CN105654037A - Myoelectric signal gesture recognition method based on depth learning and feature images - Google Patents
Myoelectric signal gesture recognition method based on depth learning and feature images Download PDFInfo
- Publication number
- CN105654037A CN105654037A CN201510971796.XA CN201510971796A CN105654037A CN 105654037 A CN105654037 A CN 105654037A CN 201510971796 A CN201510971796 A CN 201510971796A CN 105654037 A CN105654037 A CN 105654037A
- Authority
- CN
- China
- Prior art keywords
- data
- signal
- image
- feature
- depth
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
- G06F2218/02—Preprocessing
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F3/00—Input arrangements for transferring data to be processed into a form capable of being handled by the computer; Output arrangements for transferring data from processing unit to output unit, e.g. interface arrangements
- G06F3/01—Input arrangements or combined input and output arrangements for interaction between user and computer
- G06F3/011—Arrangements for interaction with the human body, e.g. for user immersion in virtual reality
- G06F3/015—Input arrangements based on nervous system activity detection, e.g. brain waves [EEG] detection, electromyograms [EMG] detection, electrodermal response detection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F3/00—Input arrangements for transferring data to be processed into a form capable of being handled by the computer; Output arrangements for transferring data from processing unit to output unit, e.g. interface arrangements
- G06F3/01—Input arrangements or combined input and output arrangements for interaction between user and computer
- G06F3/017—Gesture based interaction, e.g. based on a set of recognized hand gestures
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
- G06N3/084—Backpropagation, e.g. using gradient descent
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
- G06F2218/08—Feature extraction
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
- G06F2218/12—Classification; Matching
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- General Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- General Health & Medical Sciences (AREA)
- Health & Medical Sciences (AREA)
- Human Computer Interaction (AREA)
- Artificial Intelligence (AREA)
- Biomedical Technology (AREA)
- Computing Systems (AREA)
- Biophysics (AREA)
- Evolutionary Computation (AREA)
- Data Mining & Analysis (AREA)
- Computational Linguistics (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Molecular Biology (AREA)
- Dermatology (AREA)
- Neurology (AREA)
- Neurosurgery (AREA)
- Life Sciences & Earth Sciences (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Signal Processing (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses a myoelectric signal gesture recognition method based on depth learning and feature images. Firstly acquired gesture myoelectric original signals are preprocessed; then feature extraction is performed, features including a time domain and a time-frequency domain are extracted through sampling windows of different sizes and probabilities and the features are converted into images; then the feature images and corresponding action tags are inputted to a depth neural network together to be trained so that a network model is obtained; and finally test data and the network model obtained through training are inputted to a depth convolution neural network to be predicted and the prediction tags of all the images of each period of action are obtained, voting of the tags is performed according to the majority voting rules and the tag with the highest number of votes is the category of the period of action. The classifier of the depth convolution neural network is recognized based on the feature images. Different gestures of the same subject can be accurately recognized by using the classification method based on the depth convolution neural network so that the gestures between different subjects can be accurately recognized.
Description
Technical field
The invention belongs to computer to combine with bio signal field, specifically based on degree of depth convolutional neural networks, the gesture that the characteristic image extracted from flesh electrical signal is corresponding is identified.
Background technology
Along with the fast development of computer vision, the new technology such as touch mutual, perception calculating, one of perception user interface (perceptualuserinterface, PUI) research emphasis becoming field of human-computer interaction. Perception user interface is a kind of person to person and the interacting activity between people and real world is the high interaction of prototype, the user interface of hyperchannel, and its target makes man-machine interaction become consistent with people with the interaction of real world to reach mutual boundary directly perceived, natural. As novel human-machine interaction form, the ultimate target of PUI is the man-machine interface realizing " focus be put on man ", and namely in interactive process, computer can adapt to the natural interaction custom of the mankind, instead of the specific operation requirement of adaptation computer of being asked for help. Judge and understand the mutual intention of the mankind better to enable computer, " integration raw, mechanical, electrical " is one of important trend of following man-machine interaction development, namely by specific sensing equipment by the cognition of organism or perceptual signal (such as flesh electrical signal) digitizing, and carry out integrated fusion with the signal of other perception or cognitive passage, complete various man-machine interaction task naturally, collaborative.
Up to the present, in research at home and abroad, a lot of machine learning method is used in flesh electrical signal gesture identification, such as artificial neural network, k nearest neighbor, linear discriminate analysis, SVMs and hidden Markov model. And degree of deep learning method is seldom employed, convolution neural network used herein is exactly a kind of degree of deep learning method, and being a little of this kind of method, it is not necessary to carry out a large amount of feature extraction operation, a small amount of feature can also obtain good recognition rate.
Convolutional neural networks (ConvolutionalNeuralNetwork, CNN) is a kind of feedforward neural network, and its artificial neuron unit can respond the surrounding cells in part coverage, has outstanding performance for large-scale image procossing.Convolutional neural networks is the neural network model of a kind of special deep layer, and its singularity is embodied in two aspects, and on the one hand connection between its neurone is that non-fully connects, and in same layer, the weight of connection between some neurone is shared on the other hand. The network structure that its non-fully connects and weights are shared makes it more to be similar to biological neural network, reduces the complexity of network model, decreases the quantity of weights.
Summary of the invention
It is an object of the invention to for the deficiencies in the prior art, it is provided that a kind of based on the flesh electrical signal gesture identification method of degree of depth study and characteristic image.
It is an object of the invention to be achieved through the following technical solutions: a kind of based on the flesh electrical signal gesture identification method of degree of depth study and characteristic image, comprise the following steps:
(1) from open data set NinaPro, obtain gesture motion flesh electricity data; Flesh electrical signal is carried out pre-treatment, comprises operations such as removing noise, signal rectification and signal normalization, specifically comprise following sub-step:
(1.1) bandpass filtering (band-passfilter), original surface flesh electricity data bandwidth is 15-500Hz, is 0-25Hz after filtering;
(1.2) signal amplifies (amplification);
(1.3) rootmean-square correction (RMSrectification);
(1.4) 1Hz low-pass filtering (zero phase 2 rank Bart irrigates this filtering, zero-phasesecondorderButterworthfilter);
(1.5) sample of band ambiguity label is removed.
(2) characteristic image generates, and specifically comprises following sub-step:
(2.1) using 50ms, 100ms, 150ms, data are sampled by the sample window of three kinds of length, and wherein the moving step length of sample window is the 25% of length of window;
(2.2) in the data after step 2.1 being sampled, the data of each passage extract feature matrix;
(2.3) the image building form of parallel channels formula is used to be stored in image as Color Channel to obtain characteristic image in all feature matrixes extracted in same sample.
(3) deep neural network model training and gesture identification, it specifically comprises following sub-step:
(3.1) revise VGGNet network and make its deep neural network model being applicable to characteristic image;
(3.2) owing to each each tested gesture motion has 10 repeating datas, therefore 10 repeating datas are divided into training data according to the ratio of 5:3:2, test data, checking data;
(3.3) training data sample and checking (validation) data sample is used to carry out network model optimization (the manual setting network convolution number of plies and the full articulamentum number of plies) and parameter adjustment (convolution core size and the full articulamentum of manual setting convolutional layer export number);
(3.4) parameter after the network structure model obtained in step (3.1) and step (3.3) and optimization is used to carry out the training of network model, at least iteration more than 300 times in training;
(3.5) test data is input in the network model trained in step (3.3) and carries out Tag Estimation;
(3.6) according to majority, the label obtained is agreed to that rule (MajorityVotingRule) is voted according to affiliated data section, obtain last classification results.
Further, in step (2.2), employ 10 kinds of monodrome features and data are carried out feature extraction, respectively: the difference (D-Value) of two frame signal amplitudes, the absolute value sum (IEMG) of signal amplitude, the absolute mean value (MAV) of signal amplitude, improve the absolute mean value 1 (MMAV1) of signal amplitude, improve the absolute mean value 2 (MMAV2) of signal amplitude, signal rootmean-square (RMS), estimate muscular contraction force non-linear detection device (v-order), the logarithmic detector (LOG) of estimation muscular contraction force, the average (DWPT-MEAN) of signal after wavelet package transforms, signal standard deviation (DWPT-SD) after wavelet package transforms.
Further, in step (2.2), employ a kind of special feature matrix generating mode, it is specially: assuming that Ninapro flesh electricity data fragment Vi, Vi are the matrixes of the capable c row of f, f is the frame number in data fragment, c is the passage number of data fragment, first Vi matrix being changed into c vector v i in the way of row guiding, the length of each vi is f, for generating feature matrix Pi. Pi is a line number and row number is all the square formation of f, if the element p in PiJ, k, wherein pJ, k�� Pi, 0��j��f �� c, 0��k��f, the line number being respectively in matrix and column number. Then pJ, k=C (vI, j, vI, k) wherein function C be fundamental function, it is intended that the difference solving two elements in vi is used as the expression of temporal aspect.
Further, in step (2.3), it may also be useful to the image building form generating feature image of parallel channels formula. The image building form of parallel channels formula is specially: the feature matrix assuming to obtain at present f �� f size of c kind feature n passage, if c passage in data is regarded as the Color Channel being similar in image, the image then being become (n �� c) �� f �� f, as input, just can make full use of convolutional neural networks.
The invention has the beneficial effects as follows: first collection gesture flesh electricity original signal is comprised signal rectification and the pre-treatment of signal filtering by the inventive method; Then carry out feature extraction, extracted the feature comprising time domain, time-frequency domain by the sample window of different size and probability, and these Feature Conversion are become image (CostMatrix), store according to certain arrangement mode; 3rd step, is input in deep neural network (DeepConvolutionalNeuralNetworks) together with the 2nd step feature is extracted the characteristic image the obtained action label corresponding with these characteristic images and trains, obtain network model; 4th step, the network model input degree of depth convolutional neural networks that test data and training are obtained is predicted, obtain the prediction label of every section of all image of action, finally according to majority, these labels being agreed to, rule (MajorityVotingRule) is voted, the highest person of poll is this section of action classification. Degree of depth convolutional neural networks sorter is identified by feature based image of the present invention. Use the sorting technique based on degree of depth convolutional neural networks can accurately identify same tested different gestures, more accurately identify different tested gesture.
Accompanying drawing explanation
Fig. 1 is the method for the invention schema;
Fig. 2 is 3 gesture collection that the present invention tests the NinaPro data set chosen, and (a) is the gesture collection of 5 kinds of wrist motion, and (b) is 8 kinds of hand gestures gesture collection, and (c) is 12 kinds of finger motion gesture collection;
Fig. 3 is parallel channels arrangement mode schematic diagram of the present invention;
Fig. 4 is degree of depth convolutional neural networks structure used in the present invention.
Embodiment
Below in conjunction with the drawings and specific embodiments, the present invention is described in further detail.
Provided by the invention a kind of based on the flesh electrical signal gesture identification method of degree of depth study and characteristic image, comprise the following steps:
(1) from open data set NinaPro, obtain gesture motion flesh electricity data; Flesh electrical signal is carried out pre-treatment, comprises operations such as removing noise, signal rectification and signal normalization, specifically comprise following sub-step:
(1.1) bandpass filtering (band-passfilter), original surface flesh electricity data bandwidth is 15-500Hz, is 0-25Hz after filtering;
(1.2) signal amplifies (amplification);
(1.3) rootmean-square correction (RMSrectification);
(1.4) 1Hz low-pass filtering (zero phase 2 rank Bart irrigates this filtering, zero-phasesecondorderButterworthfilter);
(1.5) sample of band ambiguity label is removed.
(2) characteristic image generates, and specifically comprises following sub-step:
(2.1) using 50ms, 100ms, 150ms, data are sampled by the sample window of three kinds of length, and wherein the moving step length of sample window is the 25% of length of window;
(2.2) in the data after step 2.1 being sampled, the data of each passage extract feature matrix;
Present invention uses 10 kinds of monodrome features and data are carried out feature extraction, as shown in table 1, respectively: the difference (D-Value) of two frame signal amplitudes, the absolute value sum (IEMG) of signal amplitude, the absolute mean value (MAV) of signal amplitude, improve the absolute mean value 1 (MMAV1) of signal amplitude, improve the absolute mean value 2 (MMAV2) of signal amplitude, signal rootmean-square (RMS), estimate muscular contraction force non-linear detection device (v-order), the logarithmic detector (LOG) of estimation muscular contraction force, the average (DWPT-MEAN) of signal after wavelet package transforms, signal standard deviation (DWPT-SD) after wavelet package transforms.
Table 1 new feature collection comprises feature description
The generating mode of feature matrix is as follows: assume Ninapro flesh electricity data fragment Vi, Vi is the matrix of the capable c row of f, f is the frame number in data fragment, c is the passage number of data fragment, first Vi matrix is changed into c vector v i in the way of row guiding, the length of each vi is f, for generating feature matrix Pi. Pi is a line number and row number is all the square formation of f, if the element p in PiJ, k, wherein pJ, k�� Pi, 0��j��f �� c, 0��k��f, the line number being respectively in matrix and column number. Then pJ, k=C (vI, j, vI, k) wherein function C be fundamental function, it is intended that the difference solving two elements in vi is used as the expression of temporal aspect.
(2.3) the image building form of parallel channels formula is used to be stored in image as Color Channel to obtain characteristic image in all feature matrixes extracted in same sample. As shown in Figure 3, the image building form of parallel channels formula is specially: the feature matrix assuming to obtain at present f �� f size of c kind feature n passage, if c passage in data is regarded as the Color Channel being similar in image, the image then being become (n �� c) �� f �� f, as input, just can make full use of convolutional neural networks.
(3) deep neural network model training and gesture identification
Degree of depth study structure is a kind of multilayer perceptron containing many hidden layers. Degree of depth study forms more abstract high-rise expression attribute classification or feature by combination low-level feature, represents to find the distributed nature of data. The degree of depth of the present invention study structure revises VGGNet network structure, and this kind of amendment makes VGGNet well to be identified by little and dark image, and this kind of network structure comprises following network layer:
1. convolutional layer: in convolutional neural networks, every layer of convolutional layer is made up of some convolution unit, and the parameter of each convolution unit is obtained by back-propagation algorithm optimization. The object of convolution algorithm extracts the different characteristics of input, and the first layer convolutional layer may can only extract some rudimentary features such as levels such as edge, lines and angles, and more multi-layered network can iterative extraction is more complicated from low-level features feature.
2.Relu activation function: Relu can make a part of neuronic output be 0, so just causes the openness of network, and decreases the relation of interdependence of parameter, alleviates the generation of over-fitting problem.
3.MaxPooling layer: pond is on the basis that convolution feature is extracted, equalization of each convolution feature being made even, continue to reduce concealed nodes for convolution intrinsic dimensionality, reduce the design burden of sorter.
4. full articulamentum: the full articulamentum in CNN model is the hidden layer of tradition neural network. output layer is relevant to the classification task of network model, and the last output number of full articulamentum is the gesture quantity to be identified.
5.Softmax layer: Softmax layer is the layer that softmax models applying carries out recurrence operation in last training result, softmax model is the popularization of logistic regression model in many classification problem, in many classification problem, the value of the desirable two or more of class label, solving the problem of the many classification of deep neural network, concrete network structure is as shown in Figure 4.
This step specifically comprises following sub-step:
(3.1) revise VGGNet network and make its deep neural network model being applicable to characteristic image;
(3.2) owing to each each tested gesture motion has 10 repeating datas, therefore 10 repeating datas are divided into training data according to the ratio of 5:3:2, test data, checking data;
(3.3) training data sample and checking (validation) data sample is used to carry out network model optimization (the manual setting network convolution number of plies and the full articulamentum number of plies) and parameter adjustment (convolution core size and the full articulamentum of manual setting convolutional layer export number);
(3.4) parameter after the network structure model obtained in step (3.1) and step (3.3) and optimization is used to carry out the training of network model, at least iteration more than 300 times in training;
(3.5) test data is input in the network model trained in step (3.3) and carries out Tag Estimation;
(3.6) according to majority, the label obtained is agreed to that rule (MajorityVotingRule) is voted according to affiliated data section, obtain last classification results.
Embodiment
Flesh electrical signal gesture motion is judged by feature based image of the present invention and deep neural network, according to Fig. 1, mainly comprises two parts: off-line training part and ONLINE RECOGNITION part.
Off-line training part comprises:
A. collect the flesh electrical signal of gesture by flesh electricity electrode and carry out rectification and filtering process. We use open data set NinaPro to carry out method test, and signal has been carried out rectification and bandpass filtering treatment by NinaPro data set.
B. utilizing sliding window technique that training gesture is carried out CostMatrix generation, length of window is 100ms, windows overlay per-cent is 75%.
C. the CostMatrix that each passage produces is arranged, form complete characteristic image.
E. design is applicable to degree of depth god's machine network model of characteristic image, training set data sample and checking (validation) data sample is used to carry out network model optimization and parameter tune, the parameter after the network structure model obtained and optimization is used to carry out the training of network model, at least iteration more than 300 times in training, obtain final point class model.
ONLINE RECOGNITION part comprises:
A. online acquisition is to flesh electrical signal, signal carries out bandpass filtering and rectification, carries out identical Signal Pretreatment with training data
B. utilizing sliding window technique that training gesture is carried out CostMatrix generation, length of window is 100ms, windows overlay per-cent is 75%.
C. the CostMatrix that each passage produces is arranged, form complete characteristic image
D. being input in the network model that each has trained by the image of gesture to be identified, obtain the likelihood that gesture to be identified belongs to each classification, the classification selecting maximum likelihood value corresponding is as the classification of gesture to be identified.
To 5,8,12 3 kind of gesture collection (as shown in Figure 2) use the result that three kinds of sorters identify respectively:
The deep neural network models coupling characteristic image of judgement training can obtain higher recognition rate in tested.
Claims (5)
1. one kind learns the flesh electrical signal gesture identification method with characteristic image based on the degree of depth, it is characterised in that, comprise the following steps:
(1) from open data set NinaPro, obtain gesture motion flesh electricity data; Flesh electrical signal is carried out pre-treatment, comprises operations such as removing noise, signal rectification and signal normalization, specifically comprise following sub-step:
(1.1) bandpass filtering, original surface flesh electricity data bandwidth is 15-500Hz, is 0-25Hz after filtering;
(1.2) signal amplifies;
(1.3) rootmean-square correction;
(1.4) 1Hz low-pass filtering;
(1.5) sample of band ambiguity label is removed.
(2) characteristic image generates, and specifically comprises following sub-step:
(2.1) using 50ms, 100ms, 150ms, data are sampled by the sample window of three kinds of length, and wherein the moving step length of sample window is the 25% of length of window;
(2.2) in the data after step (2.1) being sampled, the data of each passage extract feature matrix;
(2.3) the image building form of parallel channels formula is used to be stored in image as Color Channel to obtain characteristic image in all feature matrixes extracted in same sample.
(3) deep neural network model training and gesture identification, specifically comprises following sub-step:
(3.1) revise VGGNet network and make its deep neural network model being applicable to characteristic image;
(3.2) owing to each each tested gesture motion has 10 repeating datas, therefore 10 repeating datas are divided into training data according to the ratio of 5:3:2, test data, checking data;
(3.3) training data sample and checking data sample is used to carry out network model optimization (the manual setting network convolution number of plies and the full articulamentum number of plies) and parameter adjustment (convolution core size and the full articulamentum of manual setting convolutional layer export number);
(3.4) parameter after the network structure model obtained in step (3.1) and step (3.3) and optimization is used to carry out the training of network model, at least iteration more than 300 times in training;
(3.5) test data is input in the network model trained in step (3.3) and carries out Tag Estimation;
(3.6) according to majority, the label obtained is agreed to that rule is voted according to affiliated data section, obtain last classification results.
2. according to claim 1 a kind of based on the flesh electrical signal gesture identification method of degree of depth study and characteristic image, it is characterized in that, in step (2.2), employ 10 kinds of monodrome features and data are carried out feature extraction, respectively: the difference (D-Value) of two frame signal amplitudes, the absolute value sum (IEMG) of signal amplitude, the absolute mean value (MAV) of signal amplitude, improve the absolute mean value 1 (MMAV1) of signal amplitude, improve the absolute mean value 2 (MMAV2) of signal amplitude, signal rootmean-square (RMS), estimate muscular contraction force non-linear detection device (v-order), the logarithmic detector (LOG) of estimation muscular contraction force, the average (DWPT-MEAN) of signal after wavelet package transforms, signal standard deviation (DWPT-SD) after wavelet package transforms.
3. according to claim 1 a kind of based on the flesh electrical signal gesture identification method of degree of depth study and characteristic image, it is characterized in that, in step (2.2), feature matrix generating mode is specially: assuming that Ninapro flesh electricity data fragment Vi, Vi are the matrixes of the capable c row of f, f is the frame number in data fragment, c is the passage number of data fragment, first Vi matrix being changed into c vector v i in the way of row guiding, the length of each vi is f, for generating feature matrix Pi. Pi is a line number and row number is all the square formation of f, if the element p in Pij,k, wherein pj,k�� Pi, 0��j��f �� c, 0��k��f, the line number being respectively in matrix and column number. Then pj,k=C (vi,j,vi,k) wherein function C be fundamental function, it is intended that the difference solving two elements in vi is used as the expression of temporal aspect.
4. according to claim 1 a kind of based on the flesh electrical signal gesture identification method of degree of depth study and characteristic image, it is characterized in that, in step (2.3), the image building form of parallel channels formula is specially: the feature matrix assuming to obtain at present f �� f size of c kind feature n passage, if c passage in data is regarded as the Color Channel being similar in image, the image then being become (n �� c) �� f �� f, as input, just can make full use of convolutional neural networks.
5. according to claim 1 a kind of based on the flesh electrical signal gesture identification method of degree of depth study and characteristic image, it is characterized in that, in step (3.1), employ degree of depth convolutional neural networks and carry out Classification and Identification, VGGNet comparatively outstanding in degree of depth convolutional neural networks is modified so that it is be applicable to the characteristic image that step 2 generates.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510971796.XA CN105654037B (en) | 2015-12-21 | 2015-12-21 | A kind of electromyography signal gesture identification method based on deep learning and characteristic image |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510971796.XA CN105654037B (en) | 2015-12-21 | 2015-12-21 | A kind of electromyography signal gesture identification method based on deep learning and characteristic image |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN105654037A true CN105654037A (en) | 2016-06-08 |
| CN105654037B CN105654037B (en) | 2019-05-21 |
Family
ID=56477765
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201510971796.XA Active CN105654037B (en) | 2015-12-21 | 2015-12-21 | A kind of electromyography signal gesture identification method based on deep learning and characteristic image |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN105654037B (en) |
Cited By (48)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106236336A (en) * | 2016-08-15 | 2016-12-21 | 中国科学院重庆绿色智能技术研究院 | A kind of myoelectric limb gesture and dynamics control 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 |
| CN106778785A (en) * | 2016-12-23 | 2017-05-31 | 东软集团股份有限公司 | Build the method for image characteristics extraction model and method, the device of image recognition |
| CN106780484A (en) * | 2017-01-11 | 2017-05-31 | 山东大学 | Robot interframe position and orientation estimation method based on convolutional neural networks Feature Descriptor |
| CN106980367A (en) * | 2017-02-27 | 2017-07-25 | 浙江工业大学 | A kind of gesture identification method based on myoelectricity topographic map |
| CN107463946A (en) * | 2017-07-12 | 2017-12-12 | 浙江大学 | A kind of type of merchandize detection method of combination template matches and deep learning |
| CN107480697A (en) * | 2017-07-12 | 2017-12-15 | 中国科学院计算技术研究所 | A kind of myoelectricity gesture identification method and system |
| CN107862249A (en) * | 2017-10-18 | 2018-03-30 | 太原理工大学 | A kind of bifurcated palm grain identification method and device |
| CN108052884A (en) * | 2017-12-01 | 2018-05-18 | 华南理工大学 | A kind of gesture identification method based on improvement residual error neutral net |
| CN108089693A (en) * | 2016-11-22 | 2018-05-29 | 比亚迪股份有限公司 | Gesture identification method and device, intelligence wearing terminal and server |
| CN108228285A (en) * | 2016-12-14 | 2018-06-29 | 中国航空工业集团公司西安航空计算技术研究所 | A kind of human-computer interaction instruction identification method multi-modal end to end |
| CN108345873A (en) * | 2018-03-22 | 2018-07-31 | 哈尔滨工业大学 | A kind of multiple degrees of freedom body motion information analytic method based on multilayer convolutional neural networks |
| CN108388348A (en) * | 2018-03-19 | 2018-08-10 | 浙江大学 | A kind of electromyography signal gesture identification method based on deep learning and attention mechanism |
| CN108453736A (en) * | 2018-03-22 | 2018-08-28 | 哈尔滨工业大学 | A kind of multiple degrees of freedom synchronization myoelectricity control method based on deep learning |
| CN108491077A (en) * | 2018-03-19 | 2018-09-04 | 浙江大学 | A kind of surface electromyogram signal gesture identification method for convolutional neural networks of being divided and ruled based on multithread |
| CN108509910A (en) * | 2018-04-02 | 2018-09-07 | 重庆邮电大学 | Deep learning gesture identification method based on fmcw radar signal |
| CN108536278A (en) * | 2017-03-03 | 2018-09-14 | 财团法人资讯工业策进会 | identification system and identification method |
| CN108921047A (en) * | 2018-06-12 | 2018-11-30 | 江西理工大学 | A kind of multi-model ballot mean value action identification method based on cross-layer fusion |
| CN109085918A (en) * | 2018-06-28 | 2018-12-25 | 天津大学 | Acupuncture needling manipulation training method based on myoelectricity |
| CN109643363A (en) * | 2016-06-15 | 2019-04-16 | 诺基亚技术有限公司 | The mthods, systems and devices of feature extraction and object detection |
| CN109662710A (en) * | 2018-12-06 | 2019-04-23 | 杭州电子科技大学 | A kind of EMG Feature Extraction based on convolutional neural networks |
| CN109726761A (en) * | 2018-12-29 | 2019-05-07 | 青岛海洋科学与技术国家实验室发展中心 | CNN evolvement method, AUV cluster working method, device and storage medium based on CNN |
| CN109730818A (en) * | 2018-12-20 | 2019-05-10 | 东南大学 | A kind of prosthetic hand control method based on deep learning |
| CN109800733A (en) * | 2019-01-30 | 2019-05-24 | 中国科学技术大学 | Data processing method and device, electronic equipment |
| CN109815984A (en) * | 2018-12-21 | 2019-05-28 | 中国电信集团工会上海市委员会 | A kind of user behavior identification system and method based on convolutional neural networks |
| CN109871805A (en) * | 2019-02-20 | 2019-06-11 | 中国电子科技集团公司第三十六研究所 | A kind of electromagnetic signal opener recognition methods |
| CN109923500A (en) * | 2016-08-22 | 2019-06-21 | 奇跃公司 | Augmented reality display device with deep learning sensor |
| CN109924977A (en) * | 2019-03-21 | 2019-06-25 | 西安交通大学 | A kind of surface electromyogram signal classification method based on CNN and LSTM |
| CN110175551A (en) * | 2019-05-21 | 2019-08-27 | 青岛科技大学 | A kind of sign Language Recognition Method |
| CN110337269A (en) * | 2016-07-25 | 2019-10-15 | 开创拉布斯公司 | The method and apparatus that user is intended to are inferred based on neuromuscular signals |
| CN110514924A (en) * | 2019-08-12 | 2019-11-29 | 武汉大学 | The Winding in Power Transformer Fault Locating Method of depth convolutional neural networks fusion visual recognition |
| CN110573883A (en) * | 2017-04-13 | 2019-12-13 | 美国西门子医学诊断股份有限公司 | Method and apparatus for determining tag count during sample characterization |
| CN110595811A (en) * | 2019-09-11 | 2019-12-20 | 浙江工业大学之江学院 | Method for constructing health state characteristic diagram of mechanical equipment |
| CN110598628A (en) * | 2019-09-11 | 2019-12-20 | 南京邮电大学 | Electromyographic signal hand motion recognition method based on integrated deep learning |
| CN110610172A (en) * | 2019-09-25 | 2019-12-24 | 南京邮电大学 | Myoelectric gesture recognition method based on RNN-CNN architecture |
| CN110908566A (en) * | 2018-09-18 | 2020-03-24 | 珠海格力电器股份有限公司 | Information processing method and device |
| CN111222398A (en) * | 2019-10-28 | 2020-06-02 | 南京航空航天大学 | Myoelectric signal decoding method based on time-frequency feature fusion |
| CN111616706A (en) * | 2020-05-20 | 2020-09-04 | 山东中科先进技术研究院有限公司 | Surface electromyogram signal classification method and system based on convolutional neural network |
| CN111783669A (en) * | 2020-07-02 | 2020-10-16 | 南京邮电大学 | Surface electromyographic signal classification and identification method for individual user |
| CN111833397A (en) * | 2020-06-08 | 2020-10-27 | 西安电子科技大学 | Data conversion method and device for orientation-finding target positioning |
| CN112733609A (en) * | 2020-12-14 | 2021-04-30 | 中山大学 | Domain-adaptive Wi-Fi gesture recognition method based on discrete wavelet transform |
| US11017296B2 (en) | 2018-08-22 | 2021-05-25 | Ford Global Technologies, Llc | Classifying time series image data |
| CN112950922A (en) * | 2021-01-26 | 2021-06-11 | 浙江得图网络有限公司 | Fixed-point returning method for sharing electric vehicle |
| CN113627401A (en) * | 2021-10-12 | 2021-11-09 | 四川大学 | Myoelectric gesture recognition method of feature pyramid network fused with double-attention machine system |
| CN113729738A (en) * | 2021-09-13 | 2021-12-03 | 武汉科技大学 | Construction method of multi-channel electromyographic feature image |
| CN114169375A (en) * | 2021-12-11 | 2022-03-11 | 福州大学 | Myoelectric gesture recognition method based on strength-independent robust features |
| CN114515146A (en) * | 2020-11-17 | 2022-05-20 | 北京机械设备研究所 | Intelligent gesture recognition method and system based on electrical measurement |
| CN116400812A (en) * | 2023-06-05 | 2023-07-07 | 中国科学院自动化研究所 | Emergency rescue gesture recognition method and device based on surface electromyographic signals |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103279734A (en) * | 2013-03-26 | 2013-09-04 | 上海交通大学 | Novel intelligent sign language translation and man-machine interaction system and use method thereof |
| US20130317648A1 (en) * | 2012-05-25 | 2013-11-28 | California Institute Of Technology | Biosleeve human-machine interface |
| CN103440498A (en) * | 2013-08-20 | 2013-12-11 | 华南理工大学 | Surface electromyogram signal identification method based on LDA algorithm |
| CN104899594A (en) * | 2014-03-06 | 2015-09-09 | 中国科学院沈阳自动化研究所 | Hand action identification method based on surface electromyography decomposition |
-
2015
- 2015-12-21 CN CN201510971796.XA patent/CN105654037B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20130317648A1 (en) * | 2012-05-25 | 2013-11-28 | California Institute Of Technology | Biosleeve human-machine interface |
| CN103279734A (en) * | 2013-03-26 | 2013-09-04 | 上海交通大学 | Novel intelligent sign language translation and man-machine interaction system and use method thereof |
| CN103440498A (en) * | 2013-08-20 | 2013-12-11 | 华南理工大学 | Surface electromyogram signal identification method based on LDA algorithm |
| CN104899594A (en) * | 2014-03-06 | 2015-09-09 | 中国科学院沈阳自动化研究所 | Hand action identification method based on surface electromyography decomposition |
Non-Patent Citations (3)
| Title |
|---|
| NAIK 等: "Multi run ICA and surface EMG-based signal processing system for recognising hand gestures", 《8TH IEEE INTERNATIONAL CONFERENCE ON COMPUTER AND INFORMATION TECHNOLOGY》 * |
| 赵漫丹等: "基于肌电信号层级分类的手部动作识别方法", 《北京生物医学工程》 * |
| 赵章琰: "表面肌电信号检测和处理中若干关键技术研究", 《万方数据库》 * |
Cited By (73)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109643363A (en) * | 2016-06-15 | 2019-04-16 | 诺基亚技术有限公司 | The mthods, systems and devices of feature extraction and object detection |
| 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 |
| CN110337269A (en) * | 2016-07-25 | 2019-10-15 | 开创拉布斯公司 | The method and apparatus that user is intended to are inferred based on neuromuscular signals |
| CN110337269B (en) * | 2016-07-25 | 2021-09-21 | 脸谱科技有限责任公司 | Method and apparatus for inferring user intent based on neuromuscular signals |
| CN106236336A (en) * | 2016-08-15 | 2016-12-21 | 中国科学院重庆绿色智能技术研究院 | A kind of myoelectric limb gesture and dynamics control method |
| CN109923500B (en) * | 2016-08-22 | 2022-01-04 | 奇跃公司 | Augmented reality display device with deep learning sensor |
| US11797078B2 (en) | 2016-08-22 | 2023-10-24 | Magic Leap, Inc. | Augmented reality display device with deep learning sensors |
| CN109923500A (en) * | 2016-08-22 | 2019-06-21 | 奇跃公司 | Augmented reality display device with deep learning sensor |
| CN108089693A (en) * | 2016-11-22 | 2018-05-29 | 比亚迪股份有限公司 | Gesture identification method and device, intelligence wearing terminal and server |
| CN108228285A (en) * | 2016-12-14 | 2018-06-29 | 中国航空工业集团公司西安航空计算技术研究所 | A kind of human-computer interaction instruction identification method multi-modal end to end |
| CN106778785B (en) * | 2016-12-23 | 2019-09-17 | 东软集团股份有限公司 | Construct the method for image Feature Selection Model and the method, apparatus of image recognition |
| CN106778785A (en) * | 2016-12-23 | 2017-05-31 | 东软集团股份有限公司 | Build the method for image characteristics extraction model and method, the device of image recognition |
| CN106780484A (en) * | 2017-01-11 | 2017-05-31 | 山东大学 | Robot interframe position and orientation estimation method based on convolutional neural networks Feature Descriptor |
| CN106980367B (en) * | 2017-02-27 | 2020-08-18 | 浙江工业大学 | Gesture recognition method based on electromyogram |
| CN106980367A (en) * | 2017-02-27 | 2017-07-25 | 浙江工业大学 | A kind of gesture identification method based on myoelectricity topographic map |
| CN108536278B (en) * | 2017-03-03 | 2021-03-19 | 财团法人资讯工业策进会 | Identification system and identification method |
| CN108536278A (en) * | 2017-03-03 | 2018-09-14 | 财团法人资讯工业策进会 | identification system and identification method |
| CN110573883A (en) * | 2017-04-13 | 2019-12-13 | 美国西门子医学诊断股份有限公司 | Method and apparatus for determining tag count during sample characterization |
| CN110573883B (en) * | 2017-04-13 | 2023-05-30 | 美国西门子医学诊断股份有限公司 | Method and apparatus for determining tag count during sample characterization |
| CN107463946B (en) * | 2017-07-12 | 2020-06-23 | 浙江大学 | Commodity type detection method combining template matching and deep learning |
| CN107480697B (en) * | 2017-07-12 | 2020-04-03 | 中国科学院计算技术研究所 | Myoelectric gesture recognition method and system |
| CN107463946A (en) * | 2017-07-12 | 2017-12-12 | 浙江大学 | A kind of type of merchandize detection method of combination template matches and deep learning |
| CN107480697A (en) * | 2017-07-12 | 2017-12-15 | 中国科学院计算技术研究所 | A kind of myoelectricity gesture identification method and system |
| CN107862249A (en) * | 2017-10-18 | 2018-03-30 | 太原理工大学 | A kind of bifurcated palm grain identification method and device |
| CN108052884A (en) * | 2017-12-01 | 2018-05-18 | 华南理工大学 | A kind of gesture identification method based on improvement residual error neutral net |
| CN108388348B (en) * | 2018-03-19 | 2020-11-24 | 浙江大学 | Myoelectric signal gesture recognition method based on deep learning and attention mechanism |
| CN108491077A (en) * | 2018-03-19 | 2018-09-04 | 浙江大学 | A kind of surface electromyogram signal gesture identification method for convolutional neural networks of being divided and ruled based on multithread |
| CN108388348A (en) * | 2018-03-19 | 2018-08-10 | 浙江大学 | A kind of electromyography signal gesture identification method based on deep learning and attention mechanism |
| CN108491077B (en) * | 2018-03-19 | 2020-06-16 | 浙江大学 | Surface electromyographic signal gesture recognition method based on multi-stream divide-and-conquer convolutional neural network |
| CN108453736A (en) * | 2018-03-22 | 2018-08-28 | 哈尔滨工业大学 | A kind of multiple degrees of freedom synchronization myoelectricity control method based on deep learning |
| CN108345873A (en) * | 2018-03-22 | 2018-07-31 | 哈尔滨工业大学 | A kind of multiple degrees of freedom body motion information analytic method based on multilayer convolutional neural networks |
| CN108509910B (en) * | 2018-04-02 | 2021-09-28 | 重庆邮电大学 | Deep learning gesture recognition method based on FMCW radar signals |
| CN108509910A (en) * | 2018-04-02 | 2018-09-07 | 重庆邮电大学 | Deep learning gesture identification method based on fmcw radar signal |
| CN108921047B (en) * | 2018-06-12 | 2021-11-26 | 江西理工大学 | Multi-model voting mean value action identification method based on cross-layer fusion |
| CN108921047A (en) * | 2018-06-12 | 2018-11-30 | 江西理工大学 | A kind of multi-model ballot mean value action identification method based on cross-layer fusion |
| CN109085918A (en) * | 2018-06-28 | 2018-12-25 | 天津大学 | Acupuncture needling manipulation training method based on myoelectricity |
| CN109085918B (en) * | 2018-06-28 | 2020-05-12 | 天津大学 | Myoelectricity-based acupuncture needle manipulation training method |
| US11017296B2 (en) | 2018-08-22 | 2021-05-25 | Ford Global Technologies, Llc | Classifying time series image data |
| CN110908566A (en) * | 2018-09-18 | 2020-03-24 | 珠海格力电器股份有限公司 | Information processing method and device |
| CN109662710A (en) * | 2018-12-06 | 2019-04-23 | 杭州电子科技大学 | A kind of EMG Feature Extraction based on convolutional neural networks |
| CN109730818A (en) * | 2018-12-20 | 2019-05-10 | 东南大学 | A kind of prosthetic hand control method based on deep learning |
| CN109815984A (en) * | 2018-12-21 | 2019-05-28 | 中国电信集团工会上海市委员会 | A kind of user behavior identification system and method based on convolutional neural networks |
| CN109726761A (en) * | 2018-12-29 | 2019-05-07 | 青岛海洋科学与技术国家实验室发展中心 | CNN evolvement method, AUV cluster working method, device and storage medium based on CNN |
| CN109800733A (en) * | 2019-01-30 | 2019-05-24 | 中国科学技术大学 | Data processing method and device, electronic equipment |
| CN109871805B (en) * | 2019-02-20 | 2020-10-27 | 中国电子科技集团公司第三十六研究所 | Electromagnetic signal open set identification method |
| CN109871805A (en) * | 2019-02-20 | 2019-06-11 | 中国电子科技集团公司第三十六研究所 | A kind of electromagnetic signal opener recognition methods |
| CN109924977A (en) * | 2019-03-21 | 2019-06-25 | 西安交通大学 | A kind of surface electromyogram signal classification method based on CNN and LSTM |
| CN110175551A (en) * | 2019-05-21 | 2019-08-27 | 青岛科技大学 | A kind of sign Language Recognition Method |
| CN110175551B (en) * | 2019-05-21 | 2023-01-10 | 青岛科技大学 | Sign language recognition method |
| CN110514924A (en) * | 2019-08-12 | 2019-11-29 | 武汉大学 | The Winding in Power Transformer Fault Locating Method of depth convolutional neural networks fusion visual recognition |
| CN110598628B (en) * | 2019-09-11 | 2022-08-02 | 南京邮电大学 | Electromyographic signal hand motion recognition method based on integrated deep learning |
| CN110595811A (en) * | 2019-09-11 | 2019-12-20 | 浙江工业大学之江学院 | Method for constructing health state characteristic diagram of mechanical equipment |
| CN110598628A (en) * | 2019-09-11 | 2019-12-20 | 南京邮电大学 | Electromyographic signal hand motion recognition method based on integrated deep learning |
| CN110610172A (en) * | 2019-09-25 | 2019-12-24 | 南京邮电大学 | Myoelectric gesture recognition method based on RNN-CNN architecture |
| CN110610172B (en) * | 2019-09-25 | 2022-08-12 | 南京邮电大学 | Myoelectric gesture recognition method based on RNN-CNN architecture |
| CN111222398A (en) * | 2019-10-28 | 2020-06-02 | 南京航空航天大学 | Myoelectric signal decoding method based on time-frequency feature fusion |
| CN111222398B (en) * | 2019-10-28 | 2023-04-18 | 南京航空航天大学 | Myoelectric signal decoding method based on time-frequency feature fusion |
| CN111616706A (en) * | 2020-05-20 | 2020-09-04 | 山东中科先进技术研究院有限公司 | Surface electromyogram signal classification method and system based on convolutional neural network |
| CN111616706B (en) * | 2020-05-20 | 2022-07-22 | 山东中科先进技术有限公司 | Surface electromyogram signal classification method and system based on convolutional neural network |
| CN111833397A (en) * | 2020-06-08 | 2020-10-27 | 西安电子科技大学 | Data conversion method and device for orientation-finding target positioning |
| CN111833397B (en) * | 2020-06-08 | 2022-11-29 | 西安电子科技大学 | Data conversion method and device for orientation-finding target positioning |
| CN111783669A (en) * | 2020-07-02 | 2020-10-16 | 南京邮电大学 | Surface electromyographic signal classification and identification method for individual user |
| CN114515146B (en) * | 2020-11-17 | 2024-03-22 | 北京机械设备研究所 | Intelligent gesture recognition method and system based on electrical measurement |
| CN114515146A (en) * | 2020-11-17 | 2022-05-20 | 北京机械设备研究所 | Intelligent gesture recognition method and system based on electrical measurement |
| CN112733609B (en) * | 2020-12-14 | 2023-08-18 | 中山大学 | Domain-adaptive Wi-Fi gesture recognition method based on discrete wavelet transform |
| CN112733609A (en) * | 2020-12-14 | 2021-04-30 | 中山大学 | Domain-adaptive Wi-Fi gesture recognition method based on discrete wavelet transform |
| CN112950922A (en) * | 2021-01-26 | 2021-06-11 | 浙江得图网络有限公司 | Fixed-point returning method for sharing electric vehicle |
| CN113729738A (en) * | 2021-09-13 | 2021-12-03 | 武汉科技大学 | Construction method of multi-channel electromyographic feature image |
| CN113729738B (en) * | 2021-09-13 | 2024-04-12 | 武汉科技大学 | Construction method of multichannel myoelectricity characteristic image |
| CN113627401A (en) * | 2021-10-12 | 2021-11-09 | 四川大学 | Myoelectric gesture recognition method of feature pyramid network fused with double-attention machine system |
| CN114169375A (en) * | 2021-12-11 | 2022-03-11 | 福州大学 | Myoelectric gesture recognition method based on strength-independent robust features |
| CN116400812A (en) * | 2023-06-05 | 2023-07-07 | 中国科学院自动化研究所 | Emergency rescue gesture recognition method and device based on surface electromyographic signals |
| CN116400812B (en) * | 2023-06-05 | 2023-09-12 | 中国科学院自动化研究所 | Emergency rescue gesture recognition method and device based on surface electromyographic signals |
Also Published As
| Publication number | Publication date |
|---|---|
| CN105654037B (en) | 2019-05-21 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN105654037A (en) | Myoelectric signal gesture recognition method based on depth learning and feature images | |
| CN105608432B (en) | A kind of gesture identification method based on instantaneous myoelectricity image | |
| CN108491077B (en) | Surface electromyographic signal gesture recognition method based on multi-stream divide-and-conquer convolutional neural network | |
| Li et al. | Boosting-based EMG patterns classification scheme for robustness enhancement | |
| CN105740823B (en) | Dynamic gesture track recognizing method based on depth convolutional neural networks | |
| Basu et al. | Use of artificial neural network in pattern recognition | |
| CN108983973B (en) | Control method of humanoid smart myoelectric artificial hand based on gesture recognition | |
| CN108388348A (en) | A kind of electromyography signal gesture identification method based on deep learning and attention mechanism | |
| CN106980367A (en) | A kind of gesture identification method based on myoelectricity topographic map | |
| CN106293057A (en) | Gesture identification method based on BP neutral net | |
| CN113205074B (en) | Gesture recognition method fusing multi-mode signals of myoelectricity and micro-inertia measurement unit | |
| CN103226388A (en) | Kinect-based handwriting method | |
| Huang et al. | Few-shot image and sentence matching via gated visual-semantic embedding | |
| CN110399846A (en) | A kind of gesture identification method based on multichannel electromyography signal correlation | |
| CN107742095A (en) | Chinese sign Language Recognition Method based on convolutional neural networks | |
| CN111562842B (en) | Virtual keyboard design method based on electromyographic signals | |
| Jia et al. | A multimodal emotion recognition model integrating speech, video and MoCAP | |
| Wen et al. | A two-dimensional matrix image based feature extraction method for classification of sEMG: A comparative analysis based on SVM, KNN and RBF-NN | |
| CN110298263A (en) | Real-time accurate and contactless gesture identification method and system based on RFID system | |
| Luo et al. | Features extraction and multi-classification of sEMG using a GPU-Accelerated GA/MLP hybrid algorithm | |
| CN116227560A (en) | Time sequence prediction model and method based on DTW-former | |
| Ding et al. | A convolutional transformer architecture for remaining useful life estimation | |
| AlKhuraym et al. | Arabic sign language recognition using lightweight cnn-based architecture | |
| Duan et al. | A Multi-Task Deep Learning Approach for Sensor-based Human Activity Recognition and Segmentation | |
| CN108345873A (en) | A kind of multiple degrees of freedom body motion information analytic method based on multilayer convolutional neural networks |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |