CN110192864A - A kind of cross-domain electrocardiogram biometric identity recognition methods - Google Patents
A kind of cross-domain electrocardiogram biometric identity recognition methods Download PDFInfo
- Publication number
- CN110192864A CN110192864A CN201910505119.7A CN201910505119A CN110192864A CN 110192864 A CN110192864 A CN 110192864A CN 201910505119 A CN201910505119 A CN 201910505119A CN 110192864 A CN110192864 A CN 110192864A
- Authority
- CN
- China
- Prior art keywords
- ecg
- data sample
- sample signal
- domain
- data
- 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
- 238000000034 method Methods 0.000 title claims abstract description 36
- 238000013527 convolutional neural network Methods 0.000 claims abstract description 22
- 238000012549 training Methods 0.000 claims abstract description 12
- 238000012545 processing Methods 0.000 claims description 6
- 238000001228 spectrum Methods 0.000 claims description 6
- 230000004913 activation Effects 0.000 claims description 5
- 238000001914 filtration Methods 0.000 claims description 4
- 238000005070 sampling Methods 0.000 claims description 4
- 238000004364 calculation method Methods 0.000 claims description 3
- 210000002569 neuron Anatomy 0.000 claims description 3
- 230000001131 transforming effect Effects 0.000 claims description 2
- 238000000605 extraction Methods 0.000 abstract description 10
- 230000002123 temporal effect Effects 0.000 abstract 2
- 238000013480 data collection Methods 0.000 abstract 1
- 238000010586 diagram Methods 0.000 description 8
- 230000006870 function Effects 0.000 description 6
- 230000008859 change Effects 0.000 description 3
- 238000012360 testing method Methods 0.000 description 3
- 238000013528 artificial neural network Methods 0.000 description 2
- 230000003542 behavioural effect Effects 0.000 description 2
- 238000013135 deep learning Methods 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 238000010801 machine learning Methods 0.000 description 2
- 230000011218 segmentation Effects 0.000 description 2
- 238000013475 authorization Methods 0.000 description 1
- 239000012141 concentrate Substances 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000018109 developmental process Effects 0.000 description 1
- 230000005021 gait Effects 0.000 description 1
- 230000007246 mechanism Effects 0.000 description 1
- 238000012544 monitoring process Methods 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 238000007781 pre-processing Methods 0.000 description 1
- 230000008569 process Effects 0.000 description 1
- 230000000750 progressive effect Effects 0.000 description 1
- 230000004044 response Effects 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 230000009466 transformation Effects 0.000 description 1
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
- A61B5/24—Detecting, measuring or recording bioelectric or biomagnetic signals of the body or parts thereof
- A61B5/316—Modalities, i.e. specific diagnostic methods
- A61B5/318—Heart-related electrical modalities, e.g. electrocardiography [ECG]
Landscapes
- Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Heart & Thoracic Surgery (AREA)
- Medical Informatics (AREA)
- Biophysics (AREA)
- Pathology (AREA)
- Engineering & Computer Science (AREA)
- Biomedical Technology (AREA)
- Veterinary Medicine (AREA)
- Physics & Mathematics (AREA)
- Molecular Biology (AREA)
- Surgery (AREA)
- Animal Behavior & Ethology (AREA)
- General Health & Medical Sciences (AREA)
- Public Health (AREA)
- Cardiology (AREA)
- Measurement And Recording Of Electrical Phenomena And Electrical Characteristics Of The Living Body (AREA)
Abstract
The present invention provides a kind of cross-domain electrocardiogram biometric identity recognition methods.This method comprises: establishing ECG data collection, is concentrated from ECG data and choose data sample signal；Temporal signatures, frequency domain character and energy domain feature extraction are carried out to data sample signal；ECG biometric identity identification model is constructed using improved CNN convolutional neural networks, the temporal signatures of data sample signal, frequency domain character and energy characteristic of field are input to ECG biometric identity identification model, obtain the ECG biometric identity recognition result of data sample signal.Even if the interval between cross-domain ECG biometric identity recognition methods training time of the invention and application time is sufficiently large, the accuracy rate of ECG biometric identity identification can also be effectively improved.
Description
Technical Field
The invention relates to the technical field of biological characteristic identity recognition, in particular to a cross-domain electrocardiogram biological characteristic identity recognition method.
Background
Biometric identity recognition is a technology for biometric authentication using physiological or behavioral characteristics of a human body, is one of the technologies in the rapidly developing information security field, has been widely used in medical care, finance, security, monitoring, and the like, and gradually enters more fields of human activities. Studies have shown that there are many physiological and behavioral features that can be identified as biometric identities, for example: fingerprint, iris, face, ECG (electrocardiogram) and handwriting, gait, etc. With the annual increase of the ECG data volume and the continuous development of the fields of machine learning, deep learning and the like, the ECG biological feature identification method based on machine learning and deep learning attracts people's attention.
ECG, as a biometric identification method with high resistance to forgery attacks, has an identification accuracy of about 95%. However, embodiments of the present invention find that if ECG is applied in a real environment where there is a significant interval between the training time and the application time, the accuracy will drop dramatically to 40%. Existing work overlooks the practical application scenario of biometric identification, and biometric collection and application are typically distributed over a long time scale, where ECG is highly dynamic.
In order to solve the above problems, it is proposed to use a non-reference random starting point fixed-length segmentation method to perform sample acquisition so as to adapt to training of the neural network. The technique mainly obtains samples by a method of randomly placing starting points, and is irrelevant to the position of an R peak, so that a large number of samples can be obtained in a finite ECG record. Liu et al propose a sample acquisition algorithm for non-reference random starting point fixed-length segmentation. Compared with the traditional QRS wave-based sample acquisition method, the QRS wave-based sample acquisition method has the advantages that the starting point can randomly wander on the time axis, so that a large number of samples which are similar to the original record in distribution are acquired. The main disadvantage of this algorithm is that the parameters involved therein, especially the length of the single sample and the training duration, are not adjusted, and different parameter settings have a great influence on the final recognition result.
Disclosure of Invention
The embodiment of the invention provides a cross-domain electrocardiogram biological characteristic identity recognition method, which aims to overcome the problems in the prior art.
In order to achieve the purpose, the invention adopts the following technical scheme.
A cross-domain electrocardiogram biological characteristic identity recognition method comprises the following steps:
establishing an Electrocardiogram (ECG) data set, and selecting a data sample signal from the ECG data set;
extracting time domain characteristics, frequency domain characteristics and energy domain characteristics of the data sample signals;
and constructing an ECG biological characteristic identity recognition model by using the improved CNN convolutional neural network, and inputting the time domain characteristic, the frequency domain characteristic and the energy domain characteristic of the data sample signal into the ECG biological characteristic identity recognition model to obtain an ECG biological characteristic identity recognition result of the data sample signal.
Preferably, said creating an ECG data set comprises:
selecting electrocardiosignal data in a database PTBDB and an ECG-ID as an ECG data set, and carrying out noise filtering processing on power frequency interference, random noise and baseline drift of the electrocardiosignal data in the ECG data set through a Butterworth filter and an IIR filter.
Preferably, said selecting a data sample signal from the ECG data set comprises:
selecting a time starting point of a data sample signal in a given time period, and selecting an electrocardiosignal data sequence with a fixed time length backward as the data sample signal in the ECG data set by taking the time starting point as a starting point, wherein the expression of the data sample signal is as follows:
wherein,an ith sample X representing acquired s individual electrocardiographic signal data,j-th record R, P representing selected s-th individual electrocardiosignal dataiDenotes the start position of the ith data sample signal, EiIndicating the termination position of the ith data sample signal;
Ei＝Pi+t*f
wherein t represents the time length of single individual electrocardiosignal data, and f represents the sampling frequency.
Preferably, the extracting time domain features, frequency domain features and energy domain features of the data sample signal includes:
selecting the statistical characteristics of the mean value, the standard deviation, the kurtosis and the skewness of the data sample signal, and taking the statistical characteristics as the time domain characteristics of the data sample signal;
firstly, Fast Fourier Transform (FFT) is carried out on an original data sample signal, Discrete Cosine Transform (DCT) is carried out on an obtained frequency spectrum, peak information is extracted from a transformed result, and envelope information of the original data sample signal is obtained, wherein the envelope information is the frequency domain characteristic of the data sample signal;
performing energy operator operation on an original data sample signal, and then performing Fast Fourier Transform (FFT), and extracting energy domain characteristics of the data sample signal; teager energy operator ψ of a discrete time-domain signal s (n)DThe calculation of(s) is as follows:
ψD(s)＝s2(n)-s(n+1)s(n-1)
the time domain characteristics, the frequency domain characteristics and the energy domain characteristics of the data sample signals are combined, the characteristics of five channels extracted from the time domain, the frequency domain and the energy domain are respectively subjected to three-layer convolution, and the results after the convolution are directly cascaded to obtain the combined characteristics.
Preferably, the constructing an ECG biological characteristic identification model by using the improved CNN convolutional neural network, and inputting the time domain characteristic, the frequency domain characteristic and the energy domain characteristic of the data sample signal into the ECG biological characteristic identification model to obtain an ECG biological characteristic identification result of the data sample signal includes:
introducing a channel Attention module into the CNN, changing a ReLU activation function in the Attention module into Sigmoid, and constructing an ECG biological characteristic identity recognition model by using the improved CNN, wherein the structure of the ECG biological characteristic identity recognition model comprises a convolution layer, a full connection layer and an output layer, the convolution layer is used for increasing the channel Attention module, the convolution layer is used for transforming input characteristics, the full connection layer is used for defining the connection relation between neurons, and finally, a final result is generated on the output layer through Softmax;
and training the ECG biological characteristic identity recognition model, inputting the combined characteristics into the trained ECG biological characteristic identity recognition model to obtain an ECG biological characteristic identity recognition result of the data sample signal, and obtaining the personal identity of the acquirer corresponding to the data sample signal.
According to the technical scheme provided by the embodiment of the invention, the cross-domain ECG biological characteristic identification method provided by the embodiment of the invention can effectively improve the accuracy of ECG biological characteristic identification even if the interval between the training time and the application time is large enough.
Additional aspects and advantages 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
In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings needed to be used in the description of the embodiments are briefly introduced below, and it is obvious that the drawings in the following description are only some embodiments of the present invention, and it is obvious for those skilled in the art to obtain other drawings based on these drawings without creative efforts.
FIGS. 1(a) and 1(b) are schematic diagrams comparing a raw ECG signal and a filtered ECG signal according to an embodiment of the present invention;
FIG. 2 is a schematic diagram illustrating an implementation principle of a cross-domain electrocardiogram biometric identity recognition method according to an embodiment of the present invention;
FIG. 3 is a flowchart illustrating a cross-domain electrocardiogram biometric identity recognition method according to an embodiment of the present invention;
FIG. 4 is a schematic diagram of a MFCC dynamic feature extraction process provided in an embodiment of the present invention;
fig. 5 is a schematic diagram of a channel authorization module according to an embodiment of the present invention.
Detailed Description
Reference will now be made in detail to embodiments of the present invention, examples of which are illustrated in the accompanying drawings, wherein like reference numerals refer to the same or similar elements or elements having the same or similar function throughout. The embodiments described below with reference to the accompanying drawings are illustrative only for the purpose of explaining the present invention, and are not to be construed as limiting the present invention.
As used herein, the singular forms "a", "an", "the" and "the" are intended to include the plural forms as well, unless the context clearly indicates otherwise. It will be further understood that the terms "comprises" and/or "comprising," when used in this specification, specify the presence of stated features, integers, steps, operations, elements, and/or components, but do not preclude the presence or addition of one or more other features, integers, steps, operations, elements, components, and/or groups thereof. It will be understood that when an embodiment of the invention is referred to as being "connected" or "coupled" to another element, it can be directly connected or coupled to the other element or intervening elements may also be present. Further, "connected" or "coupled" as used herein may include wirelessly connected or coupled. As used herein, the term "and/or" includes any and all combinations of one or more of the associated listed items.
It will be understood by those skilled in the art that, unless otherwise defined, all terms (including 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 will be further understood that terms, such as those defined in commonly used dictionaries, should be interpreted as having a meaning that is consistent with their meaning in the context of the prior art and will not be interpreted in an idealized or overly formal sense unless expressly so defined herein.
For the convenience of understanding the embodiments of the present invention, the following description will be further explained by taking several specific embodiments as examples in conjunction with the drawings, and the embodiments are not to be construed as limiting the embodiments of the present invention.
Aiming at the condition that an obvious time interval exists between training time and application time, the invention provides a practical cross-domain electrocardiogram biological characteristic identity recognition method. According to the scheme, three dimensions of a time domain, a frequency domain and an energy domain are considered comprehensively, optimal sample acquisition parameter setting is determined, a series of characteristics which do not change dynamically along with time are extracted, a network model is optimized, and finally high identification accuracy is obtained.
The implementation principle schematic diagram of the cross-domain electrocardiogram biological characteristic identity recognition method provided by the embodiment of the invention is shown in fig. 2, the specific processing flow is shown in fig. 3, and the method comprises the following processing steps:
step S31, an ECG data set is created.
According to the embodiment of the invention, partial electrocardiosignal data in two public databases PTBDB and ECG-ID are selected as an ECG data set, more than half of the electrocardiosignal data in the databases PTBDB and ECG-ID comprise at least two times of recording, one time of recording can be used as a training set, and one time of recording can be used as a test set. The ECG data set includes an acquisition time field for the electrocardiographic signal data.
The raw ECG signals in the ECG data set are obtained from an electronic acquisition device, and therefore, three types of noise are generated: power frequency interference, random noise, and baseline drift, which would severely interfere with the classifier's judgment, therefore, require pre-processing of the data set. The embodiment of the invention designs a filter bank which comprises a Butterworth filter and an Infinite Impulse Response (IIR) filter and is used for carrying out noise filtering processing on 50HZ power frequency interference, random noise and baseline drift of an electronic element influenced by the temperature of acquisition equipment of electrocardiosignal data in an ECG data set. After noise filtering, a relatively clean ECG signal is obtained, making feature extraction and classification of the data more accurate. Fig. 1(a) and 1(b) show a raw ECG signal and a filtered ECG signal, respectively.
Step S32, a data sample signal is selected from the ECG data set.
The data set acquisition process is perfected to improve the accuracy of feature extraction and reduce the confusion degree of the classifier. Selecting a time starting point of a data sample signal in a given time period, and selecting an electrocardiosignal data sequence with a fixed time length backward in the ECG data set by taking the time starting point as a data sample signal, wherein the expression of the data sample signal is as follows:
wherein,an ith sample X representing acquired s individual electrocardiographic signal data,j-th record R, P representing selected s-th individual electrocardiosignal dataiDenotes the start position of the ith data sample signal, EiIndicating the termination of the ith data sample signal.
Ei＝Pi+t*f
Wherein t represents the time length of single individual electrocardiosignal data, and f represents the sampling frequency.
And step S33, extracting time domain characteristics, frequency domain characteristics and energy domain characteristics of the data sample signals.
The feature extraction of the data sample signal is based on a time domain, a frequency domain and an energy domain, is divided into three channels for extraction, and then is combined.
The time domain. And selecting statistical characteristics such as the mean value, the standard deviation, the kurtosis and the skewness of the data sample signal, and taking the statistical characteristics as the time domain characteristics of the data sample signal.
The frequency domain. And performing frequency domain feature extraction on the data sample signals according to Meier Frequency Cepstrum Coefficients (MFCC), Fast Fourier Transform (FFT) and Discrete Cosine Transform (DCT). The MFCC coefficient is a coefficient based on linear transformation of log energy spectrum of nonlinear mel scale (mel scale) of sound frequency, whose band division is equally divided on mel scale, more closely approximating human auditory system than the linearly spaced bands used in normal log cepstrum. Fig. 4 is a schematic diagram of an MFCC dynamic feature extraction process according to an embodiment of the present invention. The invention provides an improved cepstrum feature extraction method (namely, FFT + DCT + extraction peak sequence), which comprises the following processing procedures: similar to Meier Frequency Cepstrum Coefficients (MFCCs), Fast Fourier Transform (FFT) is performed on an original data sample signal, Discrete Cosine Transform (DCT) is performed on the obtained frequency spectrum, peak information is extracted for the transformed result, and envelope information of the original data sample signal can be obtained, where the envelope information is a frequency domain feature. The method more effectively strengthens the envelope characteristics of the ECG signal on the basis of Meier Frequency Cepstrum Coefficients (MFCC).
An energy domain. The Teager energy operator is a nonlinear operator and can concentrate energy near a peak so as to eliminate zero-mean noise in a time sequence, and the calculation mode of the discrete time domain signal s (n) Teager energy operator is as follows:
ψD(s)＝s2(n)-s(n+1)s(n-1)
and performing energy operator operation on the original data sample signal, and then performing Fast Fourier Transform (FFT) to extract the energy domain characteristics of the data sample signal. In the invention, the Teager energy operator is creatively introduced, so that the frequency spectrum of the FFT is closer to the frequency spectrum of the real ECG signal, and the extracted characteristics are more accurate.
Then, the time domain characteristics, the frequency domain characteristics and the energy domain characteristics of the data sample signals are combined, the characteristics of the five channels extracted from the time domain, the frequency domain and the energy domain are respectively subjected to three-layer convolution, and then the results after the convolution are directly cascaded to obtain combined characteristics. The combined features are used as input to a classifier.
Step S34, an ECG biological characteristic identification model is built by using the improved CNN (Convolutional Neural Networks), and the combined characteristics are input into the ECG biological characteristic identification model to obtain an ECG biological characteristic identification result of the data sample signal.
In order to solve the limitation of the CNN in learning the channel information, the invention improves the CNN, and introduces a channel Attention module into the CNN to improve the learning capability of the CNN before classification, wherein an Attention mechanism comprises a series of uneven weight parameter distribution coefficients. Fig. 5 is a schematic diagram of a channel Attention module according to an embodiment of the present invention. Wherein, the recognition accuracy is improved by one percent by changing the ReLU activation function into Sigmoid.
The channel attention module is used for helping the model endow different weights to the characteristics of different channels, and extracting more important information, so that the model can learn more key information, and more correct judgment can be made.
The activation function in the Attention module is changed from ReLU to Sigmoid, and the role of the Sigmoid function is to enhance the nonlinearity of the neural network. Experimental results show that in the Attention module, the effect of using Sigmoid is better than that of ReLU.
And (3) constructing an ECG biological characteristic identity recognition model by using the improved CNN, inputting the combined characteristics of the time domain characteristics, the frequency domain characteristics and the energy domain characteristics of the data sample signals into the ECG biological characteristic identity recognition model to obtain an ECG biological characteristic identity recognition result of the data sample signals, and thus obtaining the personal identity of the acquirer corresponding to the data sample signals.
The structure of the ECG biological characteristic identity recognition model comprises a convolution layer, a full connection layer and an output layer, wherein the convolution layer is added with a channel Attention module, the convolution layer transforms input characteristics, the full connection layer defines the connection relation between neurons, and finally a final result is generated in the output layer through Softmax. The model needs to be trained. The input combined features are data, the identity of the collector is a label, and the personal identity of the collector corresponding to the data sample signal is determined by continuously reducing the proportion of classification errors through supervised learning.
The characteristics which do not change obviously along with the dynamic change of time are extracted no matter the characteristics of time domain, frequency domain or energy domain are extracted, the individual can be expressed more accurately, the optimization of sample acquisition parameters and a classifier model is added, and the overall identification accuracy of the system can meet the basic practical requirement.
In summary, the cross-domain ECG biometric identity recognition method provided by the embodiment of the present invention can effectively improve the accuracy of ECG biometric identity recognition even if the interval between the training time and the application time is large enough.
The invention obtains enough effective samples by determining the optimal parameters of the non-reference random sampling method; extracting individual distinguishable features insensitive to time span between training and testing cycles by using depth features across time, frequency and energy domains; a channel assignment module is introduced in CNN and the activation function is modified to optimize the performance of the classifier. A series of experimental results show that even if the time interval between training and testing is large enough, the method disclosed by the invention can still realize higher identification accuracy (the PTBDB and the ECG-ID are respectively 56.93% and 85.94%), which is respectively improved by 41.5% and 20.7% compared with the conventional optimal method.
Those of ordinary skill in the art will understand that: the figures are merely schematic representations of one embodiment, and the blocks or flow diagrams in the figures are not necessarily required to practice the present invention.
From the above description of the embodiments, it is clear to those skilled in the art that the present invention can be implemented by software plus necessary general hardware platform. Based on such understanding, the technical solutions of the present invention may be embodied in the form of a software product, which may be stored in a storage medium, such as ROM/RAM, magnetic disk, optical disk, etc., and includes instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to execute the method according to the embodiments or some parts of the embodiments.
The embodiments in the present specification are described in a progressive manner, and the same and similar parts among the embodiments are referred to each other, and each embodiment focuses on the differences from the other embodiments. In particular, for apparatus or system embodiments, since they are substantially similar to method embodiments, they are described in relative terms, as long as they are described in partial descriptions of method embodiments. The above-described embodiments of the apparatus and system are merely illustrative, and the units described as separate parts may or may not be physically separate, and the parts displayed as units may or may not be physical units, may be located in one place, or may be distributed on a plurality of network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of the present embodiment. One of ordinary skill in the art can understand and implement it without inventive effort.
The above description is only for the preferred embodiment of the present invention, but the scope of the present invention is not limited thereto, and any changes or substitutions that can be easily conceived by those skilled in the art within the technical scope of the present invention are included in the scope of the present invention. Therefore, the protection scope of the present invention shall be subject to the protection scope of the claims.
Claims (5)
1. A cross-domain electrocardiogram biological characteristic identity recognition method is characterized by comprising the following steps:
establishing an Electrocardiogram (ECG) data set, and selecting a data sample signal from the ECG data set;
extracting time domain characteristics, frequency domain characteristics and energy domain characteristics of the data sample signals;
and constructing an ECG biological characteristic identity recognition model by using the improved CNN convolutional neural network, and inputting the time domain characteristic, the frequency domain characteristic and the energy domain characteristic of the data sample signal into the ECG biological characteristic identity recognition model to obtain an ECG biological characteristic identity recognition result of the data sample signal.
2. The method of claim 1, wherein said creating an ECG data set comprises:
selecting electrocardiosignal data in a database PTBDB and an ECG-ID as an ECG data set, and carrying out noise filtering processing on power frequency interference, random noise and baseline drift of the electrocardiosignal data in the ECG data set through a Butterworth filter and an IIR filter.
3. The method of claim 2, wherein said extracting a data sample signal from the ECG data set comprises:
selecting a time starting point of a data sample signal in a given time period, and selecting an electrocardiosignal data sequence with a fixed time length backward as the data sample signal in the ECG data set by taking the time starting point as a starting point, wherein the expression of the data sample signal is as follows:
wherein,an ith sample X representing acquired s individual electrocardiographic signal data,j-th record R, P representing selected s-th individual electrocardiosignal dataiDenotes the start position of the ith data sample signal, EiIndicating the termination position of the ith data sample signal;
Ei＝Pi+t*f
wherein t represents the time length of single individual electrocardiosignal data, and f represents the sampling frequency.
4. The method according to any one of claims 1 to 3, wherein the extracting the time domain feature, the frequency domain feature and the energy domain feature of the data sample signal comprises:
selecting the statistical characteristics of the mean value, the standard deviation, the kurtosis and the skewness of the data sample signal, and taking the statistical characteristics as the time domain characteristics of the data sample signal;
firstly, Fast Fourier Transform (FFT) is carried out on an original data sample signal, Discrete Cosine Transform (DCT) is carried out on an obtained frequency spectrum, peak information is extracted from a transformed result, and envelope information of the original data sample signal is obtained, wherein the envelope information is the frequency domain characteristic of the data sample signal;
performing energy operator operation on an original data sample signal, and then performing Fast Fourier Transform (FFT), and extracting energy domain characteristics of the data sample signal; teager energy operator ψ of a discrete time-domain signal s (n)DThe calculation of(s) is as follows:
ψD(s)＝s2(n)-s(n+1)s(n-1)
the time domain characteristics, the frequency domain characteristics and the energy domain characteristics of the data sample signals are combined, the characteristics of five channels extracted from the time domain, the frequency domain and the energy domain are respectively subjected to three-layer convolution, and the results after the convolution are directly cascaded to obtain the combined characteristics.
5. The method according to claim 4, wherein the constructing an ECG biometric identity recognition model by using the modified CNN convolutional neural network, and inputting the time domain features, the frequency domain features and the energy domain features of the data sample signal into the ECG biometric identity recognition model to obtain an ECG biometric identity recognition result of the data sample signal comprises:
introducing a channel Attention module into the CNN, changing a ReLU activation function in the Attention module into Sigmoid, and constructing an ECG biological characteristic identity recognition model by using the improved CNN, wherein the structure of the ECG biological characteristic identity recognition model comprises a convolution layer, a full connection layer and an output layer, the convolution layer is used for increasing the channel Attention module, the convolution layer is used for transforming input characteristics, the full connection layer is used for defining the connection relation between neurons, and finally, a final result is generated on the output layer through Softmax;
and training the ECG biological characteristic identity recognition model, inputting the combined characteristics into the trained ECG biological characteristic identity recognition model to obtain an ECG biological characteristic identity recognition result of the data sample signal, and obtaining the personal identity of the acquirer corresponding to the data sample signal.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910505119.7A CN110192864B (en) | 2019-06-12 | 2019-06-12 | Cross-domain electrocardiogram biological characteristic identity recognition method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910505119.7A CN110192864B (en) | 2019-06-12 | 2019-06-12 | Cross-domain electrocardiogram biological characteristic identity recognition method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN110192864A true CN110192864A (en) | 2019-09-03 |
| CN110192864B CN110192864B (en) | 2020-09-22 |
Family
ID=67754475
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910505119.7A Expired - Fee Related CN110192864B (en) | 2019-06-12 | 2019-06-12 | Cross-domain electrocardiogram biological characteristic identity recognition method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110192864B (en) |
Cited By (3)
| 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 |
| CN112401902A (en) * | 2020-12-04 | 2021-02-26 | 山东大学 | Electrocardio identity recognition method and system based on neural network time-frequency analysis combination |
| CN112614539A (en) * | 2020-12-31 | 2021-04-06 | 西安邮电大学 | Motor imagery detection method based on TEO-MIC algorithm |
Citations (14)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101669090A (en) * | 2007-04-26 | 2010-03-10 | 福特全球技术公司 | Emotive advisory system and method |
| CN101773394A (en) * | 2010-01-06 | 2010-07-14 | 中国航天员科研训练中心 | Identification method and identification system using identification method |
| CN104063645A (en) * | 2014-07-01 | 2014-09-24 | 清华大学深圳研究生院 | Identity recognition method based on ECG dynamic self-updating samples |
| WO2017072250A1 (en) * | 2015-10-27 | 2017-05-04 | CardioLogs Technologies | An automatic method to delineate or categorize an electrocardiogram |
| CN107137071A (en) * | 2017-04-26 | 2017-09-08 | 可瑞尔科技(扬州)有限公司 | It is a kind of to analyze the method that heart impact signal is used for calculating short-term heart beat value |
| WO2017187422A1 (en) * | 2016-04-29 | 2017-11-02 | Inesc Tec - Instituto De Engenharia De Sistemas E Computadores, Tecnologia E Ciência | Biometric method and device for identifying a person through an electrocardiogram (ecg) waveform |
| CN107766781A (en) * | 2016-08-19 | 2018-03-06 | 清华大学深圳研究生院 | A kind of method and its system of quick electrocardio identification |
| CN108158578A (en) * | 2017-12-14 | 2018-06-15 | 北京顺源开华科技有限公司 | Noise segments recognition methods, ECG signal processing method and processing device |
| WO2018162957A1 (en) * | 2017-03-10 | 2018-09-13 | Qatar University | Personalized ecg monitoring for early detection of cardiac abnormalities |
| CN108714026A (en) * | 2018-03-27 | 2018-10-30 | 杭州电子科技大学 | The fine granularity electrocardiosignal sorting technique merged based on depth convolutional neural networks and on-line decision |
| CN109157231A (en) * | 2018-10-24 | 2019-01-08 | 阿呆科技（北京）有限公司 | Portable multi-channel Depression trend assessment system based on emotional distress task |
| CN109171707A (en) * | 2018-10-24 | 2019-01-11 | 杭州电子科技大学 | A kind of intelligent cardiac figure classification method |
| CN109711383A (en) * | 2019-01-07 | 2019-05-03 | 重庆邮电大学 | Convolutional neural networks Mental imagery EEG signal identification method based on time-frequency domain |
| US20190172481A1 (en) * | 2016-03-31 | 2019-06-06 | OmniSpeech LLC | Pitch detection algorithm based on pwvt of teager energy operator |
-
2019
- 2019-06-12 CN CN201910505119.7A patent/CN110192864B/en not_active Expired - Fee Related
Patent Citations (14)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101669090A (en) * | 2007-04-26 | 2010-03-10 | 福特全球技术公司 | Emotive advisory system and method |
| CN101773394A (en) * | 2010-01-06 | 2010-07-14 | 中国航天员科研训练中心 | Identification method and identification system using identification method |
| CN104063645A (en) * | 2014-07-01 | 2014-09-24 | 清华大学深圳研究生院 | Identity recognition method based on ECG dynamic self-updating samples |
| WO2017072250A1 (en) * | 2015-10-27 | 2017-05-04 | CardioLogs Technologies | An automatic method to delineate or categorize an electrocardiogram |
| US20190172481A1 (en) * | 2016-03-31 | 2019-06-06 | OmniSpeech LLC | Pitch detection algorithm based on pwvt of teager energy operator |
| WO2017187422A1 (en) * | 2016-04-29 | 2017-11-02 | Inesc Tec - Instituto De Engenharia De Sistemas E Computadores, Tecnologia E Ciência | Biometric method and device for identifying a person through an electrocardiogram (ecg) waveform |
| CN107766781A (en) * | 2016-08-19 | 2018-03-06 | 清华大学深圳研究生院 | A kind of method and its system of quick electrocardio identification |
| WO2018162957A1 (en) * | 2017-03-10 | 2018-09-13 | Qatar University | Personalized ecg monitoring for early detection of cardiac abnormalities |
| CN107137071A (en) * | 2017-04-26 | 2017-09-08 | 可瑞尔科技(扬州)有限公司 | It is a kind of to analyze the method that heart impact signal is used for calculating short-term heart beat value |
| CN108158578A (en) * | 2017-12-14 | 2018-06-15 | 北京顺源开华科技有限公司 | Noise segments recognition methods, ECG signal processing method and processing device |
| CN108714026A (en) * | 2018-03-27 | 2018-10-30 | 杭州电子科技大学 | The fine granularity electrocardiosignal sorting technique merged based on depth convolutional neural networks and on-line decision |
| CN109157231A (en) * | 2018-10-24 | 2019-01-08 | 阿呆科技（北京）有限公司 | Portable multi-channel Depression trend assessment system based on emotional distress task |
| CN109171707A (en) * | 2018-10-24 | 2019-01-11 | 杭州电子科技大学 | A kind of intelligent cardiac figure classification method |
| CN109711383A (en) * | 2019-01-07 | 2019-05-03 | 重庆邮电大学 | Convolutional neural networks Mental imagery EEG signal identification method based on time-frequency domain |
Non-Patent Citations (3)
| Title |
|---|
| 王文峰: "《MATLAB计算机视觉与机器认知》", 31 August 2017, 北京航空航天大学出版社 * |
| 田福庆等: "《机器故障非平稳特征提取方法及其应用》", 30 June 2014, 国防工业出版社 * |
| 项延德: "基于卷积神经网络的心电信号检测和分类研究", 《中国博士学位论文全文数据库》 * |
Cited By (4)
| 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 |
| CN112401902A (en) * | 2020-12-04 | 2021-02-26 | 山东大学 | Electrocardio identity recognition method and system based on neural network time-frequency analysis combination |
| CN112614539A (en) * | 2020-12-31 | 2021-04-06 | 西安邮电大学 | Motor imagery detection method based on TEO-MIC algorithm |
| CN112614539B (en) * | 2020-12-31 | 2023-04-11 | 西安邮电大学 | Motor imagery detection method based on TEO-MIC algorithm |
Also Published As
| Publication number | Publication date |
|---|---|
| CN110192864B (en) | 2020-09-22 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108776788B (en) | Brain wave-based identification method | |
| CN110192864B (en) | Cross-domain electrocardiogram biological characteristic identity recognition method | |
| CN110353702A (en) | A kind of emotion identification method and system based on shallow-layer convolutional neural networks | |
| CN106951753B (en) | Electrocardiosignal authentication method and device | |
| CN106909784A (en) | Epileptic electroencephalogram (eeg) recognition methods based on two-dimentional time-frequency image depth convolutional neural networks | |
| CN110236533A (en) | Epileptic seizure prediction method based on the study of more deep neural network migration features | |
| CN104887263B (en) | A kind of identification algorithm and its system based on heart sound multi-dimension feature extraction | |
| CN113221673B (en) | Speaker authentication method and system based on multi-scale feature aggregation | |
| CN105615834A (en) | Sleep stage classification method and device based on sleep EEG (electroencephalogram) signals | |
| CN104173045A (en) | Epileptic seizure prewarning system | |
| CN110292377B (en) | Electroencephalogram signal analysis method based on instantaneous frequency and power spectrum entropy fusion characteristics | |
| Keshishzadeh et al. | Improved EEG based human authentication system on large dataset | |
| CN108567418A (en) | A kind of pulse signal inferior health detection method and detecting system based on PCANet | |
| Dan et al. | An identification system based on portable EEG acquisition equipment | |
| CN116153330B (en) | Intelligent telephone voice robot control method | |
| CN115969392A (en) | Cross-period brainprint recognition method based on tensor frequency space attention domain adaptive network | |
| Yang et al. | A simple deep learning method for neuronal spike sorting | |
| Leung et al. | Reconstruction of FRI signals using deep neural network approaches | |
| Ariff et al. | Study of adam and adamax optimizers on alexnet architecture for voice biometric authentication system | |
| CN114081505A (en) | Electroencephalogram signal identification method based on Pearson correlation coefficient and convolutional neural network | |
| CN113397562A (en) | Sleep spindle wave detection method based on deep learning | |
| CN117113015A (en) | Electroencephalogram signal identification method and device based on space-time deep learning | |
| CN116776284A (en) | Fault diagnosis method for electromechanical device, computer device, and storage medium | |
| CN114841216B (en) | Electroencephalogram signal classification method based on model uncertainty learning | |
| CN110443276A (en) | Time series classification method based on depth convolutional network Yu the map analysis of gray scale recurrence |
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 | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20200922 |
|
| CF01 | Termination of patent right due to non-payment of annual fee |