CN103442114A - Identity authentication method based on dynamic gesture - Google Patents
Identity authentication method based on dynamic gesture Download PDFInfo
- Publication number
- CN103442114A CN103442114A CN2013103589687A CN201310358968A CN103442114A CN 103442114 A CN103442114 A CN 103442114A CN 2013103589687 A CN2013103589687 A CN 2013103589687A CN 201310358968 A CN201310358968 A CN 201310358968A CN 103442114 A CN103442114 A CN 103442114A
- Authority
- CN
- China
- Prior art keywords
- gesture
- value
- frame
- authentication
- dtw
- 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
Images
Abstract
The invention discloses an identity authentication method based on a dynamic gesture. According to the method, an intelligent cell phone acceleration sensor is utilized to obtain dynamic information when the gesture is executed, and matching authentication is conducted on the dynamic information of the gesture through a DTW efficient method combining relaxation of end point limitation and early termination. By means of the DTW efficient method combining relaxation of end point limitation and early termination, the problem of authentication failure caused by unaligned end points between gesture sequences is solved by relaxing end point limitation of a matched path, meanwhile calculating amount is reduced by utilizing bending slope limitation and an early termination strategy, and a test shows that the method has a good result in the aspects of precision and efficiency of identity authentication.
Description
Technical field
The invention belongs to pattern recognition and identity identifying technology field, relate to a kind of dynamic gesture identity identifying method based on the android platform.
Background technology
Authentication is that system validation operator's true identity claims with it process whether identity conforms to, and universal today at mobile phone, the authenticating user identification on mobile phone also becomes a pith of information security.Handset identities authentication at present mainly is divided into authentication and the authentication based on biological characteristic of password-based.The authentication of password-based commonly used has user's pin mode and nine grids unlocking manner, and the common feature of the authentication of password-based is that password is easily revealed, and for the fail safe of password, frequently changes password, makes again password be difficult to safeguard.Authentication based on biological characteristic can be used as a good alternative method of user password, and biological characteristic is people's natural attribute, comprises people's physiological characteristic or behavioural characteristic.Physiological characteristic is inborn feature, comprises the static natures such as face phase, fingerprint, palm shape, sound, iris, retina; Behavioural characteristic forms by posteriori study or development, comprises the behavioral characteristics such as signature, keystroke, gait, dynamic gesture.Biological characteristic can not guess and forget by easy quilt as password, can be as having yet easy being stolen, so, utilize biological character for identity authentication will be a kind of safer reliably, popular authentication means conveniently.
Identity identifying technology based on biological characteristic commonly used comprises following several at present:
1. finger print identifying
Finger print identifying is a kind of biometric identity authentication techniques the most ancient and commonly used, occupies in biological characteristic authentication market and surpasses the share of half.Fingerprint is the lines on people's finger tips surface, the minutias such as abundant breakpoint, crosspoint, binding site have been comprised in these rough skin lines, these features are unique, are also constancies, can determine a people's identity by the comparison of fingerprint.Finger print identifying is exactly to utilize image processing techniques to be mated the fingerprint gathered, thereby differentiates user's identity.
2. iris authentication
Iris authentication is most convenient, the most a kind of in current all biological characteristic authentication technology, is also the biological identification technology of tool development prospect of 21st century.Iris is annular section between sclera and pupil, and it comprises abundant textural characteristics, and structure is random, is that gene determine, is difficult for being forged.Contactless iris image acquisition health is easy-to-use, is not subject to the unexpected environmental impact of light while obtaining, and stability is high.
3. face authentication
Face authentication is one of research topic that the biological characteristic authentication technical field is the most difficult, the extraction of face characteristic is more difficult, the expression that same people is different, position, direction, illumination all can produce larger impact to the extraction of face characteristic, so the accuracy of face authentication is lower than finger print identifying and iris authentication at present, but contactless face characteristic information is obtained relatively nature and is difficult for discovering, good user experiences and makes face authentication become the easiest received biological characteristic authentication mode.
4. signature authentication
Signature authentication is a kind of behavioural characteristic authentication techniques, and signature authentication is divided into static signature authentication and on-line signature authentication according to data acquisition mode difference.Static signature authentication is to be the character conversion on paper the accessible image of computer by scanner, and extracts the feature such as texture information and authenticated.On-line signature authentication gathers user's written information by special-purpose board, signature sequence is converted into to image, and the information such as the pressure write of record, acceleration, speed, according to user's writing style, the user is authenticated.
5. vena identification
Vena identification is a kind of biological characteristic authentication technology people's identity authenticated by the vein distribution patterns on people's finger, the back of the hand, palm.Vein pattern is a kind of physiological characteristic, and different people's vein pattern is all different, even the right-hand man's of same person vein pattern is all different, is difficult to forge, and the very high contactless data acquisition of fail safe, also make the user be very easy to accept.Vena identification utilizes infrared C CD camera to obtain the vein image data, uses that binaryzation of surprise attack, refinement means to extract feature to digital picture, then with main frame in the vein pattern stored mated, thereby reach the effect of authentication.
The biological characteristic authentication mode has solved the various limitation of traditional password authentication mode, but on intelligent mobile phone platform, the use amount of biological characteristic authentication mode but can not show a candle to the password authentication mode, and main cause has following 2 points: be at first because resource, the device-restrictive on cell phone platform.Basically all there is no to obtain the equipment of fingerprint on mobile phone at present, use the words of finger print identifying just to need external equipment, use inconvenient; And iris authentication is very high to the camera requirement, the camera lowest price is 7000 dollars like this, and mobile phone also is difficult to realization, and the on-line signature authentication also needs external equipment; The collecting device of vena identification also has specific (special) requirements in addition, and design is complicated, and manufacturing cost is high, and product is difficult to miniaturization, portable inapplicable.Next is some drawbacks limit of authentication mode itself.Iris authentication mode extremely difficulty reads the Black Eyes feature, and face authentication and sound authentication all are very easy to be subject to the extraneous even impact of self, and the static signature authentication easily is stolen and shifts.
Therefore, be necessary to design a kind of novel identity identifying method.
Summary of the invention
Technical problem to be solved by this invention is to provide a kind of identity identifying method based on dynamic gesture, the method adopts in conjunction with the DTW high efficiency method that relaxes end points restriction and premature termination when the coupling authentication, on DTW method basis, limit the bending slope, relax the restriction of coupling path end points and in conjunction with the premature termination strategy, computational efficiency and the gesture coupling authentication precision of DTW method can be effectively improved, desirable authentication effect can be obtained.
The technical solution of invention is as follows:
A kind of identity identifying method based on dynamic gesture comprises the following steps:
X while 1) gathering the gesture execution, y, the acceleration information of tri-directions of z is as test sample book;
2) test data is carried out to preliminary treatment; Described preliminary treatment comprises smoothing denoising and quantification;
3) adopt improved dynamic time warping (DTW) algorithm to be mated test sample book and masterplate data; [the DTW algorithm, based on the thoery of dynamic programming, bends the time shaft of test pattern unevenly, until the test data feature is alignd with template characteristic.Because smart mobile phone is having some time delays when transducer obtains acceleration information, head and the afterbody of the gesture data therefore got may comprise some insignificant data, for these nonsignificant datas of place to go to avoid larger systematic error, just do these improvement] cancel DTW method the 2nd) restriction of aliging of end points while mating between the test sample book that obtains of step and optimum template, allow the starting point in Dynamic Programming coupling path at line segment [(1, 1), (1, or [(1 L)], 1), (L, 1)], and allow terminal at line segment [((M-L+1), N), (M, or [(M N)], (N-L+1)), (M, N)] on, that is to say certain gesture the first frame can with the front L frame of another gesture in any frame mated, last frame can with another gesture end L frame in any frame mated, and the slope of restriction Dynamic Programming bending is between 0.5-2, the amount L that utilizes end points to relax and the slope meter of bending are calculated the boundary condition of Dynamic Programming,
Wherein, M is optimum template length, and N is test data length, and L is the amount that end points relaxes;
4) the 3rd) carry out Dynamic Programming within the boundary condition that obtains of step, do not need to preserve all Cumulative Distances and frame matching distance, statement column vector D preserves the Cumulative Distance of previous column, and statement column vector d preserves the Cumulative Distance of current column count;
5) calculate in Dynamic Programming and judge whether the Cumulative Distance when prostatitis all is greater than the value of cutting off from τ, if all be greater than the value of cutting off from τ authentification failure, premature termination; If also not all be greater than the value of cutting off from τ, proceed to step 6);
6) judge whether current data frame is the last frame of test data, if not last frame, the next frame assignment to current data frame, and proceed to step 5); If last frame is got D[M-L+1 ..., M] in minimum value min, and with the value of cutting off from τ relatively, if min is less than the value of cutting off from, authentication success, ending method; If min is greater than the value of cutting off from τ, authentification failure, finish this verification process.
Definite method of described optimum template and the value of cutting off from is:
For the user gathers the sample of 15 certain gestures, the efficient matching process of DTW that the end points restriction is loosened in employing calculates the DTW Cumulative Distance between sample in twos, selecting with other sample distances and minimum sample is that optimum sample is optimum template, and to select the ultimate range between optimum sample and other samples be the value of cutting off from.
In order to adapt to when people's different times is carried out gesture the variation that may occur, adopt the template adaptive strategy, again choose at set intervals new optimum template and replace original optimum template from the gesture by authentication.
In step 1) in, utilize (SensorManager) context.getSystemService (context.SENSOR_SERVICE) to obtain packaged sensor management object SensorManager of Android system, then by SensorManager, obtain acceleration transducer Sensor.TYPE_ACCELEROMETER; The set of frequency that transducer obtains data is SENSOR_DELAY_GAME;
In step 2) in, adopt simple Moving Average (SMA) filter to carry out smoothing denoising to test data; Computing formula is: the value after denoising
x wherein
ibe i test data; N gets a value in 5-10;
In step 2) in, the acceleration information of the floating type that collects is converted to the integer data of 33 grades;
In step 3) in, match time axle be divided into 5 sections, (1, L), (L+1, X
a), (X
a+ 1, X
b), (X
b+ 1, N-L), (N-L+1, N), wherein X
a=(2M-N-L)/3, X
b=(2 (2N-M)+L)/3 are all got the most close integer, and must be met X
a>=1, X
b>=L show that the pass of optimum template length M and test data length N is thus:
The test data that does not meet the above formula relation is considered to differ too large with optimum template, can't carry out dynamic bending, L value 8; Bending is divided into 5 sections, the y boundary value is expressed as follows by x:
When the every column data of test gesture is carried out to Dynamic Programming, calculate its corresponding boundary value, and only mate the lattice point in border, thereby reduce amount of calculation; The coupling of each row lattice point is calculated 3 lattice points (so do not need to preserve distance matrix and the Cumulative Distance matrix mated between all frames when method realizes, reducing memory space) of only having used previous column; L value 8.[value of M and N depends on the duration that gathers gesture data.After the user clicks " beginning gesture " button, smart mobile phone can gather with fixed frequency the data of acceleration transducer, after the user clicks " completing gesture " button, finishes image data.The data acquisition number of times carried out in this process is exactly the value of M or N.】
The present invention can utilize existing equipment on mobile phone, facilitates easy-to-usely, and can effectively to the user, be authenticated again.Authentication based on dynamic gesture is the principle according to the on-line signature authentication, utilize the acceleration transducer of intelligent mobile phone platform, acceleration information when obtaining the user and carrying out gesture is as feature, thereby the gesture sequence that prestores and gesture sequence to be certified are mated to the effect that reaches authentication.Matching process adopts dynamic time bending (DTW) principle, purpose be solve the gesture sequence in time with space on inconsistent problem, while is relaxed the end points restriction in coupling path on the basis of the efficient matching process of DTW, once, and Cumulative Distance exceeds the value of cutting off from premature termination coupling, core of the present invention is based on the efficient identity identifying method of DTW that end points restriction and premature termination are relaxed in this combination.
Design of the present invention is:
The present invention proposes a kind of gesture authentication method based on the smart mobile phone sensor device, comprising: at first gather the data sample of the some gestures of user, this data sample is carried out to preliminary treatment; Then limit the bending slope on the basis of DTW method, relax the end points restriction, calculate the boundary condition of the required lattice point of asking Euclidean distance of coupling between this data sample and optimum template, best accumulated distance between test data sample and optimum template is asked in last Dynamic Programming, and employing premature termination method, once Cumulative Distance surpasses the value of cutting off from, gesture authentification failure, if the Cumulative Distance of final optimal path is less than the value of cutting off from, gesture authentication success.
Beneficial effect:
The present invention proposes a kind of identity identifying method based on dynamic gesture.Multidate information when the method utilizes the smart mobile phone acceleration transducer to obtain the gesture execution, adopt in conjunction with the DTW high efficiency method that relaxes end points restriction and premature termination the gesture multidate information mated to authentication.This matching authentication method is a kind ofly will improve the method that authentication precision and authentication efficiency combine, relax the end points restriction and cancelled the end points alignment restriction in the DTW method, allow the starting point of dynamic programming path at line segment [(1, 1), (1, or [(1 L)], 1), (L, 1)], and terminal can be at line segment [((M-L+1), N), (M, or [(M N)], (N-L+1)), (M, N)] on, according to the slope of dynamic programming path bending, require (0.5-2) can calculate the boundary condition of dynamic programming path again, carry out Dynamic Programming within boundary condition, once certain is listed as all DTW Cumulative Distances and all is greater than the value of cutting off from, premature termination.Relaxing the restriction of coupling path end points in conjunction with the DTW high efficiency method utilization of relaxing end points restriction and premature termination has solved between the gesture sequence and has not lined up because of end points the authentification failure problem caused, also utilize bending Slope restrictions and premature termination strategy to reduce amount of calculation, experiment shows that this method is having result preferably aspect the precision of authentication and efficiency simultaneously.
The present invention proposes a kind of gesture authentication method based on the smart mobile phone sensor device, the method is on the basis of DTW method, limit the bending slope, relaxed the end points alignment restriction of two matching sequences, once while mating, the DTW Cumulative Distance has exceeded the value of cutting off from of authentication, matching process premature termination simultaneously.The optimum template strategy of employing in the gesture verification process, make cycle tests not need and all templates compare, thereby reduced amount of calculation, the ultimate range between other templates that the value of cutting off from during authentication is optimum template and same people's same gesture.Utilize original DTW method, existing DTW high efficiency method (FastDTW) and this method (EIDTW) to comprise " a " and " in " 3000 gesture samples of two kinds of gestures are authenticated, the FAR of three kinds of method validations of the present invention is 0, the FRR of original DTW method is 8.6%, the FRR of FastDTW method is 5.8%, and the FRR of this method authentication is 2.1%.From Fig. 7 and Fig. 8, also can find out in addition, the authentication efficiency of this method is 2-3 times of original DTW, single gesture is authenticated to one and need the 5-10 millisecond, and the authentication real-time is very high.
The accompanying drawing explanation
Fig. 1 is flow chart of the present invention.
Fig. 2 is classical DTW Method And Principle figure.The schematic diagram of DTW method Dynamic Programming search optimal path, in order to make to mate path inclination within reason, carry out local restriction to the coupling path, constraints is as shown in right frame figure in this figure, black color dots refers to current lattice point, and three white points position that the previous lattice point in footpath may occur that shows the way, arrive current lattice point (n, m) previous lattice point before must be (n-1, m), (n-1, m-1), in (n-1, m-2) three lattice points one of the Cumulative Distance minimum.
Fig. 3 is that existing DTW high efficiency method bends by restriction the Dynamic Programming matching area that slope calculates.
Fig. 4 is the schematic diagram that this method dynamic programming path end points loosens.Cancel the restriction of Dynamic Programming coupling path end points alignment, allowed the starting point of dynamic programming path in line segment [(1,1), (1, L)] or [(1,1), (L, 1)] upper, and terminal can be at line segment [(M-L+1, N), (M, N)] or [(M, N-L+1), (M, N)] on.
Fig. 5 is after this method is relaxed end points and limited and limit Dynamic Programming bending slope and be 0.5-2, the Dynamic Programming matching area calculated, and the solid line hexagon is Dynamic Programming matching area border.
Fig. 6 is this method premature termination schematic diagram.After meaning two gesture sequence Dynamic Programming coupling parts, the Cumulative Distance in coupling path has just surpassed the authentication value of cutting off from, then the calculating of terminator sequence further part Euclidean distance, the method premature termination.
Fig. 7 is test set while being positive sample, and the result of the present invention and original methods experiment relatively.Test set is my 60 real gestures.(a) figure is the figure as a result that adopts original DTW matching process to be authenticated: 4 samples of False Rejects in 60 samples, 1706ms consuming time; (b) be the figure as a result that adopts existing efficient DTW method to be authenticated: 2 samples of False Rejects in 60 samples, 438ms consuming time; (c) be the figure as a result that adopts the present invention to be authenticated: False Rejects 0 sample in 60 samples, 653ms consuming time;
Fig. 8 is test set while being negative sample, and the result of the present invention and original methods experiment relatively.Test set is the gesture of 33 other people imitations.(a) figure is the figure as a result that adopts original DTW matching process to be authenticated: in 33 samples, wrong acceptance is 0,941ms consuming time; (b) be the figure as a result that adopts existing efficient DTW method to be authenticated: 33 sample mistakes are accepted 0,316ms consuming time; (c) be the figure as a result that adopts the present invention to be authenticated: 33 sample mistakes are accepted 0,190ms consuming time;
Embodiment
Below with reference to the drawings and specific embodiments, the present invention is described in further details:
Identity identifying method based on dynamic gesture of the present invention has following characteristics:
1. utilize the mobile phone acceleration transducer to obtain the acceleration information of gesture to be certified; For the restriction of intelligent mobile phone battery flying power, the beginning of gesture and end are controlled with button respectively, only when needs obtain data, the mobile phone transducer are monitored, and consume as little as possible the battery of mobile phone resource, also reduce unnecessary amount of calculation;
2. gesture data to be certified is carried out to the preliminary treatment such as denoising, quantification; Be subject to the impact of shake and sensor accuracy in data acquisition, unavoidably have noise jamming, in addition, initial data by the acceleration transducer past is all floating type, floating type data computation complexity is large, lose time, and to use the integer data be the same with the effect that use floating type data reach gesture when authentication, so need be to data denoising, quantification;
3. with in conjunction with relaxing the restriction and the efficient matching process calculating test sample book of DTW of premature termination and the DTW distance between optimum template, this distance is compared with the value of cutting off from, be less than or equal to the value of cutting off from and authenticate and pass through, otherwise authentication is not passed through.
In above-mentioned step, relax the restrictive condition that the end points restriction refers to cancel the alignment of DTW highly effective algorithm coupling path end points, as shown in Figure 4, allowing Dynamic Programming to mate the starting point in path can be in line segment [(1,1), (1, or [(1,1), (L L)], 1)] upper, and terminal also can be at line segment [(M-L+1, N), (M, N)] or [(M, N-L+1), (M, N)] on.That is to say certain gesture the first frame can with the front L frame of another gesture in any frame mated, last frame can with another gesture end L frame in any frame mated.When algorithm is realized, only need all for being set to 0, just can realizing the Euclidean distance of lattice point on starting point line segment and terminal line segment mating path end points and limit and loosen.The scope of relaxing according to path end points restriction and the slope requirement of bending, can calculate the zone boundary of the lattice point of the required compute euclidian distances of Dynamic Programming, as shown in Figure 5.
Once premature termination refers to the Cumulative Distance in DTW coupling path and surpasses the value of cutting off from, just stops calculating, as shown in Figure 6.On X-axis, often take a step forward, all only used the Cumulative Distance of previous column, column vector D and d preserve respectively the Cumulative Distance of previous column and the Cumulative Distance that current column count goes out, even if calculate the Euclidean distance of the required node of ability when the Cumulative Distance of prostatitis, the DTW Cumulative Distance is a kind of laddering calculating, once the pointwise of previous section sequence is to the computational process middle distance with over the value of cutting off from, without the sequence node-by-node algorithm continuing aft section, thereby the saving calculation procedure, improve computational efficiency.
10 samples that optimum template gathers when the user uses first, select, the DTW high efficiency method that the end points restriction is relaxed in utilization calculates the DTW distance between sample in twos, suppose gesture sample Normal Distribution, select with other samples apart from minimum sample as optimum template, distance maximum between this sample and other samples is the value of cutting off from.In order to adapt to when people's different times is carried out gesture the variation that may occur, adopt the template adaptive strategy, when the test specimens given figure of certain user's gesture has surpassed n (when this module realizes, the n value is 50), just again choose new optimum template from the gesture by authentication, the method for choosing is the same when choosing first.
Embodiment 1:
Idiographic flow as shown in Figure 1, is now introduced the details that realizes of each step.
1, utilize the acceleration transducer on smart mobile phone, obtain the user and carry out three directional accelerations (x, y, z) data sequence in the gesture process.At first utilize (SensorManager) context.getSystemService (context.SENSOR_SERVICE) to obtain packaged sensor management object SensorManager of Android system, then by SensorManager, obtain acceleration transducer Sensor.TYPE_ACCELEROMETER; The set of frequency that transducer obtains data is SENSOR_DELAY_GAME, adopt the words data redundancy amount of higher frequency SENSOR_DELAY_FASTEST too large, the waste processing time, while adopting more low frequency SENSOR_DELAY_UI and SENSOR_DELAY_NORMAL, the gesture acceleration information got again very little, the coupling that is unfavorable for gesture, experimental results show that to select SENSOR_DELAY_GAME proper.While gathering gesture data, need be monitored acceleration transducer, and the operation of monitoring acceleration transducer is the comparison power consumption, use millet 1 generation mobile phone to be tested the power consumption of this operation, close the application except system program during test, the result of test is: while not monitoring acceleration transducer, 1% electric weight can be used 16 minutes and 10 seconds, and while monitoring acceleration transducer, 1% electric weight can only be used 6 minutes and 35 seconds.Restriction for the intelligent mobile phone battery flying power, the beginning of gesture and end are controlled with button respectively, when starting, gesture monitors acceleration transducer, removing immediately acceleration transducer when gesture finishes monitors, so, only when needs obtain data, the mobile phone transducer is monitored, guaranteed battery life; Another benefit that adopts beginning and conclusion button is beginning and the end position that does not need to calculate according to the data of obtaining again gesture, thereby has reduced some unnecessary amounts of calculation.
2, test data is carried out to preliminary treatment, comprise denoising, quantification.Be subject to the impact of shake and sensor accuracy in data acquisition, unavoidably have noise jamming, adopt simple Moving Average (SMA) filter to carry out smoothing denoising to test data, elimination random noise on the basis of response fast, derivation formula is as follows:
SMA
now＝(X
i+X
i-1+...+X
i-n+1)/n n＝1，2，3，... (1)
N in formula (1) is the smoothing denoising parameter, and when the n value is too small, denoising effect is not obvious, and the excessive gesture information that can make again of n value is lost, and in experimentation, sums up and learns, during n value 5-10, proper.
The initial data that the mobile phone transducer obtains is all floating type, in view of the resource limitation of mobile phone, should reduce as much as possible amount of calculation, improves computational efficiency.So calculate for fear of floating type, the floating type data are converted to the shaping data of 33 grades, studied by the gesture data to a large amount of, found that the mainly concentrate on-g of value of gesture acceleration is between g, data are seldom arranged on 2g or-2g under, so to the initial data memory nonlinear quantization collected, be quantified as-16 below-2g ,-2g is quantified as-15 to-11 between-g, and-g is quantified as-10 to 10 between g, g is quantified as 11-15 between 2g, is quantified as 16 more than 2g.
3, the purpose of dynamic time warping DTW method is to find the coupling path of the time calibration of an optimization between reference template and test data.The method is applied in speech recognition the earliest, has solved the inhomogeneous difficult problem of speech rate.And also existing the inhomogeneous problem of speed at the gesture authenticated connection, the gesture of execution may be not of uniform size, so the DTW method also is applicable to the gesture authentication.As shown in Figure 2, R means optimum template, the sequential label that m is optimum template, and m=1 is start frame, m=M is terminal gesture frame; T means the gesture test data, the sequential label that n is optimum template, and n=1 is start frame, n=N is terminal gesture frame.The DTW method is exactly hunting time warping function j=w (i), makes the time shaft of test data non-linearly be mapped to optimum template time shaft j above, and meets:
D[T (i) in formula (2), R (w (i))] be the distance metric between cycle tests i frame T (i) and template sequence j frame template vector R (j), be Euclidean distance, the Euclidean distance of the required calculating of classical DTW method is all lattice points in rectangle frame.Dist is the Cumulative Distance between two data under optimal situation.Simultaneously, the DTW method has been added some restrictions for the coupling path, and the firstth, the Experience about Monotonicity of Functions restriction, because data have timing, so warping function must meet monotonicity restriction w (i+1) >=w (i); The secondth, the continuity restriction, some special direction vector plays critical effect to matching effect sometimes, in order to guarantee correctness, requires the Time alignment function not allow any one match point of effect.The 3rd is end points alignment restriction, in Fig. 2, can find out, during DTW method coupling, beginning end points and the end caps of test data and reference template all align respectively.
Existing DTW high efficiency method is that the slope of restriction bending edge circle on the basis of classical DTW method is 0.5-2, and as shown in Figure 3, high efficiency method only need calculate the Euclidean distance of the lattice point within parallelogram, and dynamic bending is divided into to 3 sections, (1, X
a), (X
a+ 1, X
b), (X
b+ 1, N).X wherein
a=(2M-N)/3, X
ball the most close integer is got in=2 (2N-M)/3, thereby the pass obtained between the most optimum template length M and test template length N is
The test data that does not meet formula 3 relations can be thought and differs too large with optimum template, can't carry out dynamic bending.
End points restriction is loosened in combination of the present invention and the DTW high efficiency method that finishes in advance loosens the end points restriction limiting on the basis that bending slope in coupling path is 0.5-2, and employing premature termination strategy in matching process.Loosening the end points restriction is exactly the end points alignment restriction of cancelling in the DTW algorithm.The starting point that allows dynamic programming path is upper at line segment [(1,1), (1, L)] or [(1,1), (L, 1)], and terminal can be on line segment [(M-L+1, N), (M, N)] or [(M, N-L+1), (M, N)].As shown in Figure 4, the DTW highly effective algorithm that loosens end points only need calculate the lattice point Euclidean distance of realizing in polygon, the amount of calculation of DTW highly effective algorithm is some more a little relatively, but can solve the start frame of gesture, the situation that end frame does not line up, and can improve the precision of authentication.Dynamic bending is divided into to 5 sections, (1, L), (L+1, X
a), (X
a+ 1, X
b), (X
b+ 1, N-L), (N-L+1, N), wherein X
a=(2M-N-L)/3, X
b=(2 (2N-M)+L)/3 are all got the most close integer, and must be met X
a>=1, X
b>=L show that the pass of optimum template length M and test data length N is thus:
The test data that does not meet formula 4 relations can be thought and differs too large with optimum template, can't carry out dynamic bending.The L value is too little, DeGrain, and the L value too conference percent of pass (FAR) that leads to errors increases, L value 8 in the present invention.Bending is divided into 5 sections, according to the x coordinate of point in the slope of every section up-and-down boundary line segment and section, can obtain the y boundary value of every section, the y boundary value is expressed as follows by x:
When the every column data of test gesture is carried out to Dynamic Programming, calculate its corresponding boundary value, and only mate the lattice point in border, thereby reduce amount of calculation.Because 3 lattice points of previous column have only been used in the coupling calculating of each row lattice point, so do not need to preserve distance matrix and the Cumulative Distance matrix mated between all frames when method realizes, reduce memory space.
In method Reusability to the optimum template utilization DTW high efficiency method that loosens the end points restriction selected.At first, the user gathers the data sample of 10 certain gestures, the DTW high efficiency method that the end points restriction is loosened in utilization calculates the DTW Cumulative Distance between data sample in twos, select one with other sample distances and minimum sample as optimum template, the ultimate range between this sample and other samples is the value of cutting off from.In order to adapt to when people's different times is carried out gesture the variation that may occur, adopt the template adaptive strategy, just from the gesture by authentication, again choose new optimum template at set intervals, the method for choosing is the same when choosing first.
4, on the basis of the DTW highly effective algorithm that loosens end points restriction again in conjunction with the premature termination strategy, premature termination is a kind of method of using in restricted distance calculates, as shown in Figure 6, once all DTW Cumulative Distances that are listed as have previously surpassed the value of cutting off from, without the Euclidean distance of asking again further part lattice point in border, thereby saved calculation procedure, the premature termination strategy can reduce a large amount of amounts of calculation when test gesture and optimum template differ larger.Dynamic Programming step based on premature termination is as follows:
State two column vector D, d, D preserves the Cumulative Distance of previous column, and d preserves the Cumulative Distance that current column count goes out.Before initialization test data the first frame and optimum template, the Cumulative Distance of L frame is 0, the second frame data that data i is test sample book;
5, calculate the Cumulative Distance d[of optimum template data frame within i and boundary condition according to the Dynamic Programming principle]; Then vector d assignment to vector D; Judge whether data all in vector D all are greater than the value of cutting off from, if all be greater than the value of cutting off from finish in advance coupling, authentification failure; If not all being greater than the value of cutting off from, enter step 6;
6, judge whether data i is the last frame of test data, if i is not last frame, the next frame assignment of i to i, proceed to step 5; If i is last frame, obtain D[M-L+ 1 ... M] between minimum value min, and compare with the value of cutting off from, if the value of cutting off from is greater than min authentication success, finish coupling; If the value of cutting off from is less than min authentification failure, finish coupling.
Claims (4)
1. the identity identifying method based on dynamic gesture, is characterized in that, comprises the following steps:
X while 1) gathering the gesture execution, y, the acceleration information of tri-directions of z is as test sample book;
2) test data is carried out to preliminary treatment; Described preliminary treatment comprises smoothing denoising and quantification;
3) adopt improved dynamic time warping (DTW) algorithm to be mated test sample book and masterplate data; Cancel DTW method the 2nd) restriction of aliging of end points between the test sample book that obtains of step and optimum template during coupling, the permission Dynamic Programming is mated the starting point in path in line segment [(1,1), (1, L)] or [(1,1), (L, 1)] upper, and allow terminal at line segment [((M-L+1), N), (M, N)] or [(M, (N-L+1)), (M, N)] on; That is to say certain gesture the first frame can with the front L frame of another gesture in any frame mated, last frame can with another gesture end L frame in any frame mated; And the slope of restriction Dynamic Programming bending is between 0.5-2, the amount L that utilizes end points to relax and the slope meter of bending are calculated the boundary condition of Dynamic Programming;
Wherein, M is optimum template length, and N is test data length, and L is the amount that end points relaxes;
4) the 3rd) carry out Dynamic Programming within the boundary condition that obtains of step, do not need to preserve all Cumulative Distances and frame matching distance, statement column vector D preserves the Cumulative Distance of previous column, and statement column vector d preserves the Cumulative Distance of current column count;
5) calculate in Dynamic Programming and judge whether the Cumulative Distance when prostatitis all is greater than the value of cutting off from τ, if all be greater than the value of cutting off from τ authentification failure, premature termination; If also not all be greater than the value of cutting off from τ, proceed to step 6);
6) judge whether current data frame is the last frame of test data, if not last frame, the next frame assignment to current data frame, and proceed to step 5); If last frame is got D[M-L+1 ..., M] in minimum value min, and with the value of cutting off from τ relatively, if min is less than the value of cutting off from, authentication success, ending method; If min is greater than the value of cutting off from τ, authentification failure, finish this verification process.
2. a kind of gesture authentication method based on the smart mobile phone sensor device according to claim 1, is characterized in that, definite method of described optimum template and the value of cutting off from is:
For the user gathers the sample of 15 certain gestures, the efficient matching process of DTW that the end points restriction is loosened in employing calculates the DTW Cumulative Distance between sample in twos, selecting with other sample distances and minimum sample is that optimum sample is optimum template, and to select the ultimate range between optimum sample and other samples be the value of cutting off from.
3. a kind of gesture authentication method based on the smart mobile phone sensor device according to claim 1, it is characterized in that, in order to adapt to when people's different times is carried out gesture the variation that may occur, adopt the template adaptive strategy, again choose at set intervals new optimum template and replace original optimum template from the gesture by authentication.
4. according to the described a kind of gesture authentication method based on the smart mobile phone sensor device of claim 1-3 any one, it is characterized in that:
In step 1) in, utilize (SensorManager) context.getSystemService (context.SENSOR_SERVICE) to obtain packaged sensor management object SensorManager of Android system, then by SensorManager, obtain acceleration transducer Sensor.TYPE_ACCELEROMETER; The set of frequency that transducer obtains data is SENSOR_DELAY_GAME;
In step 2) in, adopt simple Moving Average (SMA) filter to carry out smoothing denoising to test data; Computing formula is: the value after denoising
x wherein
ibe i test data; N gets a value in 5-10;
In step 2) in, the acceleration information of the floating type that collects is converted to the integer data of 33 grades;
In step 3) in, match time axle be divided into 5 sections, (1, L), (L+1, X
a), (X
a+ 1, X
b), (X
b+ 1, N-L), (N-L+1, N), wherein X
a=(2M-N-L)/3, X
b=(2 (2N-M)+L)/3 are all got the most close integer, and must be met X
a>=1, X
a>=L show that the pass of optimum template length M and test data length N is thus:
The test data that does not meet the above formula relation is considered to differ too large with optimum template, can't carry out dynamic bending, L value 8; Bending is divided into 5 sections, the y boundary value is expressed as follows by x:
When the every column data of test gesture is carried out to Dynamic Programming, calculate its corresponding boundary value, and only mate the lattice point in border, thereby reduce amount of calculation; The coupling of each row lattice point is calculated 3 lattice points only having used previous column; L value 8.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201310358968.7A CN103442114B (en) | 2013-08-16 | 2013-08-16 | A kind of identity identifying method based on dynamic gesture |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201310358968.7A CN103442114B (en) | 2013-08-16 | 2013-08-16 | A kind of identity identifying method based on dynamic gesture |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN103442114A true CN103442114A (en) | 2013-12-11 |
| CN103442114B CN103442114B (en) | 2015-10-21 |
Family
ID=49695778
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201310358968.7A Expired - Fee Related CN103442114B (en) | 2013-08-16 | 2013-08-16 | A kind of identity identifying method based on dynamic gesture |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN103442114B (en) |
Cited By (23)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103677642A (en) * | 2013-12-19 | 2014-03-26 | 深圳市汇顶科技股份有限公司 | Touch screen terminal and method and system for identifying hand gestures of touch screen terminal |
| CN104580143A (en) * | 2014-11-09 | 2015-04-29 | 李若斌 | Security authentication method based on gesture recognition, terminal, server and system |
| CN105024817A (en) * | 2014-04-29 | 2015-11-04 | 中国移动通信集团湖南有限公司 | Identity authentication method and identity authentication device |
| CN105100376A (en) * | 2014-05-16 | 2015-11-25 | 中国移动通信集团湖南有限公司 | Identity authentication method and apparatus |
| CN105207783A (en) * | 2015-10-26 | 2015-12-30 | 联动优势电子商务有限公司 | Authentication method and device for user's confidence level |
| CN105530095A (en) * | 2014-09-29 | 2016-04-27 | 中移电子商务有限公司 | Method and device for user identity authentication |
| CN105617638A (en) * | 2015-12-25 | 2016-06-01 | 深圳市酷浪云计算有限公司 | Badminton racket swinging movement recognizing method and device |
| CN105807912A (en) * | 2015-01-21 | 2016-07-27 | 现代自动车株式会社 | Vehicle, method for controlling the same and gesture recognition apparatus therein |
| CN106331363A (en) * | 2016-09-09 | 2017-01-11 | 常州大学 | Blind identification and guiding-type blind guiding method based on intelligent mobile phone |
| CN106411952A (en) * | 2016-12-01 | 2017-02-15 | 安徽工业大学 | Telekinetic-dynamic-gesture-based user identity authentication method and apparatus |
| CN106470192A (en) * | 2015-08-19 | 2017-03-01 | 阿里巴巴集团控股有限公司 | Auth method, apparatus and system |
| CN106548133A (en) * | 2016-10-17 | 2017-03-29 | 歌尔科技有限公司 | A kind of template matching method and device and gesture identification method and device |
| CN107018121A (en) * | 2016-10-13 | 2017-08-04 | 阿里巴巴集团控股有限公司 | The method and device of subscriber authentication |
| CN108038412A (en) * | 2017-10-30 | 2018-05-15 | 捷开通讯(深圳)有限公司 | Terminal and its control method based on self-training gesture, storage device |
| CN108319421A (en) * | 2018-01-29 | 2018-07-24 | 维沃移动通信有限公司 | A kind of display triggering method and mobile terminal |
| CN108335699A (en) * | 2018-01-18 | 2018-07-27 | 浙江大学 | A kind of method for recognizing sound-groove based on dynamic time warping and voice activity detection |
| CN108595937A (en) * | 2018-04-27 | 2018-09-28 | 中国人民解放军国防科技大学 | Micro-sensing intelligent identity authentication method based on behavior characteristics |
| CN109302532A (en) * | 2018-11-02 | 2019-02-01 | 北京航空航天大学 | A kind of identity identifying method based on smart phone acceleration transducer |
| CN110163130A (en) * | 2019-05-08 | 2019-08-23 | 清华大学 | A kind of random forest grader and classification method of the feature pre-align for gesture identification |
| CN110298159A (en) * | 2019-06-28 | 2019-10-01 | 闽江师范高等专科学校 | A kind of smart phone dynamic gesture identity identifying method |
| CN110647730A (en) * | 2018-06-26 | 2020-01-03 | 国际商业机器公司 | Single channel input multi-factor authentication via separate processing paths |
| WO2021051873A1 (en) * | 2019-09-18 | 2021-03-25 | 平安科技（深圳）有限公司 | Method and apparatus for preprocessing pulse wave based on similarity, and computer system and readable storage medium |
| CN112562657A (en) * | 2021-02-23 | 2021-03-26 | 成都启英泰伦科技有限公司 | Personalized language offline learning method based on deep neural network |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105912910A (en) * | 2016-04-21 | 2016-08-31 | 武汉理工大学 | Cellphone sensing based online signature identity authentication method and system |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103034429A (en) * | 2011-10-10 | 2013-04-10 | 北京千橡网景科技发展有限公司 | Identity authentication method and device for touch screen |
| WO2013055953A1 (en) * | 2011-10-12 | 2013-04-18 | Qualcomm Incorporated | Authenticated gesture recognition |
-
2013
- 2013-08-16 CN CN201310358968.7A patent/CN103442114B/en not_active Expired - Fee Related
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103034429A (en) * | 2011-10-10 | 2013-04-10 | 北京千橡网景科技发展有限公司 | Identity authentication method and device for touch screen |
| WO2013055953A1 (en) * | 2011-10-12 | 2013-04-18 | Qualcomm Incorporated | Authenticated gesture recognition |
Non-Patent Citations (2)
| Title |
|---|
| LIU J, ZHONG L, WICKRAMASURIYA J, ET AL: "uWave: Accelerometer-based personalized gesture recognition and its applications", 《PERVASIVE AND MOBILE COMPUTING》 * |
| 何超 等: "一种基于改进DTW算法的动态手势识别方法", 《数字通信》 * |
Cited By (39)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103677642A (en) * | 2013-12-19 | 2014-03-26 | 深圳市汇顶科技股份有限公司 | Touch screen terminal and method and system for identifying hand gestures of touch screen terminal |
| CN105024817B (en) * | 2014-04-29 | 2018-11-02 | 中国移动通信集团湖南有限公司 | A kind of identity identifying method and device |
| CN105024817A (en) * | 2014-04-29 | 2015-11-04 | 中国移动通信集团湖南有限公司 | Identity authentication method and identity authentication device |
| CN105100376A (en) * | 2014-05-16 | 2015-11-25 | 中国移动通信集团湖南有限公司 | Identity authentication method and apparatus |
| CN105530095A (en) * | 2014-09-29 | 2016-04-27 | 中移电子商务有限公司 | Method and device for user identity authentication |
| CN105530095B (en) * | 2014-09-29 | 2019-07-30 | 中移电子商务有限公司 | A kind of method and device of user identity authentication |
| CN104580143A (en) * | 2014-11-09 | 2015-04-29 | 李若斌 | Security authentication method based on gesture recognition, terminal, server and system |
| CN105807912B (en) * | 2015-01-21 | 2020-10-20 | 现代自动车株式会社 | Vehicle, method for controlling the same, and gesture recognition apparatus therein |
| CN105807912A (en) * | 2015-01-21 | 2016-07-27 | 现代自动车株式会社 | Vehicle, method for controlling the same and gesture recognition apparatus therein |
| CN111314299A (en) * | 2015-08-19 | 2020-06-19 | 阿里巴巴集团控股有限公司 | Identity verification method, device and system |
| CN106470192B (en) * | 2015-08-19 | 2019-12-10 | 阿里巴巴集团控股有限公司 | Identity verification method, device and system |
| CN106470192A (en) * | 2015-08-19 | 2017-03-01 | 阿里巴巴集团控股有限公司 | Auth method, apparatus and system |
| CN105207783B (en) * | 2015-10-26 | 2018-11-16 | 联动优势电子商务有限公司 | A kind of User reliability authentication method and device |
| CN105207783A (en) * | 2015-10-26 | 2015-12-30 | 联动优势电子商务有限公司 | Authentication method and device for user's confidence level |
| CN105617638A (en) * | 2015-12-25 | 2016-06-01 | 深圳市酷浪云计算有限公司 | Badminton racket swinging movement recognizing method and device |
| CN105617638B (en) * | 2015-12-25 | 2019-04-05 | 深圳市酷浪云计算有限公司 | Badminton racket swing action identification method and device |
| CN106331363B (en) * | 2016-09-09 | 2019-03-22 | 常州大学 | A method of blind person's identification and guiding guide based on smart phone |
| CN106331363A (en) * | 2016-09-09 | 2017-01-11 | 常州大学 | Blind identification and guiding-type blind guiding method based on intelligent mobile phone |
| CN107018121A (en) * | 2016-10-13 | 2017-08-04 | 阿里巴巴集团控股有限公司 | The method and device of subscriber authentication |
| CN107018121B (en) * | 2016-10-13 | 2021-07-20 | 创新先进技术有限公司 | User identity authentication method and device |
| CN106548133A (en) * | 2016-10-17 | 2017-03-29 | 歌尔科技有限公司 | A kind of template matching method and device and gesture identification method and device |
| CN106548133B (en) * | 2016-10-17 | 2019-04-23 | 歌尔科技有限公司 | A kind of template matching method and device and gesture identification method and device |
| CN106411952B (en) * | 2016-12-01 | 2019-07-26 | 安徽工业大学 | One kind is every lost motion state gesture user identity identifying method and device |
| CN106411952A (en) * | 2016-12-01 | 2017-02-15 | 安徽工业大学 | Telekinetic-dynamic-gesture-based user identity authentication method and apparatus |
| CN108038412A (en) * | 2017-10-30 | 2018-05-15 | 捷开通讯(深圳)有限公司 | Terminal and its control method based on self-training gesture, storage device |
| CN108038412B (en) * | 2017-10-30 | 2021-06-15 | 捷开通讯(深圳)有限公司 | Terminal and control method and storage device thereof based on self-training gesture |
| CN108335699A (en) * | 2018-01-18 | 2018-07-27 | 浙江大学 | A kind of method for recognizing sound-groove based on dynamic time warping and voice activity detection |
| CN108319421A (en) * | 2018-01-29 | 2018-07-24 | 维沃移动通信有限公司 | A kind of display triggering method and mobile terminal |
| CN108595937A (en) * | 2018-04-27 | 2018-09-28 | 中国人民解放军国防科技大学 | Micro-sensing intelligent identity authentication method based on behavior characteristics |
| CN110647730B (en) * | 2018-06-26 | 2022-04-22 | 国际商业机器公司 | Single channel input multi-factor authentication via separate processing paths |
| CN110647730A (en) * | 2018-06-26 | 2020-01-03 | 国际商业机器公司 | Single channel input multi-factor authentication via separate processing paths |
| CN109302532A (en) * | 2018-11-02 | 2019-02-01 | 北京航空航天大学 | A kind of identity identifying method based on smart phone acceleration transducer |
| CN109302532B (en) * | 2018-11-02 | 2020-02-07 | 北京航空航天大学 | Identity authentication method based on smart phone acceleration sensor |
| CN110163130B (en) * | 2019-05-08 | 2021-05-28 | 清华大学 | Feature pre-alignment random forest classification system and method for gesture recognition |
| CN110163130A (en) * | 2019-05-08 | 2019-08-23 | 清华大学 | A kind of random forest grader and classification method of the feature pre-align for gesture identification |
| CN110298159B (en) * | 2019-06-28 | 2021-04-20 | 闽江师范高等专科学校 | Smart phone dynamic gesture identity authentication method |
| CN110298159A (en) * | 2019-06-28 | 2019-10-01 | 闽江师范高等专科学校 | A kind of smart phone dynamic gesture identity identifying method |
| WO2021051873A1 (en) * | 2019-09-18 | 2021-03-25 | 平安科技（深圳）有限公司 | Method and apparatus for preprocessing pulse wave based on similarity, and computer system and readable storage medium |
| CN112562657A (en) * | 2021-02-23 | 2021-03-26 | 成都启英泰伦科技有限公司 | Personalized language offline learning method based on deep neural network |
Also Published As
| Publication number | Publication date |
|---|---|
| CN103442114B (en) | 2015-10-21 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN103442114B (en) | A kind of identity identifying method based on dynamic gesture | |
| CN100557624C (en) | Face identification method based on the multicomponent and multiple characteristics fusion | |
| Kataria et al. | A survey of automated biometric authentication techniques | |
| KR102501838B1 (en) | Authentication apparatus and method based on electrocardiographic signals | |
| KR20210047539A (en) | EMG-based user authentication device and authentication method | |
| WO2018152711A1 (en) | Electrocardiographic authentication-based door control system and authentication method therefor | |
| WO2017113407A1 (en) | Gesture recognition method and apparatus, and electronic device | |
| Kumar et al. | Analysis of various biometric techniques | |
| Chavan et al. | Fingerprint authentication using gabor filter based matching algorithm | |
| CN105184571A (en) | Payment authentication system based on combination of hand veins and multiple fingerprints | |
| Song et al. | TDS-net: Towards fast dynamic random hand gesture authentication via temporal difference symbiotic neural network | |
| CN105024817A (en) | Identity authentication method and identity authentication device | |
| Yadav et al. | Multimodal biometric authentication system: challenges and solutions | |
| CN108960003A (en) | Based on Gabor and the palm print characteristics of chaotic maps generate and authentication method | |
| Moganeshwaran et al. | Fingerprint-fingervein multimodal biometric authentication system in field programmable gate array | |
| CN103700151A (en) | Morning run check-in method | |
| Hashemi et al. | Biometric identification through hand geometry | |
| CN107958242A (en) | A kind of Palm Print Recognition System | |
| Nayak et al. | Multimodal biometric face and fingerprint recognition using neural network | |
| Yang | Fingerprint recognition system based on big data and multi-feature fusion | |
| CN102184416A (en) | Method and device for registering biometric sample | |
| JP2005275605A (en) | Personal identification device and method | |
| CN105574508A (en) | Palm print identification and unlocking system based on mobile terminal | |
| WO2021245932A1 (en) | Information processing device, information processing method, and recording medium | |
| Kita et al. | Personal Identification with Face and Voice Features Extracted through Kinect Sensor |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C14 | Grant of patent or utility model | ||
| GR01 | Patent grant | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20151021 Termination date: 20170816 |
|
| CF01 | Termination of patent right due to non-payment of annual fee |