Identity authentication method, device and equipment based on touch vibration signal, medium and product

Abstract

The embodiment of the invention provides an identity authentication method and device based on a touch vibration signal, equipment, a medium and a product, and relates to the technical field of artificial intelligence. The method comprises the steps that when it is detected that a user presses the terminal equipment, a touch vibration signal generated when the user presses the terminal equipment is collected; preprocessing the touch vibration signal to obtain a preprocessed touch vibration signal; performing feature extraction on the preprocessed touch vibration signal to obtain a global feature and a local feature corresponding to the preprocessed touch vibration signal; performing feature fusion on the global features and the local features to obtain fused features; and processing the fusion feature through a pre-constructed identity authentication model to obtain identity information corresponding to the fusion feature. According to the invention, identity authentication can be realized quickly and accurately when complex environment changes are handled.

Description

Technical Field

[0001] Embodiments of the present invention relate to the field of artificial intelligence technology. More specifically, embodiments of the present invention relate to an identity authentication method, device, equipment, medium and product based on a touch vibration signal. Background Art

[0002] This section is intended to provide a background or context to embodiments of the invention that are recited in the claims. No description herein is admitted to be prior art by inclusion in this section.

[0003] With the widespread use of mobile smart devices, smartphones and smart wristband devices have become necessities in daily life. Especially in the field of identity authentication, biometric-based identity authentication technology has been widely used on smartphones. Common biometric recognition technologies include fingerprint recognition and facial recognition, which greatly improve the security of smartphones due to their high uniqueness and difficulty in duplication.

[0004] Although biometric recognition technologies are outstanding in terms of convenience and security, they also have certain limitations. For example, fingerprint recognition is easily affected by the degree of moisture or dryness of the finger, and facial recognition may fail in low light or when the user wears a mask. It can be seen that these methods cannot quickly and accurately achieve identity authentication when dealing with complex environmental changes. Summary of the invention

[0005] In this context, embodiments of the present invention are intended to provide a method, device, apparatus, medium and product for identity authentication based on a tactile vibration signal.

[0006] In a first aspect of an embodiment of the present invention, a method for identity authentication based on a touch vibration signal is provided, which is applied to a terminal device, and the method includes:

[0007] When it is detected that the user presses the terminal device, collecting a touch vibration signal generated by the user pressing the terminal device;

[0008] Preprocessing the touch-vibration signal to obtain a preprocessed touch-vibration signal;

[0009] Performing feature extraction on the preprocessed touch-vibration signal to obtain global features and local features corresponding to the preprocessed touch-vibration signal;

[0010] Performing feature fusion on the global feature and the local feature to obtain a fused feature;

[0011] The fused features are processed through a pre-built identity authentication model to obtain identity information corresponding to the fused features.

[0012] In an example of this implementation manner, the preprocessing the touch vibration signal to obtain the preprocessed touch vibration signal includes:

[0013] Performing gravity correction on the touch vibration signal based on a preset filter factor to obtain a first touch vibration signal;

[0014] performing standardization processing on the first touch-vibration signal to obtain a second touch-vibration signal;

[0015] Based on a preset low cutoff frequency and a preset high cutoff frequency, the second touch-vibration signal is filtered to obtain a preprocessed touch-vibration signal.

[0016] In an embodiment of the present implementation manner, the extracting features of the pre-processed tactile vibration signal to obtain global features and local features corresponding to the pre-processed tactile vibration signal includes:

[0017] Performing global extraction on the preprocessed touch-vibration signal to obtain a global feature corresponding to the preprocessed touch-vibration signal;

[0018] Encoding the preprocessed touch-vibration signal to obtain key features of the preprocessed touch-vibration signal;

[0019] The key feature is decoded to obtain a local feature of the pre-processed touch-vibration signal.

[0020] In an embodiment of the present implementation manner, the globally extracting the preprocessed tactile vibration signal to obtain the global feature corresponding to the preprocessed tactile vibration signal includes:

[0021] Based on a preset frame duration and frame shift, dividing the pre-processed touch-vibration signal into a plurality of short-time frame signals;

[0022] Performing Fourier transform on each of the short-time frame signals to obtain a frequency spectrum of the pre-processed touch-vibration signal;

[0023] Filtering the frequency spectrum through a Mel filter bank to obtain a Mel frequency energy distribution;

[0024] Calculating the Mel frequency energy distribution to obtain a logarithmic Mel frequency spectrum;

[0025] Performing discrete cosine transform on the logarithmic Mel spectrum to obtain Mel frequency coefficients;

[0026] Performing differential calculation on the Mel frequency coefficient to obtain a first-order difference and a second-order difference of the Mel frequency coefficient;

[0027] The Mel frequency coefficient, the first-order difference, and the second-order difference are determined as global features corresponding to the pre-processed tactile vibration signal.

[0028] In an example of this implementation manner, the step of fusing the global feature and the local feature to obtain the fused feature includes:

[0029] The global features are calculated based on a multi-head attention mechanism to obtain global information;

[0030] Processing the local features based on an inverse neural network to obtain local fusion features;

[0031] The global information and the local fusion feature are fused to obtain a fusion feature.

[0032] In one embodiment of this implementation, the global features are calculated based on the multi-head attention mechanism to obtain global information, including:

[0033] Performing layer normalization on the global features to obtain normalized global features;

[0034] Calculating the normalized global features based on a multi-head attention mechanism to obtain initial global information;

[0035] Performing a residual connection on the global feature and the initial global information to obtain residual connection information;

[0036] The residual connection information is nonlinearly mapped through a multi-layer perceptron to obtain global information.

[0037] In a second aspect of the embodiments of the present invention, a model processing device is provided, which is applied to a terminal device, and the device includes:

[0038] a collecting unit, configured to collect a touch vibration signal generated by the user pressing the terminal device when it is detected that the user presses the terminal device;

[0039] A preprocessing unit, used for preprocessing the touch-vibration signal to obtain a preprocessed touch-vibration signal;

[0040] an extraction unit, configured to perform feature extraction on the preprocessed tactile-vibration signal to obtain global features and local features corresponding to the preprocessed tactile-vibration signal;

[0041] A fusion unit, used for fusing the global feature and the local feature to obtain a fused feature;

[0042] The processing unit is used to process the fused feature through a pre-built identity authentication model to obtain identity information corresponding to the fused feature.

[0043] In a third aspect of an embodiment of the present invention, a computing device is provided, comprising: at least one processor, a memory, and an input-output unit; wherein the memory is used to store a computer program, and the processor is used to call the computer program stored in the memory to execute any one of the methods described in the first aspect.

[0044] In a fourth aspect of the embodiments of the present invention, a computer-readable storage medium is provided, which includes instructions, and when the instructions are executed on a computer, the computer executes the method described in any one of the first aspects.

[0045] In a fifth aspect of the embodiments of the present invention, a computer program product is provided, comprising a computer program, which implements any one of the methods in the first aspect when executed by a processor.

[0046] According to the identity authentication method, device, equipment, medium and product based on the touch vibration signal of the embodiment of the present invention, the touch vibration signal generated when the user presses the terminal device can be processed to obtain the global and local features of the touch vibration signal, and the global and local features can be fused to obtain the fused features, and the corresponding identity information can be determined based on the fused features. It can be seen that the above method reduces the impact of the environment on identity authentication, that is, when dealing with complex environmental changes, identity authentication can be achieved quickly and accurately. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] The above and other objects, features and advantages of the exemplary embodiments of the present invention will become readily understood by reading the detailed description below with reference to the accompanying drawings. In the accompanying drawings, several embodiments of the present invention are shown in an exemplary and non-limiting manner, in which:

[0048] Figure 1 A schematic diagram of a flow chart of an identity authentication method based on a vibration signal provided by an embodiment of the present invention;

[0049] Figure 2 The distribution of the false acceptance rate (FAR) and the false rejection rate (FRR) of different types of models provided in one embodiment of the present invention;

[0050] Figure 3 A schematic diagram showing a comparison of classification accuracy of different types of models provided in one embodiment of the present invention;

[0051] Figure 4 A schematic diagram of the changes in FAR and FRR of a model using a Mel Frequency Cepstral Coefficient (MFCC) and a MFCC combined with an autoencoder (AE) feature extraction method provided in one embodiment of the present invention;

[0052] Figure 5 A schematic diagram of changes in accuracy, precision, FAR and FRR before and after feature fusion provided by an embodiment of the present invention;

[0053] Figure 6 A schematic diagram of the structure of an identity authentication device based on a touch-vibration signal provided by an embodiment of the present invention;

[0054] Figure 7 A schematic diagram of the structure of a medium according to an embodiment of the present invention is schematically shown;

[0055] Figure 8 A schematic diagram of the structure of a computing device according to an embodiment of the present invention is schematically shown.

[0056] In the drawings, the same or corresponding reference numerals represent the same or corresponding parts. DETAILED DESCRIPTION

[0057] The principles and spirit of the present invention will be described below with reference to several exemplary embodiments. It should be understood that these embodiments are provided only to enable those skilled in the art to better understand and implement the present invention, and are not intended to limit the scope of the present invention in any way. On the contrary, these embodiments are provided to make the present disclosure more thorough and complete, and to fully convey the scope of the present disclosure to those skilled in the art.

[0058] Those skilled in the art will appreciate that the embodiments of the present invention may be implemented as a system, device, apparatus, method or computer program product. Therefore, the present disclosure may be specifically implemented in the following forms, namely: complete hardware, complete software (including firmware, resident software, microcode, etc.), or a combination of hardware and software.

[0059] According to an embodiment of the present invention, a method, device, equipment, medium and product for identity authentication based on a touch vibration signal are proposed.

[0060] It should be noted that any number of elements in the drawings is for illustration rather than limitation, and any naming is only for distinction and does not have any limiting meaning.

[0061] The principle and spirit of the present invention are explained in detail below with reference to several representative embodiments of the present invention.

[0062] Exemplary Methods

[0063] Reference below Figure 1 , Figure 1 This is a flow chart of an identity authentication method based on a vibration signal according to an embodiment of the present invention. It should be noted that the embodiments of the present invention can be applied to applicable terminal devices.

[0064] Figure 1 The process of the identity authentication method based on the touch vibration signal provided by an embodiment of the present invention includes:

[0065] Step S101: when it is detected that a user presses the terminal device, a touch vibration signal generated by the user pressing the terminal device is collected.

[0066] In the embodiment of the present invention, the terminal device may be a smart phone, a tablet computer, smart glasses or other devices, which is not limited in the embodiment of the present invention.

[0067] In the embodiment of the present invention, after it is detected that the user presses the terminal device, a touch vibration signal generated by the user pressing the terminal device can be collected by an accelerometer built into the terminal device.

[0068] For example, the sensor logger software can be used to collect touch and vibration signals. During the collection of touch and vibration signals, the sampling rate is set to 400 Hz to ensure sufficient time resolution to capture the slight differences in the user's pressing behavior.

[0069] In the embodiment of the present invention, the touch vibration signal may include a plurality of three-axis acceleration data in a continuous time period, and the three-axis acceleration data may be acceleration data of an x-axis, a y-axis, and a z-axis measured by an accelerometer.

[0070] Step S102 , preprocessing the touch-vibration signal to obtain a preprocessed touch-vibration signal.

[0071] In the embodiment of the present invention, after collecting the user's touch vibration signals, in order to ensure the validity and accuracy of the touch vibration signals, these original touch vibration signals must be preprocessed so that the subsequent feature extraction process can effectively use these data.

[0072] Due to the influence of gravity acceleration, the acceleration readings of the x-axis, y-axis, and z-axis will be offset. Therefore, the gravity component must be corrected and removed to obtain a pure motion acceleration signal (i.e., a touch vibration signal).

[0073] In order to solve this problem, the present invention adopts a gravity correction method based on filtering factors.

[0074] As an optional implementation manner, step S102 pre-processes the touch-vibration signal, and a method of obtaining the pre-processed touch-vibration signal may include:

[0075] Performing gravity correction on the touch vibration signal based on a preset filter factor to obtain a first touch vibration signal;

[0076] performing standardization processing on the first touch-vibration signal to obtain a second touch-vibration signal;

[0077] Based on a preset low cutoff frequency and a preset high cutoff frequency, the second touch-vibration signal is filtered to obtain a preprocessed touch-vibration signal.

[0078] Among them, by implementing this embodiment, the influence of gravity acceleration can be eliminated through gravity correction, and the obtained pre-processed touch vibration signal can be made smoother and more representative through standardization and filtering.

[0079] For example, suppose the user uses the terminal device in a stationary state and the terminal device is parallel to the ground. At this time, the acceleration component of the z-axis is close to the full value of the gravity acceleration, while the readings of the x-axis and y-axis are small. In order to correct these data, the system introduces a filter factor β (for example, set to 0.2) to adjust the original acceleration reading a i , the first touch vibration signal a' i , the calculation formula of i∈{x,y,z} is as follows:

[0080] a' i =(1-β)·(a i -g i ), i∈{x,y,z}

[0081] In addition, in order to improve the stability and consistency of subsequent processing, the first touch vibration signal of each axis may be standardized.

[0082] Specifically, by calculating the mean value μ of each axial direction of the first touch vibration signal i and standard deviation σ i , and adjust each first touch vibration signal to a form of zero mean and unit variance, that is, the form of the first touch vibration signal is:

[0083]

[0084] The Euclidean distance of the adjusted first tactile vibration signal may also be calculated to comprehensively reflect the overall change characteristics of the acceleration of the tactile vibration signal in space, thereby obtaining the second tactile vibration signal. The calculation formula of the Euclidean distance M is:

[0085]

[0086] Among them, a x 、a y and a z Represents the normalized acceleration values ​​of the x-axis, y-axis, and z-axis respectively.

[0087] In order to further eliminate the noise caused by environmental and human factors and improve the smoothness of the second touch vibration signal, a bandpass filter is applied for processing. The filter is designed with a low cutoff frequency f of 10 Hz. low and a high cutoff frequency f of 150Hz high , through the filter, can effectively suppress the low and above f high frequency components, ensuring that the retained pre-processed touch-vibration signal has better smoothness and representativeness.

[0088] Step S103 : extracting features from the preprocessed tactile-vibration signal to obtain global features and local features corresponding to the preprocessed tactile-vibration signal.

[0089] As an optional implementation manner, step S103 performs feature extraction on the pre-processed tactile vibration signal to obtain global features and local features corresponding to the pre-processed tactile vibration signal, which may include:

[0090] Performing global extraction on the preprocessed touch-vibration signal to obtain a global feature corresponding to the preprocessed touch-vibration signal;

[0091] Encoding the preprocessed touch-vibration signal to obtain key features of the preprocessed touch-vibration signal;

[0092] The key feature is decoded to obtain a local feature of the pre-processed touch-vibration signal.

[0093] Among them, by implementing this implementation, the features of the pre-processed touch vibration signal can be extracted more accurately and comprehensively by extracting global features and local features, thereby improving the accuracy of feature extraction.

[0094] In an embodiment of the present invention, the pre-processed touch-vibration signal can be encoded and decoded by an autoencoder (AE) to obtain local features of the pre-processed touch-vibration signal. AE can extract potential nonlinear features of the pre-processed touch-vibration signal, compress the pre-processed touch-vibration signal into a latent space through a series of encoding and decoding operations, and learn key features of the signal therefrom, thereby obtaining local features.

[0095] In order to optimize the AE model, the model parameters can be continuously adjusted through the back propagation algorithm and the optimizer (Adam), with the goal of minimizing the reconstruction error L MSE :

[0096]

[0097] Among them, x iis the original input signal. By minimizing the reconstruction error, AE can not only extract efficient feature representation, but also retain the key information in the signal, ensuring that the model accurately reconstructs the input data. The number of short-time frame signals is N.

[0098] Specifically, a method of performing global extraction on the preprocessed tactile vibration signal to obtain a global feature corresponding to the preprocessed tactile vibration signal may include:

[0099] Based on a preset frame duration and frame shift, dividing the pre-processed touch-vibration signal into a plurality of short-time frame signals;

[0100] Performing Fourier transform on each of the short-time frame signals to obtain a frequency spectrum of the pre-processed touch-vibration signal;

[0101] Filtering the frequency spectrum through a Mel filter bank to obtain a Mel frequency energy distribution;

[0102] Calculating the Mel frequency energy distribution to obtain a logarithmic Mel frequency spectrum;

[0103] Performing discrete cosine transform on the logarithmic Mel spectrum to obtain Mel frequency coefficients;

[0104] Performing differential calculation on the Mel frequency coefficient to obtain a first-order difference and a second-order difference of the Mel frequency coefficient;

[0105] The Mel frequency coefficient, the first-order difference, and the second-order difference are determined as global features corresponding to the pre-processed tactile vibration signal.

[0106] Among them, by implementing this embodiment, the pre-processed touch vibration signal can be processed to obtain the Mel frequency coefficient of the pre-processed touch vibration signal, and then the global features of the pre-processed touch vibration signal can be obtained based on the first-order difference and second-order difference of the Mel frequency coefficient. The global features obtained in this way can more accurately represent the characteristics of the user's pressing behavior.

[0107] In the embodiment of the present invention, MFCC is a feature extraction method commonly used in the field of signal processing, and is particularly suitable for the analysis of speech and vibration signals. MFCC extracts the frequency information of the signal into a set of representative feature coefficients by simulating the auditory system of the human ear.

[0108] In the embodiment of the present invention, a pre-processed tactile vibration signal with a sampling rate of 400 Hz can be used as input, and the signal can be divided into a plurality of short-time frame signals with a preset frame duration (e.g., 25 milliseconds), and a frame shift of 10 milliseconds. Subsequently, a short-time Fourier transform (STFT) is performed on each short-time frame signal to obtain the spectrum of the pre-processed tactile vibration signal, and the mathematical expression of the spectrum X[k] is as follows:

[0109]

[0110] Wherein, k represents the frequency index, x[n] represents the preprocessed touch vibration signal, and j represents the imaginary unit.

[0111] The spectrum is filtered by a Mel filter bank to obtain a Mel frequency energy distribution, and then the logarithm of the Mel frequency energy distribution is calculated to obtain a logarithmic Mel spectrum, and a discrete cosine transform (DCT) is applied to obtain 13 Mel frequency coefficients (MFCC coefficients):

[0112]

[0113] Where M is the logarithmic Mel spectrum, m is the Mel frequency energy distribution, and n represents the index of the MFCC coefficient, ranging from 1 to 13.

[0114] In order to capture the dynamic characteristics of the signal, the first-order difference and second-order difference of the MFCC coefficients are further calculated to generate a feature vector matrix containing 39 dimensional features, which is the global feature corresponding to the preprocessed touch-vibration signal.

[0115] In an embodiment of the present invention, feature extraction can be trained; the training data can be collected from subjects aged between 22 and 48 years old. Each subject presses the bottom frame of the mobile phone screen with his thumb 1 to 2 cm to register his personal identity. The subject presses the screen with habitual force to ensure that the pressure is natural and consistent with daily usage habits. Each press lasts for 5 seconds to capture the complete vibration response. The vibration signal in this process is recorded in real time by the accelerometer sensor of the mobile phone, and finally forms a CSV file containing time information and corresponding x / y / z three-axis data.

[0116] Step S104, performing feature fusion on the global feature and the local feature to obtain a fused feature.

[0117] As an optional implementation manner, step S104 performs feature fusion on the global feature and the local feature to obtain the fused feature, which may include:

[0118] The global features are calculated based on a multi-head attention mechanism to obtain global information;

[0119] Processing the local features based on an inverse neural network to obtain local fusion features;

[0120] The global information and the local fusion feature are fused to obtain a fusion feature.

[0121] Among them, implementing this implementation method can fuse the multi-level features of global features and local features, and enhance the expression ability of the fused features.

[0122] As an optional implementation manner, the method of calculating the global features based on the multi-head attention mechanism to obtain the global information may include:

[0123] Performing layer normalization on the global features to obtain normalized global features;

[0124] Calculating the normalized global features based on a multi-head attention mechanism to obtain initial global information;

[0125] Performing a residual connection on the global feature and the initial global information to obtain residual connection information;

[0126] The residual connection information is nonlinearly mapped through a multi-layer perceptron to obtain global information.

[0127] In the embodiment of the present invention, the calculation formula of the initial global information Attention (Q, K, V) can be:

[0128]

[0129] Among them, Q, K, and V are the query, key, and value matrices of the global feature, respectively. k is the dimension of the key, and T represents the transpose of the matrix K.

[0130] First, the input global features are layer-normalized so that the features remain uniform before entering the self-attention mechanism. After layer normalization, the feature matrix is ​​calculated through the multi-head self-attention mechanism to calculate its similarity and importance at different time steps to obtain the initial global information.

[0131] In the embodiment of the present invention, the calculation formula of the global information Z is:

[0132] z=MLP(LayerNorm(X norm +Attention(Q,K,V)))

[0133] Among them, X norm Represents global features, LayerNorm() represents the residual connection function, and MLP is a multilayer perceptron.

[0134] Through the multi-head self-attention mechanism, these calculation results are spliced ​​between different heads to form a richer feature representation. The output of the multi-head self-attention mechanism is residually connected with the original input features, and then nonlinearly mapped through a multi-layer perceptron (MLP) to enhance the expression of global information.

[0135] In an embodiment of the present invention, an inverted neural network (INN) may be used to further process the local features extracted by AE.

[0136] Specifically, the local features can be subjected to multi-layer nonlinear mapping processing to obtain local fusion features, thereby enhancing the expression ability of the local fusion features, especially when processing complex nonlinear signals, showing more outstanding effects.

[0137] Finally, the global information and local fusion features can be integrated through the feature fusion layer to generate the final fusion features. This fusion feature contains both global information and effectively combines local features, providing a richer and more accurate feature representation for the authentication model.

[0138] Step S105, processing the fused features through a pre-built identity authentication model to obtain identity information corresponding to the fused features.

[0139] In the embodiment of the present invention, the pre-built identity authentication model uses convolutional neural networks (CNN) to process the characteristics of the vibration signal. The CNN model is known for its powerful feature extraction capabilities for spatial and temporal data, and is particularly suitable for processing complex signal data.

[0140] In the specific implementation, four convolutional layers and two fully connected layers are selected to form the main architecture of the CNN model. The filter sizes of the convolutional layers are (64, 64, 256, 256) in sequence. Each convolutional layer uses the ReLU activation function to enhance the nonlinear expression ability of the features, and the pooling layer is used to reduce the size of the features, thereby effectively reducing the computational complexity while ensuring the performance of the model. In the fully connected layer part, the model still uses the ReLU activation function for nonlinear transformation, and uses the Softmax activation function in the output layer to map the prediction results to the probability distribution of the multi-classification task.

[0141] Considering the generalization ability of the model and the small number of samples, the present invention further uses data enhancement technology to expand the training data set. Specifically, the identity authentication model increases the number of each sample by 6 times by introducing random noise into the original touch vibration signal. These enhanced samples retain the key features of the original touch vibration signal while adding a small amount of randomness.

[0142] In the training process, in order to optimize the model parameters, this study used the cross-entropy loss function to measure the difference between the model prediction value and the actual label, and updated the model parameters through the gradient descent method. Ultimately, the CNN model can effectively improve the complexity of the fusion features in the classified tactile vibration signal, thereby providing strong support for the identity authentication task.

[0143] Please also read Figure 2 and Figure 3 , Figure 2 The distribution of FAR and FRR of different types of models provided by an embodiment of the present invention; Figure 3 A schematic diagram comparing the classification accuracy of different types of models provided in an embodiment of the present invention; the CNN model performs well in all three indicators, reaching 2.17% FAR, 3.28% FRR and 95.38% classification accuracy respectively. This shows that the CNN model can effectively and accurately classify user identities in most cases, showing strong generalization ability and robustness.

[0144] In comparison, other models such as Support Vector Machine (SVM), GBDT and RF also performed well, but were slightly inferior in some indicators. Overall, it shows that all models have good recognition capabilities and can accurately distinguish between legitimate and illegal users.

[0145] For example, on a laptop equipped with an Intel i7-13650HX processor and 16GB RAM, the Windows 11 operating system is running. The PyTorch framework is used to build and train deep learning models to process and analyze touch-vibration signals. The data acquisition part uses the Sensor Logger software, with the sampling frequency set to 400Hz, to collect relevant data on touch-vibration behavior from multiple experimental subjects.

[0146] In addition, we introduced AE to enhance the effect of feature extraction and compared and analyzed its impact on model performance. Specifically, on the basis of traditional feature extraction methods, AE was added to capture the potential complex features in the touch vibration signal. In the experiment, we used feature extraction methods based only on MFCC and based on MFCC combined with AE to train the identity recognition model.

[0147] Please also read Figure 4 and Figure 5 , Figure 4 A schematic diagram of the changes in FAR and FRR of a model using MFCC and MFCC combined with AE feature extraction methods provided in one embodiment of the present invention; Figure 5A schematic diagram of the changes in accuracy, precision, FAR and FRR before and after feature fusion provided by an embodiment of the present invention.

[0148] like Figure 5 As shown in the figure, after the introduction of AE, the FAR of the model decreased by 0.73%, while the FRR decreased by 1.07%. This shows that after adding the features extracted by AE, the performance of the model in identifying legitimate users from illegitimate users has been significantly improved. Compared with the simple use of MFCC feature extraction, AE can extract nonlinear and complex features in the signal, thereby improving the robustness and accuracy of the model in dealing with complex environments and different user behavior patterns.

[0149] In order to explore the impact of feature fusion on model performance, we trained the model using feature data before and after fusion. Figure 6 The comparison of relevant key indicators before and after feature fusion is listed in .

[0150] like Figure 6 As shown in the figure, the performance of the feature fusion module is better. The accuracy and precision of the model have been improved to a certain extent, which shows that the fused model is more accurate in distinguishing legitimate users from illegal users, reducing the situation of misidentifying illegal users as legitimate users.

[0151] The reduction of FAR after feature fusion indicates that the model is more reliable in dealing with illegal users and reduces the risk of falsely accepting illegal users. Although the decrease in FRR after fusion is small, it still shows the potential of feature fusion in keeping the model friendly to legitimate users.

[0152] The present invention can quickly and accurately realize identity authentication when dealing with complex environmental changes. In addition, the present invention can also make the obtained pre-processed touch vibration signal have better smoothness and representativeness. In addition, the present invention can also improve the accuracy of feature extraction. In addition, the present invention can also make the obtained global features more accurately represent the characteristics of the user's pressing behavior. In addition, the present invention can also enhance the expression ability of the fusion feature.

[0153] Exemplary Devices

[0154] After introducing the method of the exemplary embodiment of the present invention, next, refer to Figure 6 An identity authentication device based on a touch vibration signal according to an exemplary embodiment of the present invention is described. The device is applied to a terminal device and includes:

[0155] The collecting unit 601 is used to collect a touch vibration signal generated by the user pressing the terminal device when it is detected that the user presses the terminal device;

[0156] A preprocessing unit 602 is used to preprocess the touch-vibration signal to obtain a preprocessed touch-vibration signal;

[0157] An extraction unit 603 is used to extract features from the pre-processed tactile vibration signal to obtain global features and local features corresponding to the pre-processed tactile vibration signal;

[0158] A fusion unit 604 is used to fuse the global feature and the local feature to obtain a fused feature;

[0159] The processing unit 605 is used to process the fused feature through a pre-built identity authentication model to obtain identity information corresponding to the fused feature.

[0160] As an optional implementation manner, the preprocessing unit 602 preprocesses the touch-vibration signal to obtain the preprocessed touch-vibration signal in the following manner:

[0161] Performing gravity correction on the touch vibration signal based on a preset filter factor to obtain a first touch vibration signal;

[0162] performing standardization processing on the first touch-vibration signal to obtain a second touch-vibration signal;

[0163] Based on a preset low cutoff frequency and a preset high cutoff frequency, the second touch-vibration signal is filtered to obtain a preprocessed touch-vibration signal.

[0164] Among them, by implementing this embodiment, the influence of gravity acceleration can be eliminated through gravity correction, and the obtained pre-processed touch vibration signal can be made smoother and more representative through standardization and filtering.

[0165] As an optional implementation manner, the extraction unit 603 performs feature extraction on the pre-processed tactile vibration signal to obtain the global features and local features corresponding to the pre-processed tactile vibration signal in the following manner:

[0166] Performing global extraction on the preprocessed touch-vibration signal to obtain a global feature corresponding to the preprocessed touch-vibration signal;

[0167] Encoding the preprocessed touch-vibration signal to obtain key features of the preprocessed touch-vibration signal;

[0168] The key feature is decoded to obtain a local feature of the pre-processed touch-vibration signal.

[0169] Among them, by implementing this implementation, the features of the pre-processed touch vibration signal can be extracted more accurately and comprehensively by extracting global features and local features, thereby improving the accuracy of feature extraction.

[0170] As an optional implementation manner, the extraction unit 603 performs global extraction on the pre-processed tactile vibration signal to obtain the global feature corresponding to the pre-processed tactile vibration signal in the following manner:

[0171] Based on a preset frame duration and frame shift, dividing the pre-processed touch-vibration signal into a plurality of short-time frame signals;

[0172] Performing Fourier transform on each of the short-time frame signals to obtain a frequency spectrum of the pre-processed touch-vibration signal;

[0173] Filtering the frequency spectrum through a Mel filter bank to obtain a Mel frequency energy distribution;

[0174] Calculating the Mel frequency energy distribution to obtain a logarithmic Mel frequency spectrum;

[0175] Performing discrete cosine transform on the logarithmic Mel spectrum to obtain Mel frequency coefficients;

[0176] Performing differential calculation on the Mel frequency coefficient to obtain a first-order difference and a second-order difference of the Mel frequency coefficient;

[0177] The Mel frequency coefficient, the first-order difference, and the second-order difference are determined as global features corresponding to the pre-processed tactile vibration signal.

[0178] Among them, by implementing this embodiment, the pre-processed touch vibration signal can be processed to obtain the Mel frequency coefficient of the pre-processed touch vibration signal, and then the global features of the pre-processed touch vibration signal can be obtained based on the first-order difference and second-order difference of the Mel frequency coefficient. The global features obtained in this way can more accurately represent the characteristics of the user's pressing behavior.

[0179] As an optional implementation manner, the fusion unit 604 performs feature fusion on the global feature and the local feature to obtain the fused feature in the following manner:

[0180] The global features are calculated based on a multi-head attention mechanism to obtain global information;

[0181] Processing the local features based on an inverse neural network to obtain local fusion features;

[0182] The global information and the local fusion feature are fused to obtain a fusion feature.

[0183] Among them, implementing this implementation method can fuse the multi-level features of global features and local features, and enhance the expression ability of the fused features.

[0184] The present invention can quickly and accurately realize identity authentication when dealing with complex environmental changes. In addition, the present invention can also make the obtained pre-processed touch vibration signal have better smoothness and representativeness. In addition, the present invention can also improve the accuracy of feature extraction. In addition, the present invention can also make the obtained global features more accurately represent the characteristics of the user's pressing behavior. In addition, the present invention can also enhance the expression ability of the fusion feature.

[0185] Exemplary Media

[0186] After introducing the method and apparatus of the exemplary embodiment of the present invention, next, reference is made to Figure 7 For a description of a computer-readable storage medium according to an exemplary embodiment of the present invention, please refer to Figure 7 , the computer-readable storage medium shown is a CD 70, on which a computer program (i.e., a program product) is stored. When the computer program is executed by the processor, the steps recorded in the above method implementation are implemented. For example, when it is detected that the user presses the terminal device, the touch vibration signal generated by the user pressing the terminal device is collected; the touch vibration signal is preprocessed to obtain a preprocessed touch vibration signal; feature extraction is performed on the preprocessed touch vibration signal to obtain global features and local features corresponding to the preprocessed touch vibration signal; feature fusion is performed on the global features and the local features to obtain fused features; the fused features are processed by a pre-constructed identity authentication model to obtain identity information corresponding to the fused features; the specific implementation method of each step is not repeated here.

[0187] It should be noted that examples of the computer-readable storage medium may also include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other optical or magnetic storage media, which are not listed here one by one.

[0188] Exemplary Computing Devices

[0189] After introducing the method, apparatus and medium of the exemplary embodiments of the present invention, next, reference is made to Figure 8 A computing device for identity authentication based on a tactile vibration signal according to an exemplary embodiment of the present invention.

[0190] Figure 8 A block diagram of an exemplary computing device 80 , which may be a computer system or a server, is shown that is suitable for use in implementing embodiments of the present invention. Figure 8The computing device 80 shown is only an example and should not bring any limitation to the functionality and scope of use of the embodiments of the present invention.

[0191] like Figure 8 As shown, the components of the computing device 80 may include, but are not limited to: one or more processors or processing units 801, a system memory 802, and a bus 803 connecting different system components (including the system memory 802 and the processing unit 801).

[0192] The computing device 80 typically includes a variety of computer system readable media. These media can be any available media that can be accessed by the computing device 80, including volatile and non-volatile media, removable and non-removable media.

[0193] The system memory 802 may include computer system readable media in the form of volatile memory, such as random access memory (RAM) 8021 and / or cache memory 8022. The computing device 80 may further include other removable / non-removable, volatile / non-volatile computer system storage media. By way of example only, the ROM 8023 may be used to read and write non-removable, non-volatile magnetic media ( Figure 8 is not shown in the Figure 8 As shown in FIG. 8 , a disk drive for reading and writing a removable non-volatile disk (e.g., a “floppy disk”) and an optical disk drive for reading and writing a removable non-volatile optical disk (e.g., a CD-ROM, a DVD-ROM, or other optical media) may be provided. In these cases, each drive may be connected to bus 803 via one or more data medium interfaces. System memory 802 may include at least one program product having a set (e.g., at least one) of program modules that are configured to perform the functions of various embodiments of the present invention.

[0194] A program / utility 8025 having a set (at least one) of program modules 8024 may be stored, for example, in system memory 802, and such program modules 8024 include, but are not limited to, an operating system, one or more application programs, other program modules, and program data, each of which or some combination may include an implementation of a network environment. Program modules 8024 generally perform the functions and / or methods of the embodiments described herein.

[0195] The computing device 80 may also communicate with one or more external devices 804 (e.g., a keyboard, a pointing device, a display, etc.). Such communication may be performed via an input / output (I / O) interface 805. Furthermore, the computing device 80 may also communicate with one or more networks (e.g., a local area network (LAN), a wide area network (WAN), and / or a public network, such as the Internet) via a network adapter 806. Figure 8 As shown, the network adapter 806 communicates with other modules (such as the processing unit 801, etc.) of the computing device 80 via the bus 803. It should be understood that although Figure 8 Not shown, other hardware and / or software modules may be used in conjunction with computing device 80 .

[0196] The processing unit 801 executes various functional applications and data processing by running the program stored in the system memory 802. For example, when it is detected that the user presses the terminal device, the touch vibration signal generated by the user pressing the terminal device is collected; the touch vibration signal is preprocessed to obtain a preprocessed touch vibration signal; the preprocessed touch vibration signal is feature extracted to obtain the global features and local features corresponding to the preprocessed touch vibration signal; the global features and the local features are feature fused to obtain a fused feature; the fused feature is processed by a pre-constructed identity authentication model to obtain the identity information corresponding to the fused feature. The specific implementation of each step is not repeated here. It should be noted that although several units / modules or sub-units / sub-modules of the identity authentication device based on the touch vibration signal are mentioned in the above detailed description, this division is only exemplary and not mandatory. In fact, according to an embodiment of the present invention, the features and functions of two or more units / modules described above can be concretized in one unit / module. Conversely, the features and functions of one unit / module described above can be further divided into multiple units / modules to be concretized.

[0197] In the description of the present invention, it should be noted that the terms “first”, “second” and “third” are only used for descriptive purposes and cannot be understood as indicating or implying relative importance.

[0198] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0199] In the several embodiments provided by the present invention, it should be understood that the disclosed systems, devices and methods can be implemented in other ways. The device embodiments described above are only schematic. For example, the division of the units is only a logical function division. There may be other division methods in actual implementation. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some communication interfaces, and the indirect coupling or communication connection of devices or units can be electrical, mechanical or other forms.

[0200] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0201] In addition, each functional unit in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit.

[0202] If the functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a non-volatile computer-readable storage medium that is executable by a processor. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or the 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 for enabling a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the methods described in each embodiment of the present invention. The aforementioned storage medium includes: various media that can store program codes, such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk.

[0203] Finally, it should be noted that the above-described embodiments are only specific implementations of the present invention, which are used to illustrate the technical solutions of the present invention, rather than to limit them. The protection scope of the present invention is not limited thereto. Although the present invention is described in detail with reference to the above-described embodiments, ordinary technicians in the field should understand that any technician familiar with the technical field can still modify the technical solutions recorded in the above-described embodiments within the technical scope disclosed by the present invention, or can easily think of changes, or make equivalent replacements for some of the technical features therein; and these modifications, changes or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present invention, and should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention shall be subject to the protection scope of the claims.

[0204] In addition, although the operations of the method of the present invention are described in a specific order in the drawings, this does not require or imply that the operations must be performed in this specific order, or that all the operations shown must be performed to achieve the desired results. Additionally or alternatively, some steps may be omitted, multiple steps may be combined into one step, and / or one step may be decomposed into multiple steps.

[0205] In an exemplary embodiment, a computer program product is provided, including a computer program, and when the computer program is executed by a processor, the steps in the above method embodiments are implemented.

Claims

1. An identity authentication method based on a touch vibration signal, applied to a terminal device, characterized in that: The method comprises: When it is detected that the user presses the terminal device, collecting a touch vibration signal generated by the user pressing the terminal device; Preprocessing the touch-vibration signal to obtain a preprocessed touch-vibration signal; Performing feature extraction on the preprocessed touch-vibration signal to obtain global features and local features corresponding to the preprocessed touch-vibration signal; Performing feature fusion on the global feature and the local feature to obtain a fused feature; The fused features are processed through a pre-built identity authentication model to obtain identity information corresponding to the fused features.

2. The identity authentication method based on touch vibration signal according to claim 1, characterized in that: The preprocessing of the touch-vibration signal to obtain a preprocessed touch-vibration signal includes: Performing gravity correction on the touch vibration signal based on a preset filter factor to obtain a first touch vibration signal; performing standardization processing on the first touch-vibration signal to obtain a second touch-vibration signal; Based on a preset low cutoff frequency and a preset high cutoff frequency, the second touch-vibration signal is filtered to obtain a preprocessed touch-vibration signal.

3. The identity authentication method based on touch vibration signal according to claim 1, characterized in that: The extracting features of the pre-processed tactile vibration signal to obtain global features and local features corresponding to the pre-processed tactile vibration signal includes: Performing global extraction on the preprocessed touch-vibration signal to obtain a global feature corresponding to the preprocessed touch-vibration signal; Encoding the preprocessed touch-vibration signal to obtain key features of the preprocessed touch-vibration signal; The key feature is decoded to obtain a local feature of the pre-processed touch-vibration signal.

4. The identity authentication method based on touch vibration signal according to claim 3, characterized in that: The globally extracting the preprocessed tactile vibration signal to obtain a global feature corresponding to the preprocessed tactile vibration signal includes: Based on a preset frame duration and frame shift, dividing the pre-processed touch-vibration signal into a plurality of short-time frame signals; Performing Fourier transform on each of the short-time frame signals to obtain a frequency spectrum of the pre-processed touch-vibration signal; Filtering the frequency spectrum through a Mel filter bank to obtain a Mel frequency energy distribution; Calculating the Mel frequency energy distribution to obtain a logarithmic Mel frequency spectrum; Performing discrete cosine transform on the logarithmic Mel spectrum to obtain Mel frequency coefficients; Performing differential calculation on the Mel frequency coefficient to obtain a first-order difference and a second-order difference of the Mel frequency coefficient; The Mel frequency coefficient, the first-order difference, and the second-order difference are determined as global features corresponding to the pre-processed tactile vibration signal.

5. The identity authentication method based on touch vibration signal according to claim 1, characterized in that: The step of fusing the global feature and the local feature to obtain a fused feature includes: The global features are calculated based on a multi-head attention mechanism to obtain global information; Processing the local features based on an inverse neural network to obtain local fusion features; The global information and the local fusion feature are fused to obtain a fusion feature.

6. The identity authentication method based on touch vibration signal according to claim 5, characterized in that: The global features are calculated based on the multi-head attention mechanism to obtain global information, including: Performing layer normalization on the global features to obtain normalized global features; Calculating the normalized global features based on a multi-head attention mechanism to obtain initial global information; Performing a residual connection on the global feature and the initial global information to obtain residual connection information; The residual connection information is nonlinearly mapped through a multi-layer perceptron to obtain global information.

7. An identity authentication device based on a touch vibration signal, applied to a terminal device, characterized in that: The device comprises: a collecting unit, configured to collect a touch vibration signal generated by the user pressing the terminal device when it is detected that the user presses the terminal device; A preprocessing unit, used for preprocessing the touch-vibration signal to obtain a preprocessed touch-vibration signal; an extraction unit, configured to perform feature extraction on the preprocessed tactile-vibration signal to obtain global features and local features corresponding to the preprocessed tactile-vibration signal; A fusion unit, used for fusing the global feature and the local feature to obtain a fused feature; The processing unit is used to process the fused feature through a pre-built identity authentication model to obtain identity information corresponding to the fused feature.

8. A computing device, characterized in that The computing device comprises: at least one processor, memory, and input-output unit; The memory is used to store a computer program, and the processor is used to call the computer program stored in the memory to execute the method according to any one of claims 1 to 10.

9. A computer-readable storage medium comprising instructions, characterized in that: When the method is executed on a computer, the computer is enabled to execute the method according to any one of claims 1 to 10.

10. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the method according to any one of claims 1 to 6 is implemented.

Images