CN110363120A - Intelligent terminal based on vibration signal touches authentication method and system - Google Patents
Intelligent terminal based on vibration signal touches authentication method and system Download PDFInfo
- Publication number
- CN110363120A CN110363120A CN201910583135.8A CN201910583135A CN110363120A CN 110363120 A CN110363120 A CN 110363120A CN 201910583135 A CN201910583135 A CN 201910583135A CN 110363120 A CN110363120 A CN 110363120A
- Authority
- CN
- China
- Prior art keywords
- signal
- vibration
- vibration signal
- touching
- intelligent terminal
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F21/00—Security arrangements for protecting computers, components thereof, programs or data against unauthorised activity
- G06F21/30—Authentication, i.e. establishing the identity or authorisation of security principals
- G06F21/31—User authentication
- G06F21/32—User authentication using biometric data, e.g. fingerprints, iris scans or voiceprints
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
- G06F2218/08—Feature extraction
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
- G06F2218/12—Classification; Matching
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Computer Security & Cryptography (AREA)
- Software Systems (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Health & Medical Sciences (AREA)
- Biomedical Technology (AREA)
- Evolutionary Computation (AREA)
- Computational Linguistics (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Biophysics (AREA)
- Data Mining & Analysis (AREA)
- Mathematical Physics (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- Health & Medical Sciences (AREA)
- Computer Hardware Design (AREA)
- User Interface Of Digital Computer (AREA)
- Collating Specific Patterns (AREA)
Abstract
A kind of intelligent terminal touching authentication method and system based on vibration signal, when intelligent terminal detects that finger touches, by actively generating certain vibration signal and by IMU sensor collection vibration signal, and extract biological characteristic, behavioural characteristic and independent touching behavioural characteristic respectively from the vibration signal received;Then classified using the neural network based on twin network (siamese network) framework to biological characteristic, realize that the unrelated intelligent terminal of behavior touches certification.The present invention actively issues vibration signal using intelligent terminal, and the biological characteristic for capturing touching finger can be independent of the behavioural characteristic of touching to identify different users.
Description
Technical field
The present invention relates to a kind of technology in user authentication field, it is specifically a kind of it is based on vibration signal with behavior without
The intelligent terminal of pass touches authentication method and system.
Background technique
In recent years, with the continuous development of intelligent mobile terminal, it is stored in intelligent mobile terminal equipment, especially intelligence is whole
Privacy and sensitive data in end are continuously increased, and result in the huge hidden danger of privacy and leaking data.Therefore, safe and efficient
Intelligent terminal user Verification System has become current research hotspot.
Existing work is broadly divided into three classes, and first kind work is the intelligent terminal user Verification System based on password, main
It to include numerical ciphers, gesture password and graphical passwords etc..The common problem of this kind of customer certification systems, which is susceptible to, peeps
Depending on there are biggish security risks;Second class work be certain external biological characteristics based on user, mainly include fingerprint,
Recognition of face, iris recognition, Application on Voiceprint Recognition etc..The common problem of this kind of work is to generally require extras and environmentally sensitive
(humidity, illumination, noise etc.), and be easy by Replay Attack (replay attack);The work of third class is touched using finger
The behavioural characteristic for touching screen carries out user authentication, mainly includes touch position, touching dynamics, touching time etc..This kind of work
Common problem is susceptible to impersonation attack (mimic attack), and user experience is poor.
Summary of the invention
The present invention In view of the above shortcomings of the prior art, proposes that a kind of intelligent terminal touching based on vibration signal is recognized
Method and system are demonstrate,proved, actively issue vibration signal using intelligent terminal, capture the biological characteristic of touching finger, it is different to identify
User, can be independent of the behavioural characteristic (such as contact position, dynamics, duration) of touching.
The present invention is achieved by the following technical solutions:
The present invention relates to a kind of, and the intelligent terminal based on vibration signal touches authentication method, when intelligent terminal detects finger
When touch, by actively generating certain vibration signal and by IMU sensor collection vibration signal, and birth is therefrom extracted respectively
Object feature, behavioural characteristic and independent touching behavioural characteristic;Then using based on twin network (siamese network) frame
The neural network of structure classifies to biological characteristic, realizes that the unrelated intelligent terminal of behavior touches certification.
The certain vibration signal is generated by regulating switch signal, each of which cycle period includes vibration impulse letter
Number, motor activation signal and remained shock eliminate part.
The vibration signal received is preferably first divided into the transient state vibration stage, stable vibration by the collection vibration signal
Stage and decline stage.
The extraction, including the feature extraction based on Wavelet transformation, the feature extraction based on Cepstrum Transform, touch position
Feature extraction, touching dynamics feature extraction, specifically: it extracts to obtain the time-frequency obtained after wavelet transformation from the transient state vibration stage
Spectrum, the peak value for extracting cepstrum from the stable vibration stage, extracting audio signal from the microphone of mobile terminal and its corresponding time work
For touch position feature, extract energy value of the vibration signal near resonant frequencyAs touching dynamics
Feature, in which: frFor the resonant frequency of vibration signal, Δ f defines energy balane bandwidth, and f (t) is original signal, right in
Answer the vibration signal in transient state vibration stage.
The touching behavioural characteristic specifically refers to: user is when using intelligent terminal, contact of the finger with intelligent terminal,
Including with smart machine front, side and it is subsequent contact when behavioural characteristic, including touch position and touching dynamics it is special
Sign.
The neural network based on the twin network architecture, including two parallel sub-networks, each sub-network has
Identical structure and weight and respectively according to the characteristic value X extracted in the vibration signal of input1,X2It is obtained after sub-network pair
The character representation C answeredW(X1),CW(X2), and then obtain the distance of two character representations: DW(X1,X2)=| | Cw(X1)-Cw(X2)|
|。
The sub-network is the structure of time-delay neural network (TDNN), includes two convolutional layers and a full articulamentum.
It is a basic convolution module in each convolutional layer as core, one batch of standardization (BN) layer handles gradient problem and one
ReLU layers are used as activation primitive.
The training sample of the neural network has same label, i.e., from the same user and touching behavioural characteristic
Dissmilarity, i.e. touching behavioural characteristic similarity are lower than threshold value.
The touching behavioural characteristic similarity is measured by Pearson correlation coefficient:
WhenGreater than preset threshold h, then it is assumed that two samples have
Similar touching behavioural characteristic, otherwise it is assumed that the touching behavioural characteristic of the two is dissimilar.
The classification refers to: according to the distance for the corresponding character representation of two inputs that neural network obtains, judging two
Whether a input belongs to unification user, so realize touching behavior whether from registered users judgement.
Technical effect
Compared with prior art, with behavior unrelated touching of the present invention by building based on intelligent mobile terminal authenticates system
System, independent of any external equipment, realizes accurate user authentication.In system building process, the present invention combines intelligence
The rich sensors and easy exploiting of mobile terminal are combined using the vibrating motor and IMU sensor on intelligent mobile terminal
The related technology of vibration signal processing and the relevant knowledge of twin neural network, finally realize that a robustness is good and precision is high
Touching Verification System, can effectively resist Replay Attack (replay attack) and impersonation attack (mimic attack), and
It has no requirement to the touching mode of user, greatly improves user experience.
Detailed description of the invention
Fig. 1 is the different touching schematic diagrams of manpower and mobile phone;
Fig. 2 is system construction drawing of the invention;
Fig. 3 is vibration signal design diagram;
Fig. 4 is that vibration signal divides schematic diagram;
Fig. 5 is vibration signal wavelet transformation schematic three dimensional views;
Fig. 6 is vibration signal Cepstrum Transform schematic three dimensional views;
Fig. 7 is touch position feature extraction schematic diagram;
Fig. 8 is touching dynamics feature extraction schematic diagram;
Fig. 9 is that twin network training data select tactful schematic diagram;
Figure 10 is twin network architecture figure of the invention;
Figure 11 is the confusion matrix figure of user authentication accuracy rate;
Figure 12 is the success rate figure of impersonation attack under varying environment;
Figure 13 is the success rate figure of Replay Attack under varying environment.
Specific embodiment
As shown in Figure 1, touching intelligent terminal for manpower has different modes, it is desirable to be connect by capturing with intelligent terminal
The biological characteristic of the finger of touching can carry out certification identification to the user of touching, that is, realize under a variety of different touching modes
The unrelated touching user authentication of behavior.To achieve it, actively being sensed with IMU using the vibrating motor on intelligent terminal
Device, by the influence for the vibration signal that active is propagated in analysis finger touching, to capture the finger contacted with intelligent terminal
Biological characteristic.
The present embodiment includes being divided into two stages of registering and logging:
A registration phase: when user's touch intelligent mobile phone, first by utilizing the vibration motor master carried on intelligent terminal
The raw specific vibration signal of movable property, and pass through IMU sensor collection vibration signal simultaneously.Collecting signal then, it further
By the processing calibrated and divided, the signal segment comprising user information is obtained.Based on these signal segments, signal is extracted respectively
Feature (mixed biologic feature and behavioural characteristic) and independent touching behavioural characteristic.It is then based on signal characteristic and the behavior of extraction
Feature, by the neural network framework and corresponding training sample selection strategy that are based on twin network (siamese network)
To weaken influence of the touching behavioural characteristic to the signal characteristic extracted, and further the training touching user unrelated with behavior recognizes
Demonstrate,prove sorter model.
B entry stage: then the movement of capture finger touching and active transmission and the specific vibration signal of reception first are pressed
The signal characteristic that vibration signal is extracted according to the mode of registration phase, the classification obtained using the signal characteristic and registration phase of extraction
Device model can carry out authentication for touching person.The entry stage does not require the mode that finger is touched.
As shown in Fig. 2, the present embodiment is related to a kind of system for realizing the above method, comprising: vibration signal processing module, letter
Number characteristic extracting module, the unrelated classifier of behavior and authentication module, in which: vibration signal processing module and signal characteristic abstraction
Module is connected and is transferred through pretreated vibration signal information, and signal characteristic abstraction module classifier unrelated with behavior is connected simultaneously
It transmits from the signal characteristic and behavior characteristic information by being extracted in pretreated vibration signal, the unrelated classifier of behavior and certification
Module is connected and is transferred through trained sorter model information, and authentication module receives the signal characteristic that authentication phase is extracted and is connected
And it is transmitted to trained sorter model and obtains the authentication result of lander.
As shown in figure 3, the certain vibration signal is generated by regulating switch signal, specifically: one is generated first
Extremely short vibration impulse signal (< 0.1ms), and the about 10ms that pauses is used to do signal alignment and subsequent touching positioning, then swashs
Vibrating motor 90ms living is allowed to reach the stable vibration stage via the transient state vibration stage, stops 10ms then to eliminate remained shock, leads to
It crosses activation and stops continuing the vibration signal of 5 formation 500ms, since the duration is short, such vibration signal is almost lossless
User experience.
The collection vibration signal, i.e., received using IMU sensor, preferably when transmitting terminal and receiving end carry out
Between on alignment.Since transmission speed of the vibration signal in intelligent terminal is generally higher than 2000m/s, it is only necessary to less than 0.05ms
The intelligent terminal of a common size can be spread all over.And the sample frequency of IMU sensor corresponds to 1ms's often less than 1000Hz
Temporal analytical density, much larger than the transmission time of vibration signal, therefore the sampled point institute that IMU sensor is received vibration signal is right
The time answered is considered the sending time of vibration signal, to complete signal alignment.
The collection vibration signal, for the information for being included using vibration signal different phase, the vibration that will be received
Dynamic signal carries out segmentation as shown in Figure 4, specifically includes:
1) based on the information and signal alignment for sending signal, signal is divided into the vibration stage (A+B stage in Fig. 4) of 90ms
With the decline stage (C-stage in Fig. 4) of 10ms.
2) using the variance of vibration signal frequency variation as threshold value, the transient state vibration stage of vibration signal is further discriminated between
(A stage in Fig. 4) and the stable vibration stage (B-stage in Fig. 4).I.e. when the variance of vibration signal frequency variation is greater than threshold value h
When, it is believed that the stage is vibrated in transient state, when the variance of vibration signal frequency variation is less than threshold value h, it is believed that be in stable vibration
Stage.
The extraction, including the feature extraction based on Wavelet transformation, the feature extraction based on Cepstrum Transform, touch position
Feature, touching dynamics feature extraction, specific steps include:
1) feature is extracted from the transient state vibration stage for receiving signal, due to vibrating stage, the frequency of vibration signal in transient state
Constantly variation, needs to convert original vibration signal, is allowed to obtain good time domain and frequency domain resolution simultaneously.Therefore,
It is converted using vibration signal of the continuous wavelet transform (CWT) to the transient state vibration stage:
Wherein: CWTf(a, τ) is when obtaining
Frequency spectrum;F (t) is original signal, in the vibration signal in middle corresponding transient state vibration stage;ψa,τIt (t) is wavelet basis function, selection
Morlet function reaches better time domain and frequency domain resolution as wavelet basis function, the time-frequency spectrum that will be obtained after wavelet transformation
Signal characteristic as the transient state vibration stage.
As shown in figure 5, carrying out the three-dimensional time-frequency spectrum after Wavelet transformation for one section of transient state vibration signal, it can be seen that at that time
Domain and frequency domain resolution are all very high, it is shown that the details that vibration frequency changes over time also implies touching finger and believes vibration
Number influence details.
2) based on the feature extraction of Cepstrum Transform: while feature is extracted from the stable vibration stage for receiving signal, in stabilization
Vibration stage, vibration signal are stablized on resonant frequency, other relatively weak frequency contents, i.e. sideband (side- are masked
Band frequencies), and touch influence of the finger to the stable vibration stage and be often reflected on sideband.Therefore, in order to
The frequency content for showing sideband is obtained in stable vibration stage signal using Cepstrum Transform comprising the various frequencies including sideband
Rate ingredient: Cy(q)=F-1(log Sy(f (t))), in which: Cy(q) cepstrum to obtain;Sy(f (t)) is the power spectrum of signal
Density (PSD); F-1Corresponding inverse Fourier transform (IFFT), using obtained cepstrum as the signal characteristic in stable vibration stage.
As shown in fig. 6, for the cepstrum for the vibration signal touched corresponding to two different users, it can be seen that two users exist
Entirely different feature is shown on cepstrum.
3) touch position feature: the touch position information in order to obtain user is carried out using the arrival time (ToA) of signal
Measurement.Due to the transmission speed of vibration signal too fast (>200m/s), the sample frequency of IMU is too low (<1000Hz), uses intelligence
Microphone in terminal extracts touch position feature to capture the acoustic signals generated simultaneously with vibration signal, specific such as Fig. 7
Shown, vibration signal mainly reaches microphone via three paths, and first is to vibrate to transmit by interior of mobile phone, and Article 2 is
Sound wave linear transmission through the air, Article 3 are propagated by touching the reflection of finger.Due to different propagated speed
Distance it is different, the impulse stage (< 0.1ms) for sending vibration signal receives in microphone will form multiple and different peaks on signal
It is worth (as shown in Figure 7).The path of the corresponding interior of mobile phone vibration transmission of the peak value (peak A) occurred at first, second peak value (peak value
B) the path of corresponding sound wave linear transmission through the air, the corresponding reflection by touching finger of third peak value (peak C)
The path of propagation.Therefore, by these peak values corresponding time, the length of respective path can be calculated, touch position is reacted
Feature.Therefore, it extracts microphone to receive the peak value in signal and correspond to the time, as touch position feature.
4) touch dynamics feature extraction: the touching dynamics information in order to obtain user utilizes the energy variation of vibration signal
It measures.In order to avoid interference, energy value of the vibration signal near resonant frequency is calculatedWherein:
frFor the resonant frequency of vibration signal, Δ f defines energy balane bandwidth, 5Hz is set as in.
As shown in figure 8, for the situation of change of energy value E under different touching dynamics, it can be seen that touching dynamics is bigger, energy
Value E is about small, and apparent corresponding relationship is presented.Therefore, energy value E of the vibration signal near resonant frequency is extracted, as touching
Dynamics feature.
The unrelated classifier of the behavior is based on the neural network of twin network (siamese network) framework, tool
Body is the twin network that pairs of training sample is needed in training process, each training sample (is included signal characteristic, used
Family label and touching behavioural characteristic), user tag and touching behavioural characteristic are used as reference and is trained sample selection, specifically
As shown in table in Fig. 9, for there is one group of training sample of same label (from the same user), and if only if them
Touching behavioural characteristic dissmilarity when, they are elected to be training sample.Similarly, for having different labels (not from two
Same user) one group of training sample, and if only if they are elected to be training sample when their touching behavioural characteristic is similar.Two
Sample x1,x2Touching behavioural characteristic similarity measured by Pearson correlation coefficient:
WhenGreater than preset threshold value h, then it is assumed that two samples have similar touching behavioural characteristic, otherwise it is assumed that the two
It is dissimilar to touch behavioural characteristic.To force classifier to complete classification independent of touching behavioural characteristic, so that twin net
Network can further extract the biology unrelated with touching behavior from signal characteristic (mixed biologic feature and touching behavioural characteristic)
Feature.
The twin network is as shown in Figure 10, including two parallel sub-networks, each sub-network knot having the same
Structure and weight and respectively according to the characteristic value X extracted in the vibration signal of input1,X2Corresponding feature is obtained after sub-network
Indicate CW(X1),CW(X2), and then obtain the distance of two character representations: DW(X1,X2)=| | Cw(X1)-Cw(X2)||。
The present embodiment is related to a kind of application scenarios: SAMSUNG Galaxy S6, SAMSUNG Galaxy S7 are chosen,
Google Pixel, HTC U Ultra and Huawei Mate8 as description touching Verification System prototype give 15 not
Same volunteer carries out using with the actual effect of assessment.Experimentation carries out under three kinds of varying environments, i.e. laboratory environment
(lab), market environment (mall) and bar environment (bar), and different types of touching is studied, including the outer touching of the palm
(Off-Hand) and in the palm (In-Hand) is touched.
The present embodiment specifically includes the following steps:
Step 1: all 10 registrations enter system as legitimate user in 15 volunteers, in addition 5 are used as attacker
Trial enters system by way of touching at random.
Step 2: 5 attackers attempt to enter system using imitation (mimic) attack.
Step 3: 5 attackers attempt to enter system using playback (replay) attack.
There are three most important evaluation indexes:
The accuracy rate (Accuracy) of user authentication authenticates successful probability for different user.
The false rejection rate (False Reject Rate) of user experience is described, i.e., will register user's erroneous authentication to be non-
Registration user simultaneously causes the probability that cannot be introduced into system.
The false acceptance rate (False Accept Rate) of success attack rate is described, i.e., attacker is in unregistered situation
The lower probability for being successfully included in system.
Assessment result is as shown in the table, the volunteer of experiment is participated in for 15, it is shown that authenticate the accurate of different user
Rate.It can be seen from the figure that for the identification between registration user 92.3% can be reached, for potential nonregistered user
Identification can then reach 98.7%, show that proposed touching Verification System can safely and effectively complete user authentication work.
As shown in figure 11, for three kinds of different environment, common commercial smart phone there are two types of showing and showing is compared
The false rejection rate of Verification System (face authentication system of Alipay and the speech lock Verification System of wechat).As can be seen that branch
Precious face authentication system (bar environment) in illumination condition difference is paid to be difficult to correctly authenticate registration user, and the speech lock of wechat
Verification System (market environment and bar environment) when ambient noise is larger has higher false rejection rate.And the touching proposed
Verification System keeps lower false rejection rate under circumstances, substantially increases user experience.
As shown in figure 12, for varying environment and different touching modes, it is shown that system connects the mistake of impersonation attack
By rate.As can be seen that the system of proposition can effectively resist impersonation attack, and mistake connects under various environment and touching mode
2% or less is maintained at by rate.Figure 13 shows under varying environment and different touching modes that system connects the mistake of Replay Attack
By rate.Although false acceptance rate is integrally slightly above impersonation attack, remain in acceptable range, i.e., 5% or less.As a result
Show that the system proposed can effectively resist Replay Attack.
Above-mentioned specific implementation can by those skilled in the art under the premise of without departing substantially from principle and objective in different ways
Local directed complete set is carried out to it, protection scope be subject to and claims and do not limited by above-mentioned specific implementation, within its scope
Each implementation constrained.
Claims (9)
1. a kind of intelligent terminal based on vibration signal touches authentication method, which is characterized in that when intelligent terminal detects finger
When touch, by actively generating certain vibration signal and by IMU sensor collection vibration signal, and birth is therefrom extracted respectively
Object feature, behavioural characteristic and independent touching behavioural characteristic;Then using the neural network based on the twin network architecture to life
Object feature is classified, and realizes that the unrelated intelligent terminal of behavior touches certification;
The certain vibration signal is generated by regulating switch signal, each of which cycle period includes vibration impulse signal, horse
Part is eliminated up to activation signal and remained shock;
The touching behavioural characteristic specifically refers to: user is when using intelligent terminal, contact of the finger with intelligent terminal, including
With the front of smart machine, side and it is subsequent contact when behavioural characteristic, including touch position and touching dynamics feature.
2. the intelligent terminal according to claim 1 based on vibration signal touches authentication method, characterized in that the receipts
Collect vibration signal, the vibration signal received is first divided into transient state vibration stage, stable vibration stage and decline stage.
3. the intelligent terminal according to claim 1 based on vibration signal touches authentication method, characterized in that described mentions
It takes, including the feature extraction based on Wavelet transformation, the feature extraction based on Cepstrum Transform, touch position feature extraction, touching power
Feature extraction is spent, specifically: extract to obtain the time-frequency spectrum obtained after wavelet transformation, from the stable vibration stage from the transient state vibration stage
Extract cepstrum, extracted from the microphone of mobile terminal audio signal peak value and its corresponding time as touch position feature, mention
Take energy value of the vibration signal near resonant frequencyAs touching dynamics feature, in which: frFor vibration
The resonant frequency of signal, Δ f define energy balane bandwidth, and f (t) is original signal, in the vibration in middle corresponding transient state vibration stage
Dynamic signal.
4. the intelligent terminal according to claim 1 or 3 based on vibration signal touches authentication method, characterized in that described
Extraction, specifically include:
1) feature is extracted from the transient state vibration stage for receiving signal, due to vibrating the stage in transient state, the frequency of vibration signal is continuous
Variation, needs to convert original vibration signal, is allowed to obtain good time domain and frequency domain resolution simultaneously, therefore, utilizes
Continuous wavelet transform converts the vibration signal in transient state vibration stage:
Wherein: CWTf(a, τ) is the time-frequency spectrum obtained;f
It (t) is original signal, in the vibration signal in middle corresponding transient state vibration stage;ψa,τ(t) it is wavelet basis function, selects Morlet letter
Number is used as wavelet basis function to reach better time domain and frequency domain resolution, using the time-frequency spectrum obtained after wavelet transformation as transient state
The signal characteristic in vibration stage;
2) based on the feature extraction of Cepstrum Transform: being obtained in stable vibration stage signal using Cepstrum Transform comprising including sideband
Various frequency contents: Cy(q)=F-1(logSy(f (t))), in which: Cy(q) cepstrum to obtain;Sy(f (t)) is signal
Power spectral density;F-1Corresponding inverse Fourier transform (IFFT), using obtained cepstrum as the signal characteristic in stable vibration stage;
3) touch position feature: vibration signal reaches microphone via three paths, sends the impulse stage of vibration signal in wheat
Gram wind receives the path that the peak value occurred at first on signal correspond to interior of mobile phone vibration transmission, and second peak value, which corresponds to, passes through air
In sound wave linear transmission path, microphone is extracted in the path that the corresponding reflection by touching finger of third peak value is propagated
The peak value in signal and corresponding time are received, as touch position feature;
4) it touches dynamics feature extraction: calculating energy value of the vibration signal near resonant frequencyWherein:
frFor the resonant frequency of vibration signal, Δ f defines energy balane bandwidth.
5. the intelligent terminal according to claim 1 based on vibration signal touches authentication method, characterized in that the base
In the neural network of the twin network architecture, including two parallel sub-networks, each sub-network structure having the same and weight
And respectively according to the characteristic value X extracted in the vibration signal of input1,X2Corresponding character representation C is obtained after sub-networkW
(X1),CW(X2), and then obtain the distance of two character representations: DW(X1,X2)=| | Cw(X1)-Cw(X2)||。
6. the intelligent terminal according to claim 5 based on vibration signal touches authentication method, characterized in that the son
Network is the structure of time-delay neural network, includes two convolutional layers and a full articulamentum, is one basic in each convolutional layer
For convolution module as core, one BN layers are used as activation primitive for handling gradient problem and one ReLU layers;
The training sample of the neural network has same label, that is, comes from the same user and touch behavioural characteristic not phase
Seemingly, i.e. touching behavioural characteristic similarity is lower than threshold value.
7. the intelligent terminal according to claim 6 based on vibration signal touches authentication method, characterized in that the touching
Behavioural characteristic similarity is touched to be measured by Pearson correlation coefficient:WhenGreatly
In preset threshold h, then it is assumed that two samples have similar touching behavioural characteristic, otherwise it is assumed that the touching behavioural characteristic of the two is not
It is similar.
8. the intelligent terminal according to claim 6 based on vibration signal touches authentication method, characterized in that point
Class refers to: according to the distance for the corresponding character representation of two inputs that neural network obtains, judging whether two inputs belong to system
One user, so realize touching behavior whether from registered users judgement.
9. a kind of system for realizing any of the above-described claim the method characterized by comprising vibration signal processing mould
Block, signal characteristic abstraction module, the unrelated classifier of behavior and authentication module, in which: vibration signal processing module and signal are special
Sign extraction module is connected and is transferred through pretreated vibration signal information, signal characteristic abstraction module classifier unrelated with behavior
It is connected and transmits from the signal characteristic and behavior characteristic information by being extracted in pretreated vibration signal, the unrelated classifier of behavior
It is connected with authentication module and is transferred through trained sorter model information, it is special that authentication module receives the signal that authentication phase is extracted
Sign, which is connected and is transmitted to trained sorter model, obtains the authentication result of registrant.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910583135.8A CN110363120B (en) | 2019-07-01 | 2019-07-01 | Intelligent terminal touch authentication method and system based on vibration signal |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910583135.8A CN110363120B (en) | 2019-07-01 | 2019-07-01 | Intelligent terminal touch authentication method and system based on vibration signal |
Publications (2)
Publication Number | Publication Date |
---|---|
CN110363120A true CN110363120A (en) | 2019-10-22 |
CN110363120B CN110363120B (en) | 2020-07-10 |
Family
ID=68217593
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201910583135.8A Active CN110363120B (en) | 2019-07-01 | 2019-07-01 | Intelligent terminal touch authentication method and system based on vibration signal |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN110363120B (en) |
Cited By (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN112214635A (en) * | 2020-10-23 | 2021-01-12 | 昆明理工大学 | Fast audio retrieval method based on cepstrum analysis |
CN112949403A (en) * | 2021-02-01 | 2021-06-11 | 浙江大学 | Reliable user authentication method and system based on biological characteristics of mandible |
CN112989888A (en) * | 2019-12-17 | 2021-06-18 | 华为技术有限公司 | Fingerprint anti-counterfeiting method and electronic equipment |
CN113324648A (en) * | 2021-07-09 | 2021-08-31 | 中国铁道科学研究院集团有限公司铁道科学技术研究发展中心 | Portable high-speed railway wheel rail vibration space-time synchronization test method and system |
TWI749875B (en) * | 2020-02-14 | 2021-12-11 | 聯發科技股份有限公司 | Apparatuses and methods for providing a virtual input key |
CN116127365A (en) * | 2023-04-14 | 2023-05-16 | 山东大学 | Authentication method based on vibration and applied to intelligent head-mounted equipment |
Citations (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20060013445A1 (en) * | 2002-07-29 | 2006-01-19 | Lange Daniel H | Method and apparatus for electro-biometric identiy recognition |
KR101281852B1 (en) * | 2012-06-28 | 2013-07-03 | 고려대학교 산학협력단 | Biometric authentication device and method using brain signal |
CN104765455A (en) * | 2015-04-07 | 2015-07-08 | 中国海洋大学 | Man-machine interactive system based on striking vibration |
CN105426713A (en) * | 2014-09-11 | 2016-03-23 | 齐科斯欧公司 | Method and apparatus for differentiating touch screen users based on touch event analysis |
CN105808996A (en) * | 2016-05-12 | 2016-07-27 | 北京小米移动软件有限公司 | Method and device for unlocking screen of terminal |
CN106874833A (en) * | 2016-12-26 | 2017-06-20 | 中国船舶重工集团公司第七0研究所 | A kind of mode identification method of vibration event |
CN108681709A (en) * | 2018-05-16 | 2018-10-19 | 深圳大学 | Intelligent input method and system based on osteoacusis vibration and machine learning |
-
2019
- 2019-07-01 CN CN201910583135.8A patent/CN110363120B/en active Active
Patent Citations (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20060013445A1 (en) * | 2002-07-29 | 2006-01-19 | Lange Daniel H | Method and apparatus for electro-biometric identiy recognition |
KR101281852B1 (en) * | 2012-06-28 | 2013-07-03 | 고려대학교 산학협력단 | Biometric authentication device and method using brain signal |
CN105426713A (en) * | 2014-09-11 | 2016-03-23 | 齐科斯欧公司 | Method and apparatus for differentiating touch screen users based on touch event analysis |
CN104765455A (en) * | 2015-04-07 | 2015-07-08 | 中国海洋大学 | Man-machine interactive system based on striking vibration |
CN105808996A (en) * | 2016-05-12 | 2016-07-27 | 北京小米移动软件有限公司 | Method and device for unlocking screen of terminal |
CN106874833A (en) * | 2016-12-26 | 2017-06-20 | 中国船舶重工集团公司第七0研究所 | A kind of mode identification method of vibration event |
CN108681709A (en) * | 2018-05-16 | 2018-10-19 | 深圳大学 | Intelligent input method and system based on osteoacusis vibration and machine learning |
Cited By (11)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN112989888A (en) * | 2019-12-17 | 2021-06-18 | 华为技术有限公司 | Fingerprint anti-counterfeiting method and electronic equipment |
WO2021121112A1 (en) * | 2019-12-17 | 2021-06-24 | 华为技术有限公司 | Fingerprint anti-counterfeiting method and electronic device |
US11881011B2 (en) | 2019-12-17 | 2024-01-23 | Huawei Technologies Co., Ltd. | Fingerprint anti-counterfeiting method and electronic device |
CN112989888B (en) * | 2019-12-17 | 2024-06-07 | 华为技术有限公司 | Fingerprint anti-counterfeiting method and electronic equipment |
TWI749875B (en) * | 2020-02-14 | 2021-12-11 | 聯發科技股份有限公司 | Apparatuses and methods for providing a virtual input key |
US11291910B2 (en) | 2020-02-14 | 2022-04-05 | Mediatek Inc. | Apparatuses and methods for providing a virtual input key |
CN112214635A (en) * | 2020-10-23 | 2021-01-12 | 昆明理工大学 | Fast audio retrieval method based on cepstrum analysis |
CN112949403A (en) * | 2021-02-01 | 2021-06-11 | 浙江大学 | Reliable user authentication method and system based on biological characteristics of mandible |
CN112949403B (en) * | 2021-02-01 | 2022-08-23 | 浙江大学 | Reliable user authentication method and system based on biological characteristics of mandible |
CN113324648A (en) * | 2021-07-09 | 2021-08-31 | 中国铁道科学研究院集团有限公司铁道科学技术研究发展中心 | Portable high-speed railway wheel rail vibration space-time synchronization test method and system |
CN116127365A (en) * | 2023-04-14 | 2023-05-16 | 山东大学 | Authentication method based on vibration and applied to intelligent head-mounted equipment |
Also Published As
Publication number | Publication date |
---|---|
CN110363120B (en) | 2020-07-10 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN110363120A (en) | Intelligent terminal based on vibration signal touches authentication method and system | |
Zhang et al. | Wi-Fi sensing for joint gesture recognition and human identification from few samples in human-computer interaction | |
CN1972186B (en) | A mobile identity authentication system and its authentication method | |
CN105429969B (en) | A kind of user ID authentication method and equipment | |
AU2016200905A1 (en) | A system and method for identifying and analyzing personal context of a user | |
CN102413100A (en) | Voice-print authentication system having voice-print password picture prompting function and realization method thereof | |
Wang et al. | Deep neural networks for CSI-based authentication | |
CN105574496A (en) | Face talkback recognition system | |
CN103106717A (en) | Intelligent warehouse voice control doorkeeper system based on voiceprint recognition and identity authentication method thereof | |
CN101540170B (en) | Voiceprint recognition method based on biomimetic pattern recognition | |
CN110929242B (en) | Method and system for carrying out attitude-independent continuous user authentication based on wireless signals | |
CN107784215B (en) | Audio unit based on intelligent terminal carries out the user authen method and system of labiomaney | |
Halevi et al. | Context-aware defenses to RFID unauthorized reading and relay attacks | |
CN108182418A (en) | A kind of thump recognition methods based on multidimensional acoustic characteristic | |
CN213601611U (en) | Law enforcement appearance with voiceprint recognition function | |
CN113870862A (en) | Voiceprint recognition model training method, voiceprint recognition method and related equipment | |
CN110621038B (en) | Method and device for realizing multi-user identity recognition based on WiFi signal detection gait | |
CN107248410A (en) | The method that Application on Voiceprint Recognition dustbin opens the door | |
CN106792512A (en) | A kind of infrared identification alignment system and method for being applied to mobile terminal | |
CN111833907A (en) | Man-machine interaction method, terminal and computer readable storage medium | |
CN110415708A (en) | Method for identifying speaker, device, equipment and storage medium neural network based | |
Shah et al. | Echo-ID: Smart user identification leveraging inaudible sound signals | |
WO2021103913A1 (en) | Voice anti-counterfeiting method and apparatus, terminal device, and storage medium | |
CN114003883A (en) | Portable digital identity authentication equipment and identity authentication method | |
CN206209623U (en) | A kind of wireless security mouse system based on micro-acceleration sensor |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |