WO2020019171A1

WO2020019171A1 - Identity authentication method and apparatus, electronic device, and storage medium

Info

Publication number: WO2020019171A1
Application number: PCT/CN2018/096894
Authority: WO
Inventors: 李烨; 孙方敏; 毛晨飞
Original assignee: 中国科学院深圳先进技术研究院
Priority date: 2018-07-24
Filing date: 2018-07-24
Publication date: 2020-01-30

Abstract

The embodiments of the present invention relate to the technical field of identity authentication, and provide an identity authentication method and apparatus, an electronic device, and a storage medium, the method comprising: acquiring a gait acceleration signal collected by an acceleration sensor arranged on the body of a user; implementing gait cycle detection of the gait acceleration signal to obtain a plurality of gait cycles; segmenting the plurality of gait cycles into a plurality of gait segmentation segments, and performing pattern extraction on each gait segmentation segment to obtain a gait pattern corresponding to each gait segmentation segment; performing eigenvector extraction on each gait pattern to obtain gait features of each gait pattern; and, on the basis of the gait features of each gait pattern, constructing an identity identification model in order to determine the identity of the user. Compared to the prior art, the embodiments of the present invention acquire gait acceleration signals of different users and determine gait cycles, satisfying the individual differences of different users, and implementing accurate identification of user identity based on gait identification.

Description

Identity authentication method, device, electronic equipment and storage medium

Technical field

Embodiments of the present invention relate to the technical field of identity authentication, and in particular, to an identity authentication method, device, electronic device, and storage medium.

Background technique

Gait refers to a person's walking posture. Each person's gait is unique and unique. Therefore, gait is a behavioral biological characteristic that contains personal identity information. In recent years, with the development of micro-electro-mechanical system technology, low-cost miniaturized acceleration sensors have developed rapidly and are widely used in various mobile wearable devices, which has made the collection of human gait information and gait-based identity authentication widely used. the study.

At present, the research methods of gait recognition can be divided into three types: based on machine vision, based on pressure sensors, and based on wearable sensors. Among them, based on machine vision, a camera is used to capture a series of gait images of a user walking, and then an image matching algorithm is used to implement identity authentication. However, it is susceptible to light, occlusion, distance, etc .; pressure-based sensors use pressure sensors to capture the user's gait characteristics and are easily affected by the external environment; wearable sensors are used to extract people through acceleration sensors worn at different positions on the human body Gait signals and implement user authentication or identification. Compared with the machine vision-based method and the pressure sensor-based method, the wearable sensor-based method can more directly and faithfully reflect human gait characteristics, but in the prior art, the user's pace difference was not considered during the detection and was ignored. Individual users are different, so there is a problem of inaccurate detection.

Summary of the Invention

An object of the embodiments of the present invention is to provide an identity authentication method, device, electronic device, and storage medium, so as to realize accurate identification of a user identity based on gait identification.

In order to achieve the foregoing objective, the technical solutions adopted in the embodiments of the present invention are as follows:

In a first aspect, an embodiment of the present invention provides an identity authentication method. The method includes: acquiring a gait acceleration signal collected by an acceleration sensor provided on a user; and performing a gait period detection on the gait acceleration signal to obtain Multiple gait cycles; divide multiple gait cycles into multiple gait segments, and extract patterns for each gait segment to get the gait pattern corresponding to each gait segment; for each Feature vectors are extracted from the gait patterns to obtain the gait characteristics of each gait pattern; according to the gait characteristics of each gait pattern, an identity recognition model is constructed to confirm the user identity.

In a second aspect, an embodiment of the present invention further provides an identity authentication device, which includes an acquisition module, a detection module, a pattern extraction module, a feature vector extraction module, and a model construction module. The acquisition module is configured to acquire a gait acceleration signal collected by an acceleration sensor provided on the user; the detection module is configured to perform a gait cycle detection on the gait acceleration signal to obtain multiple gait cycles; and the pattern extraction module is configured to Divide multiple gait periods into multiple gait segmentation segments, and perform pattern extraction on each gait segmentation segment to obtain the gait pattern corresponding to each gait segmentation segment; the feature vector extraction module is configured to Feature vectors are extracted from the gait patterns to obtain the gait characteristics of each gait pattern; the model building module is configured to construct an identity recognition model based on the gait characteristics of each gait pattern to confirm the user identity.

According to a third aspect, an embodiment of the present invention further provides an electronic device that is in communication connection with a wearable mobile terminal of a user. The wearable mobile terminal is provided with an acceleration sensor. The electronic device further includes: Or multiple processors; a memory configured to store one or more programs, and when the one or more programs are executed by the one or more processors, cause the one or more processors to achieve the foregoing identity Authentication method.

According to a fourth aspect, an embodiment of the present invention further provides a computer-readable storage medium on which a computer program is stored. When the computer program is executed by a processor, the foregoing identity authentication method is implemented.

Relative to the prior art, an identity authentication method, device, electronic device, and storage medium provided by embodiments of the present invention. First, a gait acceleration signal collected by an acceleration sensor provided on a user is acquired, and the gait acceleration signal is stepped. Gait cycle detection yields multiple gait cycles; then, multiple gait cycles are divided into multiple gait segmentation segments, and pattern extraction is performed for each gait segmentation segment to obtain the step corresponding to each gait segmentation segment. Gait pattern; then use the MFCC algorithm to extract feature vectors for each gait pattern to obtain the gait characteristics of each gait pattern; finally, based on the gait characteristics of each gait pattern, build an identity recognition model to confirm the user Identity. Compared with the prior art, the embodiment of the present invention obtains gait acceleration signals of different users and determines the gait period, which can meet the individual differences of different users and achieve accurate identification of user identity based on gait recognition.

In order to make the above-mentioned objects, features, and advantages of the present invention more comprehensible, preferred embodiments are described below in detail with reference to the accompanying drawings, as follows.

BRIEF DESCRIPTION OF THE DRAWINGS

In order to explain the technical solution of the embodiments of the present invention more clearly, the drawings used in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of the present invention, and therefore are not It should be regarded as a limitation on the scope. For those of ordinary skill in the art, other related drawings can be obtained based on these drawings without paying creative work.

FIG. 1 shows a block diagram of interaction between an electronic device and an acceleration sensor according to an embodiment of the present invention.

FIG. 2 is a schematic block diagram of an electronic device according to an embodiment of the present invention.

FIG. 3 shows a flowchart of an identity authentication method according to an embodiment of the present invention.

FIG. 4 shows a schematic diagram of a gait acceleration signal collected by an acceleration sensor.

FIG. 5 is a schematic diagram of Z-axis acceleration data in the gait acceleration signal shown in FIG. 4.

FIG. 6 is a schematic diagram of an autocorrelation signal of the gait acceleration signal shown in FIG. 4.

7 is a schematic diagram of a gait segmentation segment of the gait acceleration signal shown in FIG. 4,

FIG. 8 shows a schematic diagram of a gait period detection experiment result.

Figure 9 shows a schematic diagram of gait segmentation data.

FIG. 10 is a schematic diagram showing the results of a gait pattern extraction experiment.

FIG. 11 is a schematic block diagram of an identity authentication apparatus according to an embodiment of the present invention.

Icons: 10-electronic equipment; 20-acceleration sensor; 101-memory; 102-storage controller; 103-processor; 104-internal interface; 105-communication unit; 200-identity authentication device; 201-acquisition module; 202- Preprocessing module; 203-detection module; 204-pattern extraction module; 205-feature vector extraction module; 206-model construction module.

detailed description

In the following, the technical solutions in the embodiments of the present invention will be clearly and completely described with reference to the drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, but not all of the embodiments. The components of embodiments of the invention, generally described and illustrated in the figures herein, can be arranged and designed in a variety of different configurations. Therefore, the following detailed description of the embodiments of the present invention provided in the drawings is not intended to limit the scope of the claimed invention, but merely represents selected embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative work fall into the protection scope of the present invention.

It should be noted that similar reference numerals and letters indicate similar items in the following drawings, so once an item is defined in one drawing, it need not be further defined and explained in subsequent drawings. Meanwhile, in the description of the present invention, the terms “first”, “second”, and the like are only used to distinguish descriptions, and cannot be understood to indicate or imply relative importance.

In recent years, the rapid development of wearable devices has been widely used in a variety of application fields such as network communications, telemedicine, out-of-hospital rehabilitation, mobile payment, and interactive games. Its increasingly powerful functions (for example, information communication, mobile payment, personal banking, Positioning, etc.) and a large amount of stored personal information (for example, communication information, physiological health parameters, pictures, voice, video, etc.) make security and privacy protection an important requirement for wearable devices. At present, the security and privacy protection of wearable devices mainly include password encryption methods and biometric authentication methods. Among them, the password encryption method has the problems of easy theft and easy cracking of passwords, etc., and cannot well ensure the safety of wearable devices. With the increase of wearable devices, passwords are easily forgotten by users, which brings inconvenience to the use of wearable devices. Biometric authentication methods include fingerprint recognition and face recognition, which have better confidentiality, but require specialized fingerprint or face scanning components, are complicated to operate, high in cost, and cannot be continuously authenticated. It can be seen that the existing methods for security and privacy protection of wearable devices still have defects.

The gait recognition-based identity authentication method mainly analyzes the gait acceleration signals collected by the accelerometer in the wearable device to perform user identity authentication, which has the characteristics of not requiring the user's active cooperation and more accurate identification. Because gait is a behavioral biological characteristic that contains human identity information, the identity authentication method based on gait recognition is more accurate.

Please refer to FIG. 1, which illustrates a schematic block diagram of interaction between an electronic device 10 and at least one acceleration sensor 20 according to an embodiment of the present invention. The electronic device 10 communicates with the acceleration sensor 20. The electronic device 10 obtains a gait acceleration signal collected by the acceleration sensor 20 and performs gait recognition on the gait acceleration signal to implement user identity authentication. The electronic device 10 may be, but is not limited to, a smart electronic device such as a desktop computer, a notebook computer, and the like. The operating system of the electronic device 10 may be, but is not limited to, an IOS (iPhone operating system) system, a Windows system, and the like. The acceleration sensor 20 can be worn on the user by being integrated in a wearable device. The wearable device can be a mobile phone, a bracelet, an anklet, etc. The user can wear the wearable device on the body according to his actual situation, such as wrist, arm, Chest, waist, thighs, etc.

Please refer to FIG. 2, which is a schematic block diagram of an electronic device 10 according to an embodiment of the present invention. The electronic device 10 includes a memory 101, a memory controller 102, a processor 103, an internal interface 104, and a communication unit 105. The components of the memory 101, the storage controller 102, the processor 103, the internal interface 104, and the communication unit 105 are directly or indirectly electrically connected to each other to implement data transmission or interaction. For example, these components can be electrically connected to each other through one or more communication buses or signal lines. The identity authentication device 200 includes at least one software function module that can be stored in the memory 101 in the form of software or firmware or solidified in the operating system of the electronic device 10. The processor 103 is configured to execute executable modules stored in the memory 101, such as software function modules or computer programs included in the identity authentication device 200.

The memory 101 may be, but is not limited to, Random Access Memory (RAM), Read-Only Memory (ROM), Programmable Read-Only Memory (PROM), and Erasable read-only memory (Erasable Programmable Read-Only Memory, EPROM), electrically erasable read-only memory (Electric Erasable Programmable Read-Only Memory, EEPROM), etc.

The processor 103 may be an integrated circuit chip and has a signal processing capability. The above-mentioned processor 103 may be a general-purpose processor, including a central processing unit (CPU), a network processor (NP), a voice processor, and a video processor; it may also be a digital signal processor, Application-specific integrated circuits, field programmable gate arrays or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components. Various methods, steps, and logical block diagrams disclosed in the embodiments of the present invention may be implemented or executed. The general-purpose processor may be a microprocessor or the processor 103 may be any conventional processor or the like.

The internal interface 104 is configured to couple various input / output devices to the processor 103 and the memory 101. In some embodiments, the internal interface 104, the processor 103, and the memory controller 102 may be implemented in a single chip. In other examples, they can be implemented by separate chips.

The communication unit 105 is configured to establish a connection with the acceleration sensor 20 provided on the user through a wireless network, so as to enable the electronic device 10 to send and receive data through the wireless network.

First embodiment

Please refer to FIG. 3, which illustrates a flowchart of an identity authentication method according to an embodiment of the present invention. The identity authentication method includes the following steps:

Step S101: Acquire a gait acceleration signal collected by an acceleration sensor provided on a user.

In the embodiment of the present invention, the acceleration sensor 20 may be worn on a user by being integrated in a wearable device, and the acceleration sensor 20 may collect acceleration data of three orthogonal axes X, Y, and Z axes at the same time, that is, That is, during the user's walking, the acceleration sensor 20 outputs X-axis acceleration data, Y-axis acceleration data, and Z-axis acceleration data. The electronic device 10 obtains the gait acceleration signal collected by the acceleration sensor 20 through the communication unit 105. The electronic device 10 can obtain the gait acceleration signal collected by the acceleration sensor 20 through the communication unit 105. The gait acceleration signal may include multiple sample points (for example, 1200), each sample point includes X-axis acceleration data, Y-axis acceleration data and Z-axis acceleration data.

In step S102, the gait acceleration signal is pre-processed to reduce noise in the gait acceleration signal.

In the embodiment of the present invention, because the sampling clock of the acceleration sensor 20 integrated in the wearable device is unstable, the time interval between two consecutive sample points is inconsistent. Therefore, a linear interpolation method may be used for the acquired steps. The state acceleration signal is processed to ensure that the time interval between the sample points is fixed.

In addition, in the process of collecting the gait acceleration signal by the acceleration sensor 20, the gait acceleration signal obtained by the electronic device 10 inevitably contains a large amount of noise due to the influence of the uneven position of the sensor and the ground inequality. Level wavelet decomposition and reconstructed noise removal method attenuates the noise components in the gait acceleration signal. Specifically, the db6 wavelet orthogonal basis function can be used to perform two-level wavelet decomposition of the gait acceleration signal containing noise, keeping large-scale low All wavelet coefficients at the resolution. For wavelet coefficients at high resolutions at all scales, a threshold can be set. All wavelet coefficients whose amplitude is lower than the threshold are set to 0. Wavelet coefficients higher than the threshold are completely retained or contracted. After processing, the wavelet coefficients obtained after processing are reconstructed by inverse wavelet transform to recover an effective gait acceleration signal, which reduces the signal noise caused by the acquisition environment.

Step S103: Gait cycle detection is performed on the gait acceleration signal to obtain multiple gait cycles.

In the embodiment of the present invention, a gait cycle may be a time interval between the first heel contacting the ground and the same heel contacting the ground again. Since each user's walking posture, speed, and step length are different, each user's The gait cycle is unique. Different users correspond to different gait cycles. Generally, the gait acceleration signal collected by the acceleration sensor 20 includes multiple gait cycles. Therefore, the pre-processed gait acceleration signal needs to be stepped. Figure 4 shows the gait acceleration signal collected by the acceleration sensor 20. Figure 4 shows that the Z-axis acceleration data in the gait acceleration signal has a strong periodicity, so the Z-axis acceleration data can be used to Perform gait cycle detection. Specific methods can include:

First, obtain a plurality of minimum points of the Z-axis acceleration data in the gait acceleration signal. Specifically, randomly select the same point, and compare the acceleration value of the sample point with the acceleration values of adjacent sample points. If the acceleration value of the sample point is smaller than the acceleration values of the sample points adjacent to it, determine that the sample point is a minimum point, and follow the same method to traverse all the sample points in the Z-axis acceleration data to obtain Z Multiple minimum points in the axis acceleration data. For example, FIG. 5 is a schematic diagram of the Z-axis acceleration data of the first 600 sample points in the gait acceleration signal shown in FIG. 4. The sample points marked with “*” in FIG. 5 Both are minimum points.

Secondly, according to the acceleration value of each minima point, noise points in multiple minima points are filtered out. Due to the influence of factors such as weight during the user's walking, there may be multiple minima points obtained. Some noise points need to be filtered out. Specifically, the calibration differences and averages of the minimum points can be calculated based on the acceleration values of multiple minimum points. Then, a threshold is determined to filter out based on the calibration differences and averages of the minimum points. For the noise points of multiple minima, the threshold can be calculated according to the formula Threshold = mean + 0.5 * std, where Threshold represents the threshold, std represents the standard deviation, and mean represents the mean; finally, multiple minima are found according to the threshold The noise points among the value points are determined as the minimum points whose acceleration value is greater than the threshold value, and all the found noise points are filtered out.

Next, calculate the auto-correlation coefficient of the Z-axis acceleration data and determine the estimated step size based on the auto-correlation coefficient. Specifically, you can first calculate the auto-correlation coefficient of each sample point in the Z-axis acceleration data. The autocorrelation coefficient is normalized to obtain the autocorrelation signal; then the autocorrelation signal is subjected to smooth filtering processing to filter out noise in the autocorrelation signal, and the autocorrelation coefficient of the Z-axis acceleration data shown in FIG. 5 is calculated. The correlation signal is shown in Figure 6; finally, the interval between the first minimum point and the third minimum point in the autocorrelation signal obtained in the previous step is determined as the estimated step length L of the gait cycle.

Finally, multiple gait cycles are extracted based on the multiple minima points and the estimated step size after filtering out the noise points. Because the minima points obtained in the second step are not all the starting points of the gait cycle or The end point, so the minimum point obtained in the second step needs to be further filtered, and then the minimum point and the estimated step length L to find the start point and the end point of each gait cycle. The following takes the detection of the first gait cycle as an example:

First, calculate the first interval d1 between the first minimum point and the second minimum point, that is, count the number of sample points between the first minimum point and the second minimum point. The relationship between the first interval d1 and the estimated step size is used to find the start and end points of the first gait cycle. Specifically, if

That is, the first interval d1 is smaller than the estimated step size.

Then it is judged that neither the first minimum point nor the second minimum point is the starting point of the gait period. At this time, the acceleration values of the first minimum point and the second minimum point are compared and eliminated The minimum point with the larger acceleration value; if

That is, the first interval d1 is larger than the estimated step size.

Less than or equal to the estimated step size

Then determine that the first minimum point and the second minimum point are not the end points of the gait cycle;

Second, based on the first minimum point, calculate the second interval d2 between the third minimum point and the second minimum point, and use the second interval d2 and the estimated step size. To find the starting point and ending point of the first gait cycle. Specifically, if

Then the minimum value point with the larger acceleration value among the third minimum value point and the second minimum value point is eliminated; if

Then the second minimum point is eliminated, and the starting point and the ending point of the first gait cycle are detected by the above methods as the first minimum point and the third minimum point, respectively;

Third, if

That is, the first interval d1 is larger than the estimated step size.

Since the interval between adjacent minima points will not exceed a gait period, when the first interval d1 is larger than the estimated step size,

The starting point and ending point of the first gait cycle are the first minimum point and the second minimum point, respectively.

After the first gait cycle is detected, the end point of the previous gait cycle is the start point of the next gait cycle. Iterate through each minimum point in the above method to find the end point of the next gait cycle. Until all gait cycles are detected.

In step S104, a plurality of gait periods are divided into a plurality of gait segmentation segments, and pattern extraction is performed on each gait segmentation segment to obtain a gait pattern corresponding to each gait segmentation segment.

In the embodiment of the present invention, gait period detection is performed on the gait acceleration signal. After obtaining multiple gait periods, 4 gait periods are used as one gait segment, and between two adjacent gait segments. Set a 50% overlap to divide multiple gait cycles into multiple gait segmentation segments. The segmentation results are shown in Figure 7.

In the embodiment of the present invention, after obtaining multiple gait segmentation segments, pattern extraction is performed on each gait segmentation segment to obtain a gait pattern corresponding to each gait segmentation segment. Specifically, first, for each The gait acceleration signals corresponding to the gait segmentation are subjected to data fusion to obtain the first fusion data and the second fusion data corresponding to each gait segmentation segment, where the first fusion data is the X-axis acceleration data and the Y-axis acceleration data. Fusion. The first fusion data is the fusion of X-axis acceleration data, Y-axis acceleration data, and Z-axis acceleration data. That is, using the X, Y, and Z axes in each gait segment, use the formula

and

Then establish two axes M _XY axis and M _XYZ axis, the first fusion data is M _XY axis acceleration data, the second fusion data is M _XYZ axis acceleration data; then, the first fusion data corresponding to each gait segment , The second fusion data and the Z-axis acceleration data are combined to obtain a gait pattern corresponding to each gait segment, that is, the Z, M _XY, and M _XYZ axes in each gait segment are selected The data is the gait pattern corresponding to each gait segment, which is

In step S105, feature vector extraction is performed for each gait pattern to obtain a gait feature of each gait pattern.

In the embodiment of the present invention, each gait mode includes first fusion data, second fusion data, and Z-axis acceleration data, that is,

The specific method for extracting feature vectors for each gait pattern can be: First, use the MFCC algorithm to calculate the first fusion data, the second fusion data, and the Z-axis acceleration data in each gait pattern, respectively, to obtain the corresponding The first eigenvector, the second eigenvector, and the third eigenvector of the. That is, the MFCC algorithm is used to extract the Z-axis acceleration data, M _XY- axis acceleration data, and M _XYZ- axis acceleration data in each gait pattern P. MFCC coefficient, the first eigenvector is the MFCC coefficient of the Z-axis acceleration data, the second eigenvector is the MFCC coefficient of the M _XY- axis acceleration data, the third eigenvector is the MFCC coefficient of the M _XYZ- axis acceleration data, and the first eigenvector is , The second feature vector, and the third feature vector are all 18-dimensional MFCC vectors; then, the first feature vector, the second feature vector, and the third feature vector are fused to obtain the gait features of each gait pattern. That is, the Z axis, M _XY and M _XYZ axes corresponding to the three axes 18 dimensional MFCC vector integration together to form a 54-dimensional feature vectors, each step gait pattern Feature is a 54-dimensional feature vector.

In step S106, an identity recognition model is constructed according to the gait characteristics of each gait pattern to confirm the identity of the user.

In the embodiment of the present invention, in order to prevent the calculation amount from being too large, the gait features obtained above are subjected to PCA dimensionality reduction processing. At the same time, in order to improve the convergence accuracy of the identity recognition model, the gait characteristics after dimensionality reduction processing are normalized. Processing, and then use common classification algorithms in machine learning (for example, support vector machines, neural networks, K-valued nearest neighbors, etc.) to classify gait features to confirm user identity. Specifically, first, according to the gait characteristics of each gait pattern, a training data set and a test data set are constructed. Since each gait acceleration signal collected by the acceleration sensor 20 has a corresponding recording label when stored, the The label is configured to represent the data recorded several times, so all gait features with even number of records can be used as training data set, and all gait features with odd number of records can be used as test data set; then, the training The data set is input to a machine learning classifier to train an identity recognition model. The machine learning classifier can be a support vector machine classifier, a neural network classifier, a k-nearest neighbor classifier, etc. Finally, the test data set is input into the trained identity recognition model. , Output the category of each gait feature in the test data set. When performing user identity authentication, gait sample training is performed on the user in advance, and the extracted gait features are stored. During user identity authentication, machine learning algorithms are used to identify and authenticate the gait. If there is a classifier output, If it is +1, it will pass the authentication; if all the output of the classifier is -1, it will not pass the authentication.

In order to better illustrate that the identity authentication method provided by the embodiment of the present invention has higher recognition accuracy than that of the prior art, an experiment is used for explanation below, as follows:

Using the Google Android HTC Nexus One mobile phone's built-in acceleration sensor 20 to collect gait acceleration data, a total of 38 subjects data, including 28 males and 10 females, with an average age of 23 to 28 years. During the data collection process, the mobile phone was placed in the subject's pants pocket in a constant direction and set to the SENSOR_DELAY_FASTEST mode on the Android SDK, with a sampling rate of approximately 27 Hz. Each volunteer collected data more than 15 times, and each time the data was recorded, the number of times the data was recorded was recorded and labeled.

First, perform a gait cycle detection experiment. The experimental results are shown in Figure 8. The marked points in Figure 8 are the starting or ending points of the gait cycle. You can see that all the starting and ending points are extremely small. Value points, and the interval between adjacent minimum value points is uniform, and it is also within the neighborhood of the estimated step size, and has a strong periodicity;

Then, the gait mode is extracted. Because the acceleration data of the X and Y axes are unstable, two axes M _XY and M _{XYZ are} established through the X, Y, and Z axes in each gait segment. There are five axes of data in the segment. Figure 9 shows the acceleration data of the five axes. In the embodiment of the present invention, acceleration data of the Z axis, M _XY axis, and M _XYZ axis are used as a gait mode, and FIG. 10 is acceleration data in a gait mode.

Finally, gait feature recognition was performed. Data from 30 subjects in the data set were selected, including 22 males and 8 females, with a total of 559 recorded data, including 287 even-numbered data and 272 odd-numbered data. There are 5026 gait features, of which there are 2596 even-numbered recorded data, 2430 odd-numbered recorded data, with even-numbered recorded data as the training data set, and odd-numbered recorded data as the test data set. The training data set is imported into each In different classifiers (for example, support vector machines, neural networks, K-value nearest neighbors), the identity recognition model is obtained, and then the test data set is imported into the identity recognition model for predictive recognition, so that the recognition accuracy and classification can be obtained. The result table is shown in Table 1.

Table 1 Classification result table

分类算法Classification algorithm	准确率Accuracy
支持向量机Support Vector Machines	90.33％90.33%
神经网络Neural Networks	86.42％86.42%
K值最邻近K value is nearest	77.65％77.65%

As can be seen from Table 1, the SVM has the highest accuracy.

Compared with the prior art, the embodiments of the present invention have the following advantages:

First, find the autocorrelation coefficient of the gait acceleration signal and find the minimum point. Estimate the step size based on the first and third minimum points of the autocorrelation coefficient, and eliminate the gait based on the estimated step size. The minimum point of the cycle to find the exact starting and ending points of the gait cycle;

Secondly, due to the unstable acceleration data of the X and Y axes, the two axes M _XY and M _{XYZ are} established through the X, Y, and Z axes in each gait segment to eliminate this effect and extract The acceleration data of the Z axis, the M _XY axis, and the M _XYZ axis are used as the gait patterns, thereby obtaining relatively stable gait pattern data.

Finally, the MFCC coefficient of the acceleration data of the Z-axis, M _XY axis, and M _XYZ axis is calculated using the MFCC algorithm as the gait feature, which can more fully and truly reflect the gait feature, and is extracted in the time or frequency domain only than the traditional one. Gait characteristics are more accurate.

Second embodiment

Please refer to FIG. 11, which is a schematic block diagram of an identity authentication apparatus 200 according to an embodiment of the present invention. The identity authentication device 200 includes an acquisition module 201, a pre-processing module 202, a detection module 203, a pattern extraction module 204, a feature vector extraction module 205, and a model construction module 206.

The obtaining module 201 is configured to obtain a gait acceleration signal collected by an acceleration sensor provided on a user.

The pre-processing module 202 is configured to pre-process the gait acceleration signal to reduce noise in the gait acceleration signal.

The detection module 203 is configured to perform a gait period detection on the gait acceleration signal to obtain multiple gait periods.

In the embodiment of the present invention, the gait acceleration signal includes Z-axis acceleration data, and the detection module 203 is specifically configured to obtain multiple minimum points in the Z-axis acceleration data; according to the acceleration value of each minimum point, filtering Remove the noise points from multiple minima points; calculate the autocorrelation coefficient of the Z-axis acceleration data, and determine the estimated step size based on the autocorrelation coefficient; based on the multiple minima points and estimates after filtering out the noise points Step size, extract multiple gait cycles.

The pattern extraction module 204 is configured to divide multiple gait periods into multiple gait segmentation segments, and perform pattern extraction on each gait segmentation segment to obtain a gait pattern corresponding to each gait segmentation segment.

In the embodiment of the present invention, the pattern extraction module 204 is specifically configured to perform data fusion on a gait acceleration signal corresponding to each gait segment, and obtain first fusion data and second fusion data corresponding to each gait segment. ; Combining the first fusion data, the second fusion data, and the Z-axis acceleration data corresponding to each gait segment, to obtain a gait pattern corresponding to each gait segment.

The feature vector extraction module 205 is configured to perform feature vector extraction on each gait pattern to obtain a gait feature of each gait pattern.

In the embodiment of the present invention, each gait pattern includes first fusion data, second fusion data, and Z-axis acceleration data, and the feature vector extraction module 205 is specifically configured to use the MFCC algorithm for the first step in each gait pattern. A fusion data, a second fusion data, and a Z-axis acceleration data are respectively calculated to obtain corresponding first feature feature vector, second feature vector and third feature vector; the first feature feature vector, the second feature vector and The third feature vector is fused to obtain the gait features of each gait pattern.

The model construction module 206 is configured to construct an identity recognition model according to the gait characteristics of each gait pattern to confirm the identity of the user.

In the embodiment of the present invention, the pattern extraction module 204 is specifically configured to construct a training data set and a test data set according to the gait characteristics of each gait pattern; input the training data set into a machine learning classifier, and train out The identity recognition model; inputting the test data set into the trained identity recognition model, and outputting the category of each gait feature in the test data set to confirm the user identity.

An embodiment of the present invention also discloses a computer-readable storage medium on which a computer program is stored. When the computer program is executed by the processor 103, the identity authentication method disclosed in the foregoing embodiment of the present invention is implemented.

In summary, an identity authentication method, device, electronic device, and storage medium provided by embodiments of the present invention include: obtaining a gait acceleration signal collected by an acceleration sensor provided on a user; and gait acceleration signal Perform gait cycle detection to obtain multiple gait cycles; divide multiple gait cycles into multiple gait segmentation segments, and perform pattern extraction on each gait segmentation segment to obtain the corresponding value for each gait segmentation segment. Gait pattern; feature vector extraction is performed for each gait pattern to obtain the gait characteristics of each gait pattern; according to the gait characteristics of each gait pattern, an identity recognition model is constructed to confirm the user identity. Compared with the prior art, the embodiment of the present invention obtains gait acceleration signals of different users and determines the gait period, which can meet the individual differences of different users and achieve accurate identification of user identity based on gait recognition.

In the several embodiments provided in this application, it should be understood that the disclosed apparatus and method may also be implemented in other ways. The device embodiments described above are merely schematic. For example, the flowchart and block diagrams in the accompanying drawings show the architecture, functions, and functions of devices, methods, and computer program products according to various embodiments of the present invention. operating. In this regard, each block in the flowchart or block diagram may represent a module, program segment, or portion of code, which contains one or more components configured to implement a specified logical function. Executable instructions. It should also be noted that in some alternative implementations, the functions marked in the blocks may also occur in a different order than those marked in the drawings. For example, two consecutive blocks may actually be executed substantially in parallel, and they may sometimes be executed in the reverse order, depending on the functions involved. It should also be noted that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, can be implemented in a dedicated hardware-based system that performs the specified function or action. , Or it can be implemented with a combination of dedicated hardware and computer instructions.

In addition, the functional modules in the various embodiments of the present invention may be integrated together to form an independent part, or each of the modules may exist alone, or two or more modules may be integrated to form an independent part.

If the functions are implemented in the form of software function modules and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention is essentially a part that contributes to the existing technology or a part of the technical solution can be embodied in the form of a software product. The computer software product is stored in a storage medium, including Several instructions are used to cause a computer device (which may be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in various embodiments of the present invention. The foregoing storage media include: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), magnetic disks or optical disks and other media that can store program codes . It should be noted that in this article, relational terms such as first and second are used only to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply that these entities or operations There is any such actual relationship or order among them. Moreover, the terms "including", "comprising", or any other variation thereof are intended to encompass non-exclusive inclusion, such that a process, method, article, or device that includes a series of elements includes not only those elements but also those that are not explicitly listed Or other elements inherent to such a process, method, article, or device. Without more restrictions, the elements defined by the sentence "including a ..." do not exclude the existence of other identical elements in the process, method, article, or equipment including the elements.

The above descriptions are merely preferred embodiments of the present invention and are not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and changes. Any modification, equivalent replacement, and improvement made within the spirit and principle of the present invention shall be included in the protection scope of the present invention. It should be noted that similar reference numerals and letters indicate similar items in the following drawings, so once an item is defined in one drawing, it need not be further defined and explained in subsequent drawings.

Claims

An identity authentication method, comprising:

Obtain a gait acceleration signal collected by an acceleration sensor set on the user;

Performing a gait period detection on the gait acceleration signal to obtain multiple gait periods;

Divide multiple gait periods into multiple gait segmentation segments, and perform pattern extraction on each gait segmentation segment to obtain the gait pattern corresponding to each gait segmentation segment;

Feature vector extraction is performed for each gait pattern to obtain the gait feature of each gait pattern;

According to the gait characteristics of each gait pattern, an identity recognition model is constructed to confirm the identity of the user.
The method according to claim 1, wherein the gait acceleration signal comprises Z-axis acceleration data, and the step of performing a gait cycle detection on the gait acceleration signal to obtain multiple gait cycles comprises:

Acquiring a plurality of minimum points in the Z-axis acceleration data;

Filtering out noise points among the multiple minima points according to the acceleration value of each minima point;

Calculating an auto-correlation coefficient of the Z-axis acceleration data, and determining an estimated step size based on the auto-correlation coefficient;

A plurality of gait periods are extracted according to a plurality of minimum point points after filtering out noise points and the estimated step length.
The method according to claim 2, wherein the step of performing pattern extraction on each gait segmentation segment to obtain a gait pattern corresponding to each gait segmentation segment comprises:

Perform data fusion on the gait acceleration signal corresponding to each gait segment, to obtain first fusion data and second fusion data corresponding to each gait segment;

The first fusion data, the second fusion data, and the Z-axis acceleration data corresponding to each gait segment are combined to obtain a gait pattern corresponding to each gait segment.
The method according to claim 3, wherein each gait pattern includes first fusion data, second fusion data, and Z-axis acceleration data, and the feature extraction is performed on each gait pattern to obtain each The steps of the gait feature of each gait pattern include:

Use the MFCC algorithm to separately calculate the first fusion data, the second fusion data, and the Z-axis acceleration data in each gait mode to obtain corresponding first feature vector, second feature vector, and third feature vector;

The first feature feature vector, the second feature vector, and the third feature vector are fused to obtain a gait feature of each gait pattern.
The method according to claim 1, wherein the step of constructing an identity recognition model to confirm the identity of the user according to the gait characteristics of each gait pattern comprises:

According to the gait characteristics of each gait pattern, a training data set and a test data set are constructed;

Input the training data set into a machine learning classifier to train the identity recognition model;

The test data set is input into the trained identity recognition model, and the category of each gait feature in the test data set is output to confirm the identity of the user.
The method according to claim 1, wherein after the step of acquiring a gait acceleration signal collected by an acceleration sensor provided on a user, the method further comprises:

The gait acceleration signal is pre-processed to reduce noise in the gait acceleration signal.
An identity authentication device, characterized in that the device includes:

An acquisition module configured to acquire a gait acceleration signal collected by an acceleration sensor set on the user;

A detection module configured to perform a gait cycle detection on the gait acceleration signal to obtain multiple gait cycles;

The pattern extraction module is configured to divide multiple gait periods into multiple gait segmentation segments, and perform pattern extraction on each gait segmentation segment to obtain a gait pattern corresponding to each gait segmentation segment;

A feature vector extraction module configured to perform feature vector extraction on each gait pattern to obtain gait features of each gait pattern;

The model building module is configured to construct an identity recognition model according to the gait characteristics of each gait pattern to confirm the identity of the user.
The apparatus according to claim 7, further comprising:

The pre-processing module is configured to pre-process the gait acceleration signal to reduce noise in the gait acceleration signal.
An electronic device is characterized in that the electronic device is communicatively connected with a wearable mobile terminal of a user, an acceleration sensor is provided in the wearable mobile terminal, and the electronic device further includes:

One or more processors;

A memory configured to store one or more programs, when the one or more programs are executed by the one or more processors, causing the one or more processors to implement any one of claims 1-6 Item.
A computer-readable storage medium having stored thereon a computer program, characterized in that when the computer program is executed by a processor, the method according to any one of claims 1-6 is implemented.