CN112401902B - Electrocardio identity recognition method and system based on neural network time-frequency analysis combination - Google Patents
Electrocardio identity recognition method and system based on neural network time-frequency analysis combination Download PDFInfo
- Publication number
- CN112401902B CN112401902B CN202011398560.9A CN202011398560A CN112401902B CN 112401902 B CN112401902 B CN 112401902B CN 202011398560 A CN202011398560 A CN 202011398560A CN 112401902 B CN112401902 B CN 112401902B
- Authority
- CN
- China
- Prior art keywords
- neural network
- frequency
- time
- layer
- feature vector
- 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.)
- Active
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
- A61B5/117—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
- A61B5/7203—Signal processing specially adapted for physiological signals or for diagnostic purposes for noise prevention, reduction or removal
-
- 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
- A61B5/7225—Details of analog processing, e.g. isolation amplifier, gain or sensitivity adjustment, filtering, baseline or drift compensation
-
- 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
- A61B5/7235—Details of waveform analysis
- A61B5/725—Details of waveform analysis using specific filters therefor, e.g. Kalman or adaptive filters
-
- 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
- A61B5/7235—Details of waveform analysis
- A61B5/7264—Classification of physiological signals or data, e.g. using neural networks, statistical classifiers, expert systems or fuzzy systems
-
- 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
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/044—Recurrent networks, e.g. Hopfield networks
-
- 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
- Health & Medical Sciences (AREA)
- Engineering & Computer Science (AREA)
- Life Sciences & Earth Sciences (AREA)
- Physics & Mathematics (AREA)
- General Health & Medical Sciences (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Artificial Intelligence (AREA)
- Molecular Biology (AREA)
- Signal Processing (AREA)
- Veterinary Medicine (AREA)
- Pathology (AREA)
- Public Health (AREA)
- Animal Behavior & Ethology (AREA)
- Surgery (AREA)
- Heart & Thoracic Surgery (AREA)
- Medical Informatics (AREA)
- Theoretical Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Physiology (AREA)
- Psychiatry (AREA)
- Mathematical Physics (AREA)
- Evolutionary Computation (AREA)
- Data Mining & Analysis (AREA)
- Software Systems (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Computing Systems (AREA)
- Computational Linguistics (AREA)
- Fuzzy Systems (AREA)
- Power Engineering (AREA)
- Measurement And Recording Of Electrical Phenomena And Electrical Characteristics Of The Living Body (AREA)
Abstract
The invention provides an electrocardiogram identity recognition method and system based on neural network time-frequency analysis combination. Wherein, the method comprises preprocessing the electrocardiosignal; performing time-frequency conversion on the preprocessed electrocardiosignals, extracting a cepstrum coefficient of each heart beat as an identification feature to obtain an initial frequency domain feature vector, and inputting the initial frequency domain feature vector to a first neural network model to obtain an abstracted frequency domain feature vector; extracting the time domain characteristics of the preprocessed electrocardiosignals based on a second neural network model to obtain time domain characteristic vectors; and splicing the abstracted frequency domain characteristic vector and the time domain characteristic vector, inputting the final full-connection layer with the activation function of softmax for identification, and outputting an identification result.
Description
Technical Field
The invention relates to the field of biological feature recognition, in particular to an electrocardio identity recognition method and system based on the combination of time-frequency analysis of a neural network.
Background
The statements in this section merely provide background information related to the present disclosure and may not necessarily constitute prior art.
Electrocardiosignals are a biological feature with high safety, which has a relatively obvious difference among different individuals and can be used for identification in addition to being used as a basis for disease diagnosis. In addition, the electrocardiosignals are generated in the human body and are difficult to steal or forge, so the identification based on the electrocardiosignals (called electrocardio identification for short) has good development prospect. Common methods for extracting the electrocardiosignal features include a filter bank method, a threshold discrimination method, wavelet transformation and the like.
In addition, with the development of deep learning, there are various technical methods for solving the problem of identity recognition by using a deep network, such as the common models of BP neural network, RNN, LSTM, etc., but these network models all have certain drawbacks. The inventor finds that the BP neural network is very easy to fall into a local minimum value, RNN cannot solve the long-term dependence problem, LSTM cannot perform parallel computation, and resources consumed by training are huge.
Disclosure of Invention
In order to solve at least one technical problem in the background art, the invention provides an electrocardiogram identity recognition method and system based on the combination of time-frequency analysis of a neural network, which can further improve the recognition performance and efficiency of the electrocardiogram identity recognition system.
In order to achieve the purpose, the invention adopts the following technical scheme:
the invention provides an electrocardiogram identity recognition method based on the combination of time-frequency analysis of a neural network.
An electrocardio identity recognition method based on the combination of time-frequency analysis of a neural network comprises the following steps:
preprocessing the electrocardiosignal;
performing time-frequency conversion on the preprocessed electrocardiosignals, extracting a cepstrum coefficient of each heart beat as an identification feature to obtain an initial frequency domain feature vector, and inputting the initial frequency domain feature vector to a first neural network model to obtain an abstracted frequency domain feature vector; extracting the time domain characteristics of the preprocessed electrocardiosignals based on a second neural network model to obtain time domain characteristic vectors;
and splicing the abstracted frequency domain characteristic vector and the time domain characteristic vector, inputting the final full-connection layer with the activation function of softmax for identification, and outputting an identification result.
The second aspect of the invention provides an electrocardiogram identity recognition system based on the combination of time-frequency analysis of a neural network.
An electrocardio identity recognition system based on neural network time-frequency analysis combination comprises:
the preprocessing module is used for preprocessing the electrocardiosignal;
the feature extraction module is used for performing time-frequency conversion on the preprocessed electrocardiosignals, extracting a cepstrum coefficient of each heartbeat as an identification feature to obtain an initial frequency domain feature vector, and inputting the initial frequency domain feature vector to the first neural network model to obtain an abstracted frequency domain feature vector; extracting the time domain characteristics of the preprocessed electrocardiosignals based on a second neural network model to obtain time domain characteristic vectors;
and the identification module is used for splicing the abstracted frequency domain characteristic vector and the time domain characteristic vector, inputting the spliced frequency domain characteristic vector and time domain characteristic vector into a full connection layer with the final activation function of softmax for identification, and outputting an identification result.
A third aspect of the invention provides a computer-readable storage medium.
A computer-readable storage medium, on which a computer program is stored, which, when being executed by a processor, carries out the steps of the method for cardiac electrical identity recognition based on a combination of time-frequency analysis and neural network analysis as described above.
A fourth aspect of the invention provides a computer apparatus.
A computer device comprises a memory, a processor and a computer program stored in the memory and operable on the processor, wherein the processor executes the program to implement the steps of the method for identifying an electrocardiogram based on the combination of time-frequency analysis and neural network.
Compared with the prior art, the invention has the beneficial effects that:
the electrocardiosignal is subjected to feature extraction from the angles of time domain and frequency domain, so that the unique features of each type of signal can be more comprehensively found, the interference of noise is avoided, and the good identification performance is achieved; the improved TDNN model is selected for time domain feature extraction, compared with other traditional models, the time domain feature extraction method is more stable, the convergence speed is extremely high, and less time resources and computing resources can be consumed, so that the high efficiency of the identification process is ensured.
Advantages of additional aspects of the invention will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of the invention.
Drawings
The accompanying drawings, which are incorporated in and constitute a part of this specification, are included to provide a further understanding of the invention, and are incorporated in and constitute a part of this specification, illustrate exemplary embodiments of the invention and together with the description serve to explain the invention and not to limit the invention.
FIG. 1 is a flow chart of an electrocardiogram identity recognition method based on the combination of time-frequency analysis of a neural network according to an embodiment of the present invention;
FIG. 2 is a filter bank distribution plot for frequency converting Mel to linear frequency;
FIG. 3 is a schematic diagram of a Time Delay Neural Network (TDNN);
fig. 4 is an overall network model.
Detailed Description
The invention is further described with reference to the following figures and examples.
It is to be understood that the following detailed description is exemplary and is intended to provide further explanation of the invention as claimed. Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs.
It is noted that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to be limiting of exemplary embodiments according to the invention. As used herein, the singular forms "a", "an" and "the" are intended to include the plural forms as well, and it should be understood that when the terms "comprises" and/or "comprising" are used in this specification, they specify the presence of stated features, steps, operations, devices, components, and/or combinations thereof, unless the context clearly indicates otherwise.
Example one
As shown in fig. 1, the method for identifying an electrocardiogram based on the combination of time-frequency analysis and neural network of this embodiment includes:
s101: the electrocardiosignal is preprocessed.
Specifically, the acquired electrocardiosignals are preprocessed, and the signals are passed through a fourth-order low-pass Butterworth filter with a cut-off frequency of 40Hz to remove noises such as myoelectric interference and baseline drift, so that an R peak in the electrocardiosignals is detected and divided into a plurality of heartbeats.
The acquired original electrocardiosignals have noises such as baseline drift, electromyographic interference and the like, the method removes the noises based on a four-order low-pass Butterworth filter with the cut-off frequency of 40Hz, and then segments the heartbeat cycle based on the R peak of the electrocardiosignals. The R peak is the most obvious position of waveform change in a heartbeat cycle, is not easy to be interfered by noise, and is easy to detect an accurate position. The number of the heart beat sampling points obtained after the division may be different, the sampling frequencies of different data sets may also be different, and each heart beat needs to be resampled, so that the number of the sampling points of each heart beat is 256.
The specific process of detecting the position of the R peak by using the Pan-Tompkins algorithm comprises filtering, derivation, squaring, integration, adaptive threshold and searching, and specifically comprises the following steps:
first, a low pass filter with a cutoff frequency of 11Hz and a 35dB attenuation at 60Hz is used to remove noise, and then a high pass filter with a cutoff frequency of 5Hz is used to eliminate baseline drift;
solving a first derivative of the signal to remove a direct current component and enhancing the slope of the QRS wave;
carrying out square operation on the derived signals to change the samples into nonnegative numbers, and further enhancing the slope of the QRS wave;
integrating the signal through a smoothing window to enable the output to be smoother;
and finally, searching the R wave crest in the signal through an adaptive threshold value.
S102: performing time-frequency conversion on the preprocessed electrocardiosignals, extracting a cepstrum coefficient of each heart beat as an identification feature to obtain an initial frequency domain feature vector, and inputting the initial frequency domain feature vector to a first neural network model to obtain an abstracted frequency domain feature vector; and extracting the time domain characteristics of the preprocessed electrocardiosignals based on a second neural network model to obtain time domain characteristic vectors.
Specifically, the signal is subjected to time-frequency conversion and frequency-domain analysis, and the method has good self-adaptability. The cepstrum coefficient of each heart beat is extracted based on frequency domain analysis to serve as an identification feature, the cepstrum coefficient of each frame is extracted aiming at each heart beat divided into multiple frames, and the cepstrum coefficients of the frames are connected to obtain a final 72-dimensional feature vector.
The electrocardiosignals belong to non-stationary signals, the Fourier transform cannot be directly carried out, the electrocardiosignals can be subjected to framing, each frame of signals can be regarded as short-time stationary signals, and the stationary signals can be subjected to Fourier transform to better carry out spectrum analysis. Each frame signal length is 14 sampling points, and the adjacent two frames have a frame stack of 12 sampling points, so that one heart beat can be divided into 122 frames, and a Hamming window is respectively added to each frame for smoothing discontinuous changes at the cut-off part and reducing frequency spectrum leakage. If one of the frame signals after the framing is S (n), the windowed signal is S' (n) ═ S (n) × w (n), and the expression of the hamming window is as follows:
where N is the length of the signal per frame.
For discrete signals in a computer, the frequency spectrum of the signals can be acquired very efficiently by using Fast Fourier Transform (FFT), the number of transform points needs to be set before transformation, the number of the transform points needs to be an integer power of 2, the higher the number of the transform points is, the higher the resolution of the frequency spectrum is, and the method is more beneficial to distinguishing different frequency components in the frequency spectrum. Fruit of Chinese wolfberryExperiments show that the FFT of 512 points is respectively carried out on each windowed frame signal S' (n) so as to obtain the frequency spectrum of each frame more quickly under the condition of keeping high resolution, and the power spectrum | X of the signal is obtained by taking the square of the modulus of the obtained frequency spectruma(k)|2。
A Mel filter bank h (k) is constructed consisting of several (e.g. 40) triangular filters, each with a response of 1 at the centre frequency, evenly distributed over the Mel frequency. The linear frequency to Mel frequency mapping formula is:
the filter bank distribution for conversion of Mel frequency to linear frequency according to this formula is shown in figure 2.
The specific calculation method of the filter bank h (k) is as follows:
1) determining the lowest (generally 0Hz) and highest (generally sampling rate) linear frequency of the electrocardiosignals and the number of Mel filters;
2) calculating the Mel frequency corresponding to the lowest and highest linear frequency;
3) calculating the distance between the center frequencies of two adjacent Mel filters: (highest Mel frequency-lowest Mel frequency)/(number of filters + 1);
4) converting each central Mel frequency to a linear frequency;
5) calculate the index of the frequency corresponding to the mid-point of the FFT.
The power spectrum | X of each frame signala(k)|2Obtaining the logarithmic energy output by each filter through a filter bank H (k) and taking a logarithm calculation, wherein the calculation formula is as follows:
where M is the number of filters, Hm(k) Is the mth triangular filter. The number of filters is 40, so that 40 logarithmic energies can be obtained per frame of signal.
Performing DCT transformation on the obtained logarithmic energy S (m), wherein the formula is as follows:
and obtaining an L-order coefficient, wherein L is 23, 23-order cepstrum coefficients can be obtained from each frame of signal, 122 frames are obtained in total for each heartbeat, and 2806-dimensional feature vectors can be obtained after splicing.
S103: and splicing the abstracted frequency domain characteristic vector and the time domain characteristic vector, inputting the final full-connection layer with the activation function of softmax for identification, and outputting an identification result.
The constructed time-frequency network model is divided into a time domain feature extraction part and a frequency domain feature extraction part. The specific process of identification is shown in figure 1.
The present embodiment uses a modified time-delay neural network (TDNN) to extract the time-domain features of the electrocardiosignal. TDNN is a one-dimensional convolution network, and the network model is as shown in FIG. 3:
the input of the L1 layer is 1-dimensional original heart beat, the window length of a convolution kernel is 5, the step length is 1, and 64-dimensional feature map is obtained by each convolution; the L2 layer uses the idea of cavity convolution in the two-dimensional CNN for reference, and performs one-dimensional cavity convolution, namely discontinuous convolution, on the feature map output by the L1, so that the output obtained by each convolution can embody the features earlier and later than the current moment, the abstract capability of feature extraction is improved, and the output is 64-dimensional feature vectors; the L3 layers are also discontinuous convolutions, but the convolution interval is larger than the L2 layers, each convolution outputting a 128-dimensional feature vector, where each layer uses relu as the activation function and adds dropout to prevent overfitting. The final pooling layer is to calculate the average value and the standard deviation of the output of the L3 layer in the time dimension, so as to obtain 128 average values and 128 standard deviations, and the two are spliced together to finally obtain a 256-dimensional time domain feature vector.
Inputting the obtained 2806-dimensional frequency domain feature vector into a 3-layer BP neural network, wherein according to an experimental result, the input layer of the BP neural network is 2806 neurons, two hidden layers are 1024 and 512 neurons respectively, an activation function is equal to relu, and finally an abstracted frequency domain feature vector with 512 dimensions is obtained.
And splicing the time domain characteristic vectors and the frequency domain characteristic vectors, inputting the final full-connection layer with the activation function of softmax for recognition training, and using a cross entropy function as a loss function. The overall network model is shown in fig. 4.
And giving a heart beat, inputting the heart beat into the trained network, automatically extracting the characteristics of the heart beat in time domain and frequency domain by the network, and finally outputting a recognition result.
The method extracts the features of the electrocardiosignals from the angles of time domain and frequency domain, can more comprehensively discover the unique features of each type of signals, avoids the interference of noise, and has good identification performance; the improved TDNN model is selected for time domain feature extraction, compared with other traditional models, the time domain feature extraction method is more stable, the convergence speed is extremely high, and less time resources and computing resources can be consumed, so that the high efficiency of the identification process is ensured.
Example two
The embodiment provides an electrocardiogram identity recognition system based on the combination of time-frequency analysis of a neural network, which comprises:
the preprocessing module is used for preprocessing the electrocardiosignal;
the feature extraction module is used for performing time-frequency conversion on the preprocessed electrocardiosignals, extracting a cepstrum coefficient of each heartbeat as an identification feature to obtain an initial frequency domain feature vector, and inputting the initial frequency domain feature vector to the first neural network model to obtain an abstracted frequency domain feature vector; extracting the time domain characteristics of the preprocessed electrocardiosignals based on a second neural network model to obtain time domain characteristic vectors;
and the identification module is used for splicing the abstracted frequency domain characteristic vector and the time domain characteristic vector, inputting the spliced frequency domain characteristic vector and time domain characteristic vector into a full connection layer with the final activation function of softmax for identification, and outputting an identification result.
Each module in the electrocardiogram identity recognition system based on the combination of the time-frequency analysis of the neural network in the embodiment corresponds to each step in the electrocardiogram identity recognition method based on the combination of the time-frequency analysis of the neural network in the embodiment one, and the description is omitted here.
EXAMPLE III
The present embodiment provides a computer-readable storage medium, on which a computer program is stored, and when the program is executed by a processor, the program implements the steps in the method for identifying an electrocardiogram identity based on the combination of time-frequency analysis and neural network as described above.
Example four
The embodiment provides a computer device, which includes a memory, a processor, and a computer program stored in the memory and capable of running on the processor, wherein the processor implements the steps of the method for identifying an electrocardiogram identity based on the combination of time-frequency analysis and neural network when executing the program.
As will be appreciated by one skilled in the art, embodiments of the present invention may be provided as a method, system, or computer program product. Accordingly, the present invention may take the form of a hardware embodiment, a software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, optical storage, and the like) having computer-usable program code embodied therein.
The present invention is described with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each flow and/or block of the flow diagrams and/or block diagrams, and combinations of flows and/or blocks in the flow diagrams and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general purpose computer, special purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instruction means which implement the function specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
It will be understood by those skilled in the art that all or part of the processes of the methods of the embodiments described above can be implemented by a computer program, which can be stored in a computer-readable storage medium, and when executed, can include the processes of the embodiments of the methods described above. The storage medium may be a magnetic disk, an optical disk, a Read-Only Memory (ROM), a Random Access Memory (RAM), or the like.
The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention, and various modifications and changes may be made by those skilled in the art. Any modification, equivalent replacement, or improvement made within the spirit and principle of the present invention should be included in the protection scope of the present invention.
Claims (8)
1. An electrocardio identity recognition method based on the combination of time-frequency analysis of a neural network is characterized by comprising the following steps:
preprocessing the electrocardiosignal;
performing time-frequency conversion on the preprocessed electrocardiosignals, extracting a cepstrum coefficient of each heart beat as an identification feature to obtain an initial frequency domain feature vector, and inputting the initial frequency domain feature vector to a first neural network model to obtain an abstracted frequency domain feature vector; extracting the time domain characteristics of the preprocessed electrocardiosignals based on a second neural network model to obtain time domain characteristic vectors;
the second neural network model is a time delay neural network, and the time delay neural network is a one-dimensional convolution network;
the specific process of feature extraction is as follows:
extracting time domain features of the electrocardiosignals by using an improved time delay neural network, wherein an L1 layer is input into a 1-dimensional original heart beat, an L2 layer performs one-dimensional cavity convolution, namely discontinuous convolution, on feature maps output by L1, so that the output obtained by each convolution can embody the features before and after the current moment, an L3 layer is also discontinuous convolution, but the convolution interval is larger than that of the L2 layer, each layer uses relu as an activation function and adds dropout to prevent overfitting, a final pooling layer calculates an average value and a standard deviation of the output of the L3 layer in a time dimension, and the average value and the standard deviation are spliced together to finally obtain a time domain feature vector;
inputting the obtained frequency domain feature vector into a 3-layer BP neural network, wherein the activation function is equal to relu, and the abstracted frequency domain feature vector is obtained;
and splicing the abstracted frequency domain characteristic vector and the time domain characteristic vector, inputting the final full-connection layer with the activation function of softmax for identification, and outputting an identification result.
2. The method as claimed in claim 1, wherein in the process of preprocessing the electrocardiosignal, the position of the R peak is detected by using Pan-Tompkins algorithm, and a heartbeat is obtained by selecting a signal of a sampling point between two R peaks based on the two R peaks.
3. The method of claim 2, wherein each heartbeat is resampled such that the number of samples taken from each heartbeat is the same.
4. The method of claim 1, wherein the Mel frequency is transformed to the filter bank distribution on the linear frequency according to a mapping formula of the linear frequency and the Mel frequency.
5. The method for recognizing the electrocardiogram identity based on the combination of time-frequency analysis and neural network as claimed in claim 4, wherein the specific calculation method of the filter bank is as follows:
determining the lowest and highest linear frequencies of the electrocardiosignals and the number of Mel filters;
calculating Mel frequency corresponding to lowest and highest linear frequency;
calculating the distance between the center frequencies of two adjacent Mel filters: (highest Mel frequency-lowest Mel frequency)/(number of filters + 1);
converting each central Mel frequency into a linear frequency;
the index of the frequency corresponding to the mid-point of the FFT is calculated.
6. An electrocardio identity recognition system based on neural network time frequency analysis combines together which characterized in that includes:
the preprocessing module is used for preprocessing the electrocardiosignal;
the feature extraction module is used for performing time-frequency conversion on the preprocessed electrocardiosignals, extracting a cepstrum coefficient of each heartbeat as an identification feature to obtain an initial frequency domain feature vector, and inputting the initial frequency domain feature vector to the first neural network model to obtain an abstracted frequency domain feature vector; extracting the time domain characteristics of the preprocessed electrocardiosignals based on a second neural network model to obtain time domain characteristic vectors; the second neural network model is a time delay neural network, and the time delay neural network is a one-dimensional convolution network;
the specific process of feature extraction is as follows:
extracting time domain features of the electrocardiosignals by using an improved time delay neural network, wherein an L1 layer is input into a 1-dimensional original heart beat, an L2 layer performs one-dimensional cavity convolution, namely discontinuous convolution, on feature maps output by L1, so that the output obtained by each convolution can embody the features before and after the current moment, an L3 layer is also discontinuous convolution, but the convolution interval is larger than that of the L2 layer, each layer uses relu as an activation function and adds dropout to prevent overfitting, a final pooling layer calculates an average value and a standard deviation of the output of the L3 layer in a time dimension, and the average value and the standard deviation are spliced together to finally obtain a time domain feature vector;
inputting the obtained frequency domain feature vector into a 3-layer BP neural network, wherein the activation function is equal to relu, and the abstracted frequency domain feature vector is obtained;
and the identification module is used for splicing the abstracted frequency domain characteristic vector and the time domain characteristic vector, inputting the spliced frequency domain characteristic vector and time domain characteristic vector into a full connection layer with the final activation function of softmax for identification, and outputting an identification result.
7. A computer-readable storage medium, on which a computer program is stored, which, when being executed by a processor, carries out the steps of the method for cardiac electrical identity recognition based on a combination of neural network time-frequency analysis according to any one of claims 1 to 5.
8. Computer device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, wherein the processor when executing the program performs the steps of the method for cardiac electrical identity recognition based on neural network time-frequency analysis in combination according to any of claims 1-5.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011398560.9A CN112401902B (en) | 2020-12-04 | 2020-12-04 | Electrocardio identity recognition method and system based on neural network time-frequency analysis combination |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011398560.9A CN112401902B (en) | 2020-12-04 | 2020-12-04 | Electrocardio identity recognition method and system based on neural network time-frequency analysis combination |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN112401902A CN112401902A (en) | 2021-02-26 |
| CN112401902B true CN112401902B (en) | 2022-02-08 |
Family
ID=74829828
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202011398560.9A Active CN112401902B (en) | 2020-12-04 | 2020-12-04 | Electrocardio identity recognition method and system based on neural network time-frequency analysis combination |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112401902B (en) |
Families Citing this family (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113197583A (en) * | 2021-05-11 | 2021-08-03 | 广元市中心医院 | Electrocardiogram waveform segmentation method based on time-frequency analysis and recurrent neural network |
| CN114098691A (en) * | 2022-01-26 | 2022-03-01 | 之江实验室 | Pulse wave identity authentication method, device and medium based on Gaussian mixture model |
| CN114469139A (en) * | 2022-01-27 | 2022-05-13 | 中国农业银行股份有限公司 | Electroencephalogram signal recognition model training method, electroencephalogram signal recognition device and medium |
| CN115670446A (en) * | 2022-11-10 | 2023-02-03 | 福州大学 | Identity recognition method based on bioelectric signal fusion |
| CN115980189A (en) * | 2023-03-06 | 2023-04-18 | 中铁西南科学研究院有限公司 | Tunnel lining void detection method and system based on shock echo signal |
Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111582138A (en) * | 2020-04-30 | 2020-08-25 | 山东大学 | Electrocardio identity recognition method and system based on frequency domain cepstrum coefficient characteristics |
Family Cites Families (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| DE102008025674B4 (en) * | 2008-05-29 | 2021-01-21 | Tom Tec Imaging Systems Gmbh | Method, device and computer program product for recording medical images of a moving object |
| CN106492477A (en) * | 2015-09-03 | 2017-03-15 | 马铿杰 | Single-power drives the model plane of multiple propeller rotation |
| CN105468951B (en) * | 2015-11-17 | 2019-08-06 | 安徽华米信息科技有限公司 | Method and device, the wearable device of identification are carried out by ecg characteristics |
| CN110192864B (en) * | 2019-06-12 | 2020-09-22 | 北京交通大学 | Cross-domain electrocardiogram biological characteristic identity recognition method |
-
2020
- 2020-12-04 CN CN202011398560.9A patent/CN112401902B/en active Active
Patent Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111582138A (en) * | 2020-04-30 | 2020-08-25 | 山东大学 | Electrocardio identity recognition method and system based on frequency domain cepstrum coefficient characteristics |
Non-Patent Citations (2)
| Title |
|---|
| Robust ECG Biometrics by Fusing Temporal and Cepstral Information;Ming Li等;《IEEE》;20101231;第1316-1329页 * |
| 基于希尔伯特振动分解和卷积神经网络的融合特征心电识别算法;黄润新等;《通信技术》;20200410(第04期);第953-962页 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN112401902A (en) | 2021-02-26 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN112401902B (en) | Electrocardio identity recognition method and system based on neural network time-frequency analysis combination | |
| US20170143226A1 (en) | Action Recognition Method and Device Based on Surface Electromyography Signal | |
| Bhaskar | Performance analysis of support vector machine and neural networks in detection of myocardial infarction | |
| CN107688553B (en) | Method for detecting electrocardiographic waveform characteristics based on wavelet transform and logistic regression algorithm | |
| CN106951753B (en) | Electrocardiosignal authentication method and device | |
| CN106377247B (en) | Arrhythmia classification method based on feature selecting | |
| CN104706321A (en) | MFCC heart sound type recognition method based on improvement | |
| CN104398252A (en) | Electrocardiogram signal processing method and device | |
| CN105147252A (en) | Heart disease recognition and assessment method | |
| US10163528B2 (en) | Determining user-interested information based on wearable device | |
| CN111202512A (en) | Electrocardiogram classification method and device based on wavelet transformation and DCNN | |
| CN116898455B (en) | Sleep electroencephalogram signal detection method and system based on deep learning model | |
| CN108647584B (en) | Arrhythmia identification and classification method based on sparse representation and neural network | |
| CN106485213A (en) | A kind of utilization electrocardiosignal carries out the feature extracting method of automatic identification | |
| CN111582138A (en) | Electrocardio identity recognition method and system based on frequency domain cepstrum coefficient characteristics | |
| CN104515905A (en) | CQT multi-resolution based subject EEG signal adaptive spectrum analysis method | |
| CN110037683A (en) | The improvement convolutional neural networks and its training method of rhythm of the heart type for identification | |
| CN113116361A (en) | Sleep staging method based on single-lead electroencephalogram | |
| CN113128384B (en) | Brain-computer interface software key technical method of cerebral apoplexy rehabilitation system based on deep learning | |
| CN108682433A (en) | The heart sound kind identification method of first-order difference coefficient based on MFCC | |
| Ariff et al. | Study of adam and adamax optimizers on alexnet architecture for voice biometric authentication system | |
| Jaiswal et al. | Artificial neural network for ECG classification | |
| CN110169766A (en) | A kind of cardiogram wave detection method, apparatus based on wavelet transformation, terminal device | |
| CN108937857A (en) | A kind of identification and appraisal procedure of cardiechema signals | |
| CN110674738A (en) | Method for identifying left and right hand electroencephalogram signals and related device |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |