CN103593673B - A kind of on-line trial authentication method based on dynamic threshold - Google Patents
A kind of on-line trial authentication method based on dynamic threshold Download PDFInfo
- Publication number
- CN103593673B CN103593673B CN201310521026.6A CN201310521026A CN103593673B CN 103593673 B CN103593673 B CN 103593673B CN 201310521026 A CN201310521026 A CN 201310521026A CN 103593673 B CN103593673 B CN 103593673B
- Authority
- CN
- China
- Prior art keywords
- signature
- dtw
- coordinate
- sample
- value
- 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.)
- Expired - Fee Related
Links
Abstract
A kind of on-line trial authentication method based on dynamic threshold of the present invention, belong to information security field, the coordinate of on-line signature, pressure, velocity information are carried out comprehensive modeling by the inventive method, and then consider the effect of each factor, more effectively improve the accuracy rate that signature judges；Signature be have employed the strategy that segmentation differentiates by the inventive method, and according to my writing style of signature, automatically selects segmentation method more stable in two kinds of partition strategies, such that it is able to improve the accuracy that in signature judge process, the similarity of signature judges；The threshold value that the inventive method uses in signature judge process to be measured is to be dynamically determined according to different signatures, be conducive to embodying the deviation characteristics of everyone signature, namely take into account the problem that the feature stability of everyone signature is different, thus obtaining everyone best comparison result.
Description
Technical field
The invention belongs to information security field, be specifically related to a kind of on-line trial authentication method based on dynamic threshold.
Background technology
In society, along with developing rapidly of computer technology and Internet technology, the traditional office of people and life style are created enormous impact by information revolution, and information security has become as one of focal issue that people face.The requirement of information security is seemed more prominent by the application such as ecommerce, online transaction, online payment, web document transfer, and authentication quickly and accurately increasingly becomes a kind of exigence.Identity authentication method can be divided into common several classes: Knowledge based engineering method, for instance system password, password etc.；Based on the method for keepsake, such as identity card, IC-card etc.；Based on the recognition methods of biological characteristic, such as fingerprint, iris, retina, handwritten signature etc..Wherein based in the recognition methods of biological characteristic, behavior characteristics such as the signature that the long-term life of people is formed identifies the identity of people, it is believed that be natural, the most credible a kind of identity identifying method.And in daily life, it is seen everywhere by the authentication mode signed, file, check, contract, agreement signature etc., signature authentication has the meaning of particular importance, in long-term social life, people have received signature as identity and a kind of mode showing oneself wish.
In today that Internet and information technology develop rapidly, the mode of traditionally on paper handwritten signature cannot meet the demand of information security.By electronic device terminal, gather the signed data of signer, then pass through data analysis technique and be identified and the handwritten signature verification technology of certification increasingly receives publicity.Handwritten signature verification include signature acquisition, pretreatment, feature extraction and compare, the process such as decision-making.Different according to the method that signature obtains, signature authentication can be divided into static signature certification and the big class of Signature Verification two.Static signature authentication method, is usually after the signature image pretreatment obtained, extracts the parameter attribute of some geometry aspects, then carries out the comparison signed；Signature Verification method, by information such as electronic device terminal, the signing messages in dynamic acquisition writing process, the image such as signed, the order of strokes observed in calligraphy, speed, pressure, then sets up signature authentication model, carries out the certification signed.Owing to the dynamic signature information of Signature Verification method collection containing abundant personal characteristics, it is difficult to imitate, can the identity of effective identification signature person, so increasingly becoming the main direction of studying of signature authentication.Namely on-line trial Authentication Research is a kind of Signature Verification method, and it is the research field of a multi-crossed disciplines, relates to the knowledge of the multi-door subjects such as computer, signal processing, pattern recognition, neuro physiology.
The process nature of on-line trial certification is the extraction to signed data feature, the process comparing and differentiating, it it is some characteristic information by extracting signed data to be measured, and these characteristic informations are compared with the signature in registered template base, if comparative result is in rational range of error, then think that this signature is real signature in person, otherwise it is assumed that be the signature forged.But existing on-line trial authentication techniques are still far from perfect, reliability and precision can't be satisfactory, reason is in that: 1) handwritten signature comprises abundant behavioral characteristics, the abundant information such as including font, the order of strokes observed in calligraphy, speed, pressure, existing a lot of Signature Authentication System is authenticated independently only with single or less several features, such as only relying on font style characteristic, such imitator, possibly through substantial amounts of imitation exercise, breaks through the feature in a certain respect of signature.2) owing to handwritten signature is biometric information, so the feature occurred in characteristic quantity depends on that concrete everyone, different people have its distinctive characteristic quantity.In existing on-line trial system, judge the true or false of handwritten signature, it is according to respectively writing the similarity system design between the characteristic quantity of signature and the true signature character amount of Sample Storehouse and determining, and in the process of similarity system design, it is all use fixed threshold, thus have ignored the deviation characteristics of everyone signature, namely have ignored the problem that the feature stability of everyone signature is different, thus everyone best comparison result cannot be obtained.
Summary of the invention
For the deficiencies in the prior art, the present invention proposes a kind of on-line trial authentication method based on dynamic threshold, to reach to improve the accuracy rate of signature judgement, improve the accuracy that in signature judge process, the similarity of signature judges and the purpose of the best comparison result obtaining everyone.
A kind of on-line trial authentication method based on dynamic threshold, comprises the following steps:
Step 1, gather my multiple signature sample data, namely the movement locus of lettering pen is carried out constant duration sampling, it is thus achieved that the X-coordinate value of sampled point, Y-coordinate value and the contact force value to clipboard；And all sampling numbers evidence of signature is carried out validity check, if signed data is qualified, then allow the submission of signed data, perform step 2；Otherwise return and perform step 1；
Step 2, signature sample choose the foundation with template base；
Requirement chosen by the sample whether quantity of the signature sample that step 2-1, judgement gather reaches to set, and if so, performs step 2-2；Otherwise, execution step 1 is returned；
Step 2-2, rising of signature is started to write a little and the extreme point of coordinate is as waypoint, maybe signature acted point of starting to write, the angular velocity extreme point of signature and the extreme point of linear velocity of signing as waypoint, it is achieved the segment processing to signed data；
Step 2-3, employing dynamic time programming DTW method judge two degree of similarity signed, and obtain the final sample of template base；
Step 2-3-1, calculate all signatures coordinate between any two overall DTW value respectively, DTW value, pressure segmentation key point DTW value, the overall DTW value of speed, speed segmentation and DTW value and speed waypoint DTW value are sued for peace in coordinate fragments summation DTW value, coordinate fragments key point DTW value, pressure overall DTW value, pressure segmentation, and calculate above-mentioned DTW value average and deviation；
Step 2-3-2, delete the signature sample that in all samples, sampling number grid deviation is maximum and DTW value deviation is maximum；
Step 2-3-3, repeatedly execution step 1 are to step 2-3-3 until the volume residual of sample reaches the sample size retained needed for final sample storehouse；
Step 2-3-4, judge that whether residue sample meets signature stability condition, namely judge whether residue sample DTW deviation between any two is respectively less than the DTW threshold value that user sets, if so, then will remain the signature sample final sample as template base；Otherwise return and perform step 1；
Step 2-4, using the remaining signature sample of stablizing chosen as template base sample, calculation template storehouse sample DTW value between any two, and it is added up, determine the dynamic discriminant strategy of the residue stable signature DTW value deviation dynamic threshold of coordinate, the DTW value deviation dynamic threshold of pressure, the DTW value deviation dynamic threshold of speed and applicable current signature, and above-mentioned information is stored in template base；
Step 3, gather the signature sample data of tested user, namely the movement locus of lettering pen is carried out constant duration sampling, it is thus achieved that the X-coordinate value of sampled point, Y-coordinate value and the contact force value to clipboard；And to tested signature sampling number according to carrying out validity check, if signed data is qualified, then allow the submission of signed data, perform step 4；Otherwise return and perform step 3；
Step 4, judge the true and false of tested signature；
Step 4-1, the employing adding up tested signature are counted, the linear velocity extreme point quantity of the linear velocity extreme point type of the angular velocity extreme point quantity of the angular velocity extreme point type of the coordinate extreme point quantity of the coordinate extreme point type of the stroke count amount of rising and falling of stroke of signing, signature, signature, signature, signature, signature and signature；In conjunction with the dynamic threshold in template base, and based on above-mentioned information, signature to be measured is carried out dynamic filter, it is judged that whether tested signature meets dynamic filter condition, if so, perform step 4-2；Otherwise, then tested signature is pseudo-signature, terminates；
Step 4-2: select the dynamic discriminant strategy being suitable for current signature that tested signature is carried out segment processing；
Step 4-3: according to the coordinate of tested signed data, pressure and speed, calculates the coordinate of all signature sample data in itself and template base, DTW value between pressure and speed；
Step 4-4: according to the overall DTW value deviation of coordinate, coordinate fragments summation DTW value deviation, coordinate fragments key point DTW value deviation between the signature sample deposited in signed data to be measured and template base；The overall DTW value deviation of pressure between signature sample in signed data to be measured and template base, pressure segmentation summation DTW value deviation, pressure segmentation key point DTW value deviation, signed data and the overall DTW value deviation of speed of signature sample in template base, speed segmentation and DTW value deviation, speed waypoint DTW value deviation, respectively in connection with the dynamic threshold of signature sample coordinate DTW value deviation in template base, the dynamic threshold of pressure DTW value deviation and the dynamic threshold of speed DTW value deviation, the method adopting multi-expert ballot scoring judges the true and false of signature to be measured, even total more than DTW value less than the DTW value quantity of dynamic threshold half, then signature to be measured is judged as true signature, otherwise signature to be measured is judged as pseudo-signature, forbid that tested user logs in；
Step 4-5: differentiation result is carried out output display.
Effect described in step 1 and step 3 checks, the coordinate write including signing crosses the border inspection and signature reasonableness check, namely sign always count, whether the length and width scope of total writing time, signature reaches the threshold value that sets.
Determination described in step 2-4 and step 4-2 be suitable for current signature dynamic discriminant strategy, described dynamic discriminant strategy includes: using signature rise start to write a little and coordinate extreme point as waypoint carry out segmentation and using signature rise start to write point, signature angular velocity extreme point and sign linear velocity extreme point carry out segmentation as waypoint；Select segmentation method: select under this segmentation method, the signature sample method that DTW value deviation is minimum between any two.
Determination described in step 2-4 remains the DTW value deviation dynamic threshold of stable signature coordinate, particularly as follows: determine that residue stablizes sample coordinate DTW distance between any two, obtain whole coordinate DTW distance average, and according to each coordinate DTW distance and coordinate DTW distance average, obtain the deviation of DTW distance, select deviation maximum as the DTW value deviation dynamic threshold of coordinate；The DTW value deviation dynamic threshold of pressure is identical with the DTW value deviation dynamic threshold defining method of coordinate with the DTW value deviation dynamic threshold value determination method of speed.
The coordinate extreme point type of the signature described in step 4-1 includes X extreme point, Y extreme point, X and Y extreme point simultaneously；The angular velocity extreme point type of signature includes angular velocity minimum and angular velocity maximum.
Dynamic filter condition described in step 4-1 is: specimen sample count average ± 20% scope in；Stroke count amount of rising and falling average ± 2 stroke counts within the scope of；Signature coordinate extreme point quantity average ± 20% scope in；Signature angular velocity extreme point quantity average ± 20% scope in；Signature linear velocity extreme point quantity average ± 20% scope in.
Advantages of the present invention:
A kind of on-line trial authentication method based on dynamic threshold of the present invention, online handwritten signature verification method common with tradition is different in that: 1) coordinate of on-line signature, pressure, velocity information are carried out comprehensive modeling by the inventive method, and then consider the effect of each factor, more effectively improve the accuracy rate that signature judges；2) signature be have employed the strategy that segmentation differentiates by the inventive method, and according to my writing style of signature, automatically selects segmentation method more stable in two kinds of partition strategies, such that it is able to improve the accuracy that in signature judge process, the similarity of signature judges；3) threshold value that the inventive method uses in signature judge process to be measured is to be dynamically determined according to different signatures, be conducive to embodying the deviation characteristics of everyone signature, namely take into account the problem that the feature stability of everyone signature is different, thus obtaining everyone best comparison result.
Accompanying drawing explanation
Fig. 1 is every multidate information schematic diagram that the handwriting pad of the on-line trial authentication method based on dynamic threshold of an embodiment of the present invention and system obtains；
Fig. 2 is the system hardware connection diagram of an embodiment of the present invention；
Fig. 3 is system comprising modules and the functions structural representation of an embodiment of the present invention；
Fig. 4 is the on-line trial operation interface schematic diagram of an embodiment of the present invention；
Fig. 5 is the on-line trial authentication method flow chart based on dynamic threshold of an embodiment of the present invention；
Fig. 6 is the signature sequence data matching process schematic diagram of an embodiment of the present invention；
Fig. 7 is the signature stepwise schematic views of an embodiment of the present invention.
Detailed description of the invention
Below in conjunction with accompanying drawing, an embodiment of the present invention is described further.
As it is shown in figure 1, in the embodiment of the present invention, the signature writing pad of employing is general touch control type flat-panel display.This product can provide the coordinate at number of signature strong point, pressure information；The subsidiary writing pencil of this equipment is to determine light target current location, it is provided that the two-dimensional coordinate information of cursor position.When this handwriting input device is placed in face of user with normal direction, zero is positioned at the upper left corner of equipment, and X-axis points to right, and Y-axis is downwardly directed.
As in figure 2 it is shown, in the embodiment of the present invention, the handwriting pad used is connected with computer by USB interface, computer screen is transferred on handwriting pad by this universal touch control panel display, uses Universal USB interface, connects simple and convenient.It comprises two data line, and one, as output, transmits image；Article one, as input, transmit coordinate, pressure information that writing pencil is currently put.
As it is shown on figure 3, the on-line trial certification of the embodiment of the present invention is made up of signed data collection, signature pretreatment, feature extraction, signature template base and signature authentication.The handwritten signature verification of the embodiment of the present invention includes two stages, i.e. registration phase and authentication phase.Registration phase is the same with authentication phase step before feature extraction, namely inputs signature, pretreatment, feature extraction.After feature extraction, the feature of extraction together with the identity information of input, is stored in template database, when comparing for authentication phase by registration phase.After authentication phase, feature extraction, system finds the template that input signature is corresponding from template database, the two is carried out matching ratio relatively, then carries out decision-making according to result of the comparison.If signature to be measured meets the decision condition of authentication model, then accepting this signature is true signature, otherwise then thinks that this signature is for puppet signature.
Fig. 4 is present system operation interface schematic diagram, writes district, signature sample list, signature list to be measured, signature to be measured differentiation result output area, signature image preview region and correlation function control knob including signature.
A kind of on-line trial authentication method based on dynamic threshold, method flow is as it is shown in figure 5, comprise the following steps:
Step 1, gather my multiple signature sample data, namely the movement locus of lettering pen is carried out constant duration sampling, it is thus achieved that the X-coordinate value of sampled point, Y-coordinate value and the contact force value to clipboard；And all sampling numbers evidence of signature is carried out validity check, if signed data is qualified, then allow the submission of signed data, perform step 2；Otherwise return and perform step 1；Described effect checks, the coordinate write including signing crosses the border inspection and signature reasonableness check, namely sign always count, whether the length and width scope of total writing time, signature reaches the threshold value that sets.
In the embodiment of the present invention, the collection of my sample data of signing.Signature carries out writing of signed data by handwriting pad in person, the movement locus of lettering pen is carried out constant duration sampling by system, obtaining the X of sampled point, Y-coordinate value and the contact force value to clipboard, the data ultimately producing signature every bit are recorded as: (sampled point numbering, X-coordinate, Y coordinate, timestamp, pressure).Then to all sampling numbers of signature according to carrying out validity check, if not passing through inspection, then again write effective signature, or signer carry out specification signature writing training after carry out the collection of sample again.
In the embodiment of the present invention, system provides signature training function: signer is before formally writing signature, it is possible to use training function, is familiar with the writing feature of handwriting pad and signature pen.Select signature training function, write signature in input area, write latter 2 seconds systems of end and automatically remove signature.The user profile of input signature: before formal collecting sample and signature to be measured, first the user identity relevant information of signature is set.
Step 2, signature sample choose the foundation with template base；
Requirement (minimum quantity requirement) chosen by the sample whether quantity of the signature sample that step 2-1, judgement gather reaches to set, and if so, performs step 2-2；Otherwise, execution step 1 is returned；
Step 2-2, rising of signature is started to write a little and the extreme point of coordinate is as waypoint, maybe signature acted point of starting to write, the angular velocity extreme point of signature and the extreme point of linear velocity of signing as waypoint, it is achieved signed data is carried out segment processing；
Fig. 7 is the signature stepwise schematic views of a kind of on-line trial authentication method based on dynamic threshold of the present invention.In figure, the extreme point acting coordinate of starting to write and sign according to signature carries out segmentation.
Step 2-3, adopting dynamic time programming DTW(DynamicTimeWarping) method judges two degree of similarity signed, the final sample of acquisition template base；
Fig. 6 is the main thought schematic diagram of signature sequence data dynamic matching method DTW in the present invention, is provided with two signature sequence S and T, and both sample frequencys are the same, wherein S=(s1,s2,…,sn), T=(t1,t2,…,tm), for a signing messages, signed data is regarded as one from origin-to-destination point sequence, s1,s2,…,snRefer to corresponding for signature S the 1st of point sequence the, 2 ... nth point, the i.e. all of curve formed above in Fig. 7；Signature sequence T is similar；The target of signature sequence data coupling seeks to find an optimum point-to-point corresponding relation, and the accumulative gap making two samples is minimum.In order to find such point-to-point alignment relation, the matrix of one n × m of definition, (i j) records the distance between two point si and tj of two samples to entry of a matrix element.Matching process seeks to find one from (1,1) to (n, path W=(w m)1,w2,…,wk), max (n, m) < K < n+m so that the element sum on this path is minimum.Fig. 6 shows that two coordinate sequences signed mate the point-to-point graph of a relation obtained.For ease of observing, mutually corresponding some straight line is coupled together by figure.It can be seen that because two sequences compared present irregular change, the point-to-point relation obtained has substantial amounts of one-to-many situation, wherein has and much belong to irrational matching relationship, it should avoid as far as possible.The present invention combines the method for signature segmentation, the coupling of signature sequence data has been done and has further optimized.
Step 2-3-1, calculate all signatures coordinate between any two overall DTW value respectively, DTW value, pressure segmentation key point DTW value, the overall DTW value of speed, speed segmentation and DTW value and speed waypoint DTW value are sued for peace in coordinate fragments summation DTW value, coordinate fragments key point DTW value, pressure overall DTW value, pressure segmentation, and calculate above-mentioned DTW value average and deviation；
In the embodiment of the present invention, the calculating of DTW value between signature sample.Respectively for the coordinate of signature, pressure, speed, calculate all signature sample DTW value between any two.The calculating of DTW value is considered by segmentation and two kinds of situations of not segmentation, namely calculates the DTW value of the summation of all segmentation DTW values, segmentation key point in the DTW value in not segmentation situation, segmentation situation.Computationally state on the basis of DTW value, then calculate the average of all kinds of DTW value of all signature sample, deviation information, support information when choosing as subsequent signature sample.
Step 2-3-2, delete the signature sample that in all samples, sampling number grid deviation is maximum and DTW value deviation is maximum；
Step 2-3-3, repeatedly execution step 1 are to step 2-3-3 until the volume residual of sample reaches the sample size retained needed for final sample storehouse；
Step 2-3-4, judge that whether residue sample meets signature stability condition, namely judge whether residue sample DTW deviation between any two is respectively less than the DTW threshold value that user sets, if so, then will remain the signature sample final sample as template base；Otherwise return and perform step 1；
In the embodiment of the present invention, it is judged that the whether stable method of sample is judge according to the DTW upper threshold set, when remaining sample DTW deviation being both less than regulation DTW upper threshold between any two, then residue sample meets stability condition.
In the embodiment of the present invention, if active user is but without carrying out sample collection, or the sample collection of active user is but without being fully completed, cause that template base is but without being successfully established, even when there is the signature to be measured of current signature, signature so to be measured also cannot complete certification, and system is shown as un-authenticated state.
When after 5 samples of collection successful acquisition of sample, the DTW index of 5 samples of system-computed and deviation, it is judged that the stability of 5 samples.When 5 samples meet stability condition, sample is chosen successfully, enters the establishment stage of template base；When 5 samples are unsatisfactory for stability condition, system requirements user adds 5 new samples, proceeds to collection place of signature sample.When after 10 samples of collection successful acquisition of sample, DTW index between system-computed sample and deviation, from 10 samples, circulation weeds out 5 maximum samples of deviation, final retain 5 and treats judgment sample.When remaining 5 samples meet stability condition, sample is chosen successfully, enters the establishment stage of template base；When remaining 5 samples and being unsatisfactory for stability condition, the stability that system prompt active user writes signature is too poor, it is proposed that after user adds training again, then carry out the collection of sample.
Step 2-4, using the remaining signature sample of stablizing chosen as template base sample, calculation template storehouse sample DTW value between any two, and it is added up, determine the dynamic discriminant strategy of the residue stable signature DTW value deviation dynamic threshold of coordinate, the DTW value deviation dynamic threshold of pressure, the DTW value deviation dynamic threshold of speed and applicable current signature, and above-mentioned information is stored in template base；Described determination be suitable for current signature dynamic discriminant strategy, described dynamic discriminant strategy includes: using signature rise start to write a little and coordinate extreme point as waypoint carry out segmentation and using signature rise start to write point, signature angular velocity extreme point and sign linear velocity extreme point carry out segmentation as waypoint；Select segmentation method: select under this segmentation method, the signature sample method that DTW value deviation is minimum between any two.
In the embodiment of the present invention, it is determined that the method for dynamic threshold: stablize the maximum of sample coordinate DTW range deviation between any two according to residue, determine the dynamic threshold of coordinate DTW deviation, i.e. upper deviation.In the embodiment of the present invention, it is S1, S2, S3 that residue stablizes sample, and wherein the DTW distance of S1 and S2 is d12, the DTW distance of S1 and S3 is d13, the DTW distance of S2 and S3 is d23, seek the meansigma methods DTW of DTW distanceavg=(d12+d13+d23)/3, finally according to each DTW distance, DTW average, solve the deviation of each DTW distance, so that it is determined that dynamic threshold.
Step 3, gather the signature sample data of tested user, namely the movement locus of lettering pen is carried out constant duration sampling, it is thus achieved that the X-coordinate value of sampled point, Y-coordinate value and the contact force value to clipboard；And to tested signature sampling number according to carrying out validity check, if signed data is qualified, then allow the submission of signed data, perform step 4；Otherwise return and perform step 3；
Step 4, judge the true and false of tested signature；
Step 4-1, the employing adding up tested signature are counted, the linear velocity extreme point quantity of the linear velocity extreme point type of the coordinate extreme point type (X extreme point, Y extreme point, X and Y be extreme point simultaneously) of the stroke count amount of rising and falling of stroke of signing, signature, the coordinate extreme point quantity of signature, the angular velocity extreme point type (angular velocity minimum, angular velocity maximum) of signature, the angular velocity extreme point quantity of signature, signature and signature；In conjunction with the dynamic threshold in template base, and based on above-mentioned information, signature to be measured is carried out dynamic filter, it is judged that whether tested signature meets dynamic filter condition, if so, perform step 4-2；Otherwise, then tested signature is pseudo-signature, terminates；
Specimen sample count average ± 20% scope in；Stroke count amount of rising and falling average ± 2 stroke counts within the scope of；Signature coordinate extreme point quantity average ± 20% scope in；Signature angular velocity extreme point quantity average ± 20% scope in；Signature linear velocity extreme point quantity average ± 20% scope in.
Step 4-2: select the dynamic discriminant strategy being suitable for current signature that tested signature is carried out segment processing；
Step 4-3: according to the coordinate of tested signed data, pressure and speed, calculates the coordinate of all signature sample data in itself and template base, DTW value deviation between pressure and speed；Form the Score index of multi-expert marking, as the support information of the follow-up judgement signature true and false to be measured.
Step 4-4: according to the overall DTW value deviation of coordinate, coordinate fragments summation DTW value deviation, coordinate fragments key point DTW value deviation between the signature sample deposited in signed data to be measured and template base；The overall DTW value deviation of pressure between signature sample in signed data to be measured and template base, pressure segmentation summation DTW value deviation, pressure segmentation key point DTW value deviation, signed data and the overall DTW value deviation of speed of signature sample in template base, speed segmentation and DTW value deviation, speed waypoint DTW value deviation, respectively in connection with the dynamic threshold of signature sample coordinate DTW value deviation in template base, the dynamic threshold of pressure DTW value deviation and the dynamic threshold of speed DTW value deviation, the method adopting multi-expert ballot scoring judges the true and false of signature to be measured, even total more than DTW value less than the DTW value quantity of dynamic threshold half, then signature to be measured is judged as true signature, otherwise signature to be measured is judged as pseudo-signature, forbid that tested user logs in；
In the embodiment of the present invention, use multi objective ballot mode, for instance in above-mentioned 9 DTW values, if 6 less than in template base the corresponding dynamic threshold upper limit, then it is true for being considered as signature to be measured, be otherwise puppet.
Step 4-5: differentiation result is carried out output display.
The software system of final present invention exploitation has carried out substantial amounts of test, and starts to have put into practical application.The signed data of 10 people of system acquisition is as experiment, and everyone establishes the template base of 5 samples, and each signature user have input other people signature of imitating of my 10 true signatures and 50 afterwards, amount to 100 true signatures and 500 copy signature.Finally test result indicate that, the signature of the present invention refuses rate lower than 10% by mistake, and misclassification rate is 0.25%, and overall recognition correct rate of signing is up to 97.8%.To sum up, the on-line trial authentication method based on dynamic threshold of the present invention achieves extraordinary effect.
Claims (6)
1. the on-line trial authentication method based on dynamic threshold, it is characterised in that comprise the following steps:
Step 1, gather my multiple signature sample data, namely the movement locus of lettering pen is carried out constant duration sampling, it is thus achieved that the X-coordinate value of sampled point, Y-coordinate value and the contact force value to clipboard；And all sampling numbers evidence of signature is carried out validity check, if signed data is qualified, then allow the submission of signed data, perform step 2；Otherwise return and perform step 1；
Step 2, signature sample choose the foundation with template base；
Requirement chosen by the sample whether quantity of the signature sample that step 2-1, judgement gather reaches to set, and if so, performs step 2-2；Otherwise, execution step 1 is returned；
Step 2-2, rising of signature is started to write a little and the extreme point of coordinate is as waypoint, maybe signature acted point of starting to write, the angular velocity extreme point of signature and the extreme point of linear velocity of signing as waypoint, it is achieved the segment processing to signed data；
Step 2-3, employing dynamic time programming DTW method judge two degree of similarity signed, and obtain the final sample of template base；
Step 2-3-1, calculate all signatures coordinate between any two overall DTW value respectively, DTW value, pressure segmentation key point DTW value, the overall DTW value of speed, speed segmentation and DTW value and speed waypoint DTW value are sued for peace in coordinate fragments summation DTW value, coordinate fragments key point DTW value, pressure overall DTW value, pressure segmentation, and calculate above-mentioned DTW value average and deviation；
Step 2-3-2, delete the signature sample that in all samples, sampling number grid deviation is maximum and DTW value deviation is maximum；
Step 2-3-3, repeatedly execution step 1 are to step 2-3-3 until the volume residual of sample reaches the sample size retained needed for final sample storehouse；
Step 2-3-4, judge that whether residue sample meets signature stability condition, namely judge whether residue sample DTW deviation between any two is respectively less than the DTW threshold value that user sets, if so, then will remain the signature sample final sample as template base；Otherwise return and perform step 1；
Step 2-4, using the remaining signature sample of stablizing chosen as template base sample, calculation template storehouse sample DTW value between any two, and it is added up, determine the dynamic discriminant strategy of the residue stable signature DTW value deviation dynamic threshold of coordinate, the DTW value deviation dynamic threshold of pressure, the DTW value deviation dynamic threshold of speed and applicable current signature, and above-mentioned information is stored in template base；
Step 3, gather the signature sample data of tested user, namely the movement locus of lettering pen is carried out constant duration sampling, it is thus achieved that the X-coordinate value of sampled point, Y-coordinate value and the contact force value to clipboard；And to tested signature sampling number according to carrying out validity check, if signed data is qualified, then allow the submission of signed data, perform step 4；Otherwise return and perform step 3；
Step 4, judge the true and false of tested signature；
Step 4-1, add up the linear velocity extreme point quantity of the sampling number of tested signature, the signature stroke count amount of rising and falling of stroke, the coordinate extreme point type of signature, the coordinate extreme point quantity of signature, the angular velocity extreme point type of signature, the angular velocity extreme point quantity of signature, the linear velocity extreme point type of signature and signature；In conjunction with the dynamic threshold in template base, and based on above-mentioned information, signature to be measured is carried out dynamic filter, it is judged that whether tested signature meets dynamic filter condition, if so, perform step 4-2；Otherwise, then tested signature is pseudo-signature, terminates；
Step 4-2: select the dynamic discriminant strategy being suitable for current signature that tested signature is carried out segment processing；
Step 4-3: according to the coordinate of tested signed data, pressure and speed, calculates the coordinate of all signature sample data in itself and template base, DTW value between pressure and speed；
Step 4-4: according to the overall DTW value deviation of coordinate, coordinate fragments summation DTW value deviation, coordinate fragments key point DTW value deviation between the signature sample deposited in signed data to be measured and template base；The overall DTW value deviation of pressure between signature sample in signed data to be measured and template base, pressure segmentation summation DTW value deviation, pressure segmentation key point DTW value deviation, signed data and the overall DTW value deviation of speed of signature sample in template base, speed segmentation and DTW value deviation, speed waypoint DTW value deviation, respectively in connection with the dynamic threshold of signature sample coordinate DTW value deviation in template base, the dynamic threshold of pressure DTW value deviation and the dynamic threshold of speed DTW value deviation, the method adopting multi-expert ballot scoring judges the true and false of signature to be measured, even total more than DTW value less than the DTW value quantity of dynamic threshold half, then signature to be measured is judged as true signature, otherwise signature to be measured is judged as pseudo-signature, forbid that tested user logs in；
Step 4-5: differentiation result is carried out output display.
2. the on-line trial authentication method based on dynamic threshold according to claim 1, it is characterized in that, validity check described in step 1 and step 3, the coordinate write including signing crosses the border inspection and signature reasonableness check, namely sign always count, whether the length and width scope of total writing time, signature reaches the threshold value that sets.
3. the on-line trial authentication method based on dynamic threshold according to claim 1, it is characterized in that, determination described in step 2-4 and step 4-2 be suitable for current signature dynamic discriminant strategy, described dynamic discriminant strategy includes: using signature rise start to write a little and coordinate extreme point as waypoint carry out segmentation and using signature rise start to write point, signature angular velocity extreme point and sign linear velocity extreme point carry out segmentation as waypoint；Select segmentation method: select under this segmentation method, the signature sample method that DTW value deviation is minimum between any two.
4. the on-line trial authentication method based on dynamic threshold according to claim 1, it is characterized in that, determination described in step 2-4 remains the DTW value deviation dynamic threshold of stable signature coordinate, particularly as follows: determine that residue stablizes sample coordinate DTW distance between any two, obtain whole coordinate DTW distance average, and according to each coordinate DTW distance and coordinate DTW distance average, obtain the deviation of DTW distance, select deviation maximum as the DTW value deviation dynamic threshold of coordinate；The DTW value deviation dynamic threshold of pressure is identical with the DTW value deviation dynamic threshold defining method of coordinate with the DTW value deviation dynamic threshold value determination method of speed.
5. the on-line trial authentication method based on dynamic threshold according to claim 1, it is characterised in that the coordinate extreme point type of the signature described in step 4-1 includes X extreme point, Y extreme point, X and Y extreme point simultaneously；The angular velocity extreme point type of signature includes angular velocity minimum and angular velocity maximum.
6. the on-line trial authentication method based on dynamic threshold according to claim 1, it is characterised in that the dynamic filter condition described in step 4-1 is: specimen sample count average ± 20% scope in；Stroke count amount of rising and falling average ± 2 stroke counts within the scope of；Signature coordinate extreme point quantity average ± 20% scope in；Signature angular velocity extreme point quantity average ± 20% scope in；Signature linear velocity extreme point quantity average ± 20% scope in.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201310521026.6A CN103593673B (en) | 2013-10-27 | 2013-10-27 | A kind of on-line trial authentication method based on dynamic threshold |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201310521026.6A CN103593673B (en) | 2013-10-27 | 2013-10-27 | A kind of on-line trial authentication method based on dynamic threshold |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN103593673A CN103593673A (en) | 2014-02-19 |
| CN103593673B true CN103593673B (en) | 2016-07-20 |
Family
ID=50083805
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201310521026.6A Expired - Fee Related CN103593673B (en) | 2013-10-27 | 2013-10-27 | A kind of on-line trial authentication method based on dynamic threshold |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN103593673B (en) |
Families Citing this family (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110781469B (en) | 2014-08-07 | 2021-01-01 | 创新先进技术有限公司 | Identity authentication method and device |
| CN105095709A (en) * | 2015-09-09 | 2015-11-25 | 西南大学 | On-line signature identification method and on-line signature identification system |
| CN105844726B (en) * | 2016-03-18 | 2018-04-17 | 吉林大学 | A kind of handwritten signature C++Builder language |
| CN107302433A (en) * | 2016-04-15 | 2017-10-27 | 平安科技（深圳）有限公司 | Method of calibration, verification server and the user terminal of electronic signature |
| 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 |
| CN106778568B (en) * | 2016-12-05 | 2020-08-21 | 上海携程商务有限公司 | Method for processing verification code based on WEB page |
| CN110866499B (en) * | 2019-11-15 | 2022-12-13 | 爱驰汽车有限公司 | Handwritten text recognition method, system, device and medium |
| CN111291636A (en) * | 2020-01-19 | 2020-06-16 | 深圳壹账通智能科技有限公司 | Method, device and system for effectively identifying electronic signature and storage medium |
| CN111985319A (en) * | 2020-07-13 | 2020-11-24 | 上海眼控科技股份有限公司 | Signature identification method and device |
| CN112035825A (en) * | 2020-09-01 | 2020-12-04 | 中国银行股份有限公司 | Method for logging in application APP, client and server |
| CN113468987B (en) * | 2021-06-17 | 2023-04-18 | 重庆傲雄在线信息技术有限公司 | Electronic handwriting authentication method, system, electronic equipment and storage medium |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US5825906A (en) * | 1994-11-30 | 1998-10-20 | Nippondenso Co., Ltd. | Signature recognition system |
| CN101079707A (en) * | 2007-06-21 | 2007-11-28 | 中国科学院合肥物质科学研究院 | Identity authentication method based on reversible handwriting signature |
| CN102592142A (en) * | 2012-01-05 | 2012-07-18 | 中国科学院合肥物质科学研究院 | Computer-system-based handwritten signature stability evaluation method |
-
2013
- 2013-10-27 CN CN201310521026.6A patent/CN103593673B/en not_active Expired - Fee Related
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US5825906A (en) * | 1994-11-30 | 1998-10-20 | Nippondenso Co., Ltd. | Signature recognition system |
| CN101079707A (en) * | 2007-06-21 | 2007-11-28 | 中国科学院合肥物质科学研究院 | Identity authentication method based on reversible handwriting signature |
| CN102592142A (en) * | 2012-01-05 | 2012-07-18 | 中国科学院合肥物质科学研究院 | Computer-system-based handwritten signature stability evaluation method |
Non-Patent Citations (1)
| Title |
|---|
| 基于HMM的在线手写签名认证系统设计与实现;栾方军等;《计算机应用与软件》;20080630;第25卷(第6期);全文 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN103593673A (en) | 2014-02-19 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN103593673B (en) | A kind of on-line trial authentication method based on dynamic threshold | |
| EP2874099B1 (en) | Dynamic handwriting verification and handwriting-based user authentication | |
| CN102592152B (en) | Computer-system-based online handwriting authentication method | |
| Blanco‐Gonzalo et al. | Performance evaluation of handwritten signature recognition in mobile environments | |
| CN103258157B (en) | A kind of online handwriting authentication method based on finger information and system | |
| CN110060531A (en) | A kind of computer On-line Examining system and method using intelligent digital pen | |
| CN103023658A (en) | Identity authentication method and identity authentication system based on signature | |
| CN113656761B (en) | Business processing method and device based on biological recognition technology and computer equipment | |
| Tolosana et al. | Exploiting complexity in pen-and touch-based signature biometrics | |
| CN105184267A (en) | Face-identification-based secondary-deformation auxiliary authorization method | |
| CN102592142A (en) | Computer-system-based handwritten signature stability evaluation method | |
| Alariki et al. | TOUCH GESTURE AUTHENTICATION FRAMEWORK FOR TOUCH SCREEN MOBILE DEVICES. | |
| AU2016236466B2 (en) | Method for identification of user's interaction signature | |
| Navaz et al. | Signature Authentication Using Biometric Methods | |
| WO2015032304A1 (en) | Online handwriting and identity authentication method having capability for identifying identity of attacker | |
| JP6924770B2 (en) | Dynamic movement tracking infrastructure for spatially divided segments Signature authentication system and method | |
| Yu et al. | Mental workload classification via online writing features | |
| Lu et al. | SCUT-MMSIG: a multimodal online signature database | |
| Anikin et al. | Framework for biometric user authentication based on a dynamic handwritten signature | |
| CN101894154A (en) | Method for extracting key pattern from image file | |
| CN105892661A (en) | Machine intelligent decision-making method | |
| Lee et al. | Signing in the air by pinching multi-fingers for advanced authentication on embedded system | |
| Long et al. | High discriminant features for writer-independent online signature verification | |
| Gonçalves et al. | Time/space based biometric handwritten signature verification | |
| Miguel-Hurtado et al. | A new approach to automatic signature complexity assessment |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for 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: 20160720 Termination date: 20171027 |
|
| CF01 | Termination of patent right due to non-payment of annual fee |