108703824A (en) * | 2018-03-15 | 2018-10-26 | 哈工大机器人（合肥）国际创新研究院 | A kind of bionic hand control system and control method based on myoelectricity bracelet |
| CN109657651A (en) * | 2019-01-16 | 2019-04-19 | 杭州电子科技大学 | A kind of continuous method for estimating of lower limb knee joint based on electromyography signal |
| CN110547806A (en) * | 2019-09-11 | 2019-12-10 | 湖北工业大学 | gesture action online recognition method and system based on surface electromyographic signals |
| CN111027488A (en) * | 2019-12-11 | 2020-04-17 | 深圳先进技术研究院 | Signal classification method and device |
| CN111397896A (en) * | 2020-03-08 | 2020-07-10 | 华中科技大学 | Fault diagnosis method and system for rotary machine and storage medium |
| CN111657939A (en) * | 2020-05-27 | 2020-09-15 | 杭州电子科技大学 | Upper limb joint angle estimation method based on myoelectric signal GA-Elman |
| CN111904416A (en) * | 2020-07-27 | 2020-11-10 | 深圳先进技术研究院 | Muscle fatigue detection method, computer device, and storage medium |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US5462065A (en) * | 1994-08-17 | 1995-10-31 | Cusimano; Maryrose | Integrated movement analyziing system |
| CN102073881A (en) * | 2011-01-17 | 2011-05-25 | 武汉理工大学 | Denoising, feature extraction and pattern recognition method for human body surface electromyography signals |
-
2013
- 2013-07-16 CN CN2013103000988A patent/CN103345641A/en active Pending
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US5462065A (en) * | 1994-08-17 | 1995-10-31 | Cusimano; Maryrose | Integrated movement analyziing system |
| CN102073881A (en) * | 2011-01-17 | 2011-05-25 | 武汉理工大学 | Denoising, feature extraction and pattern recognition method for human body surface electromyography signals |
Non-Patent Citations (3)
| Title |
|---|
| 席旭刚 等: "基于相关性分析和支持向量机的手部肌电信号动作识别", 《电子与信息学报》, vol. 30, no. 10, 31 October 2008 (2008-10-31), pages 2315 - 2319 * |
| 朱嘉: "基于肌电信号的人手动作模式识别", 《中国优秀硕士学位论文全文数据库 信息科技辑》, no. 4, 15 April 2009 (2009-04-15), pages 17 - 40 * |
| 王艳景 等: "基于小波包熵和支持向量机的运动想象任务分类研究", 《仪器仪表学报》 * |
Cited By (18)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105426842B (en) * | 2015-11-19 | 2018-08-14 | 浙江大学 | Multiclass hand motion recognition method based on support vector machines and surface electromyogram signal |
| CN105426842A (en) * | 2015-11-19 | 2016-03-23 | 浙江大学 | Support vector machine based surface electromyogram signal multi-hand action identification method |
| CN105326501A (en) * | 2015-12-10 | 2016-02-17 | 宁波工程学院 | Muscle disease monitoring method based on sEMG |
| CN106019090A (en) * | 2016-05-11 | 2016-10-12 | 西安西热节能技术有限公司 | Partial discharge electromagnetic wave signal energy feature extraction method |
| CN106019090B (en) * | 2016-05-11 | 2018-10-02 | 西安西热节能技术有限公司 | Shelf depreciation electromagnetic wave signal energy feature extraction method |
| CN106293057A (en) * | 2016-07-20 | 2017-01-04 | 西安中科比奇创新科技有限责任公司 | Gesture identification method based on BP neutral net |
| CN107811635A (en) * | 2016-09-12 | 2018-03-20 | 深圳先进技术研究院 | A kind of health status sorting technique and device based on physiology signal |
| CN106527716A (en) * | 2016-11-09 | 2017-03-22 | 努比亚技术有限公司 | Wearable equipment based on electromyographic signals and interactive method between wearable equipment and terminal |
| CN106527716B (en) * | 2016-11-09 | 2020-12-29 | 努比亚技术有限公司 | Wearable device based on electromyographic signals and interaction method of wearable device and terminal |
| CN107918162A (en) * | 2017-11-16 | 2018-04-17 | 昆山光创信息技术有限公司 | A kind of intelligent identification Method of main line of communication surrounding enviroment disturbing signal |
| CN108703824A (en) * | 2018-03-15 | 2018-10-26 | 哈工大机器人（合肥）国际创新研究院 | A kind of bionic hand control system and control method based on myoelectricity bracelet |
| CN109657651A (en) * | 2019-01-16 | 2019-04-19 | 杭州电子科技大学 | A kind of continuous method for estimating of lower limb knee joint based on electromyography signal |
| CN110547806A (en) * | 2019-09-11 | 2019-12-10 | 湖北工业大学 | gesture action online recognition method and system based on surface electromyographic signals |
| CN111027488A (en) * | 2019-12-11 | 2020-04-17 | 深圳先进技术研究院 | Signal classification method and device |
| WO2021115059A1 (en) * | 2019-12-11 | 2021-06-17 | 深圳先进技术研究院 | Signal classification method and device |
| CN111397896A (en) * | 2020-03-08 | 2020-07-10 | 华中科技大学 | Fault diagnosis method and system for rotary machine and storage medium |
| CN111657939A (en) * | 2020-05-27 | 2020-09-15 | 杭州电子科技大学 | Upper limb joint angle estimation method based on myoelectric signal GA-Elman |
| CN111904416A (en) * | 2020-07-27 | 2020-11-10 | 深圳先进技术研究院 | Muscle fatigue detection method, computer device, and storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN103345641A (en) | Hand electromyographic signal motion recognition method based on wavelet entropy and support vector machine | |
| Abbaspour et al. | Evaluation of surface EMG-based recognition algorithms for decoding hand movements | |
| CN102930284B (en) | Surface electromyogram signal pattern recognition method based on empirical mode decomposition and fractal | |
| Sapsanis et al. | EMG based classification of basic hand movements based on time-frequency features | |
| Chen et al. | Surface electromyography feature extraction via convolutional neural network | |
| Al Omari et al. | Pattern recognition of eight hand motions using feature extraction of forearm EMG signal | |
| CN102961203B (en) | Method for identifying surface electromyography (sEMG) on basis of empirical mode decomposition (EMD) sample entropy | |
| CN104107042A (en) | Electromyographic signal gait recognition method based on particle swarm optimization and support vector machine | |
| CN101859377A (en) | Electromyographic signal classification method based on multi-kernel support vector machine | |
| CN110826625B (en) | Finger gesture classification method based on surface electromyographic signals | |
| Kakoty et al. | EMG feature set selection through linear relationship for grasp recognition | |
| AlOmari et al. | Analysis of extracted forearm sEMG signal using LDA, QDA, K-NN classification algorithms | |
| Bhagwat et al. | Electromyogram (EMG) based fingers movement recognition using sparse filtering of wavelet packet coefficients | |
| Bai et al. | Upper arm motion high-density sEMG recognition optimization based on spatial and time-frequency domain features | |
| CN108681685A (en) | A kind of body work intension recognizing method based on human body surface myoelectric signal | |
| Mohammadpour et al. | Comparison of EEG signal features and ensemble learning methods for motor imagery classification | |
| CN108268844A (en) | Movement recognition method and device based on surface electromyogram signal | |
| Li et al. | EEG signal classification method based on feature priority analysis and CNN | |
| CN109492546A (en) | A kind of bio signal feature extracting method merging wavelet packet and mutual information | |
| CN110363152A (en) | A kind of artificial leg road conditions recognition methods based on surface electromyogram signal | |
| Aly et al. | Bio-signal based motion control system using deep learning models: a deep learning approach for motion classification using EEG and EMG signal fusion | |
| CN113111831A (en) | Gesture recognition technology based on multi-mode information fusion | |
| Bergil et al. | Efficient hand movement detection using k-means clustering and k-nearest neighbor algorithms | |
| Jie et al. | High dimensional feature data reduction of multichannel sEMG for gesture recognition based on double phases PSO | |
| CN102968641A (en) | Electromyographic signal identification method based on spherical mean value Lyapunov exponent and correlation dimension |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20131009 |
|
| WD01 | Invention patent application deemed withdrawn after publication |