CN108763915A - Identifying code is established to generate model and generate the method, apparatus of identifying code - Google Patents
Identifying code is established to generate model and generate the method, apparatus of identifying code Download PDFInfo
- Publication number
- CN108763915A CN108763915A CN201810479656.4A CN201810479656A CN108763915A CN 108763915 A CN108763915 A CN 108763915A CN 201810479656 A CN201810479656 A CN 201810479656A CN 108763915 A CN108763915 A CN 108763915A
- Authority
- CN
- China
- Prior art keywords
- identifying code
- constraint information
- model
- generation
- code
- 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.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F21/00—Security arrangements for protecting computers, components thereof, programs or data against unauthorised activity
- G06F21/30—Authentication, i.e. establishing the identity or authorisation of security principals
- G06F21/31—User authentication
- G06F21/36—User authentication by graphic or iconic representation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/047—Probabilistic or stochastic networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Software Systems (AREA)
- General Engineering & Computer Science (AREA)
- Artificial Intelligence (AREA)
- Computational Linguistics (AREA)
- Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Computer Security & Cryptography (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Mathematical Physics (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Computer Hardware Design (AREA)
- Probability & Statistics with Applications (AREA)
- Character Discrimination (AREA)
Abstract
The present invention provides a kind of method established identifying code and generate model, and method includes：Obtain the identifying code of preset kind；It obtains constraint information collection and merges the correspondence established in constraint information set between each constraint information and each identifying code；Structure fights network comprising the generation for generating model and discrimination model；Confrontation network is generated according to the identifying code of acquisition and the training of corresponding constraint information, the generation model in the generation confrontation network obtained using training is verified code and generates model.The present invention also provides a kind of method generating identifying code, method includes：Sampling obtains constraint information from preset constraint information set；The identifying code that the constraint information input training that P n-dimensional random variable ns are obtained with sampling obtains is generated into model, the output result that model is generated using the identifying code obtains the corresponding identifying code of constraint information for sampling and obtaining.The present invention can simplify the generation step of identifying code, improve the production efficiency of identifying code.
Description
【Technical field】
The present invention relates to machine learning techniques fields more particularly to a kind of identifying code of establishing to generate model and generation verification
The method, apparatus of code.
【Background technology】
Currently, the form of identifying code includes mainly following several：Number, letter, Chinese character, image and above-mentioned element it is quiet
State combines or dynamic combined.And the prior art generates the tactful substantially similar of identifying code, by taking picture validation code generates as an example, stream
Journey can be summarized as：Define default characters；Random number is generated, word corresponding with random number is matched from preset characters concentration
Symbol；Above-mentioned steps are repeated according to the length for generating identifying code；Create blank image, drawing image background；By the word of generation
Symbol string is plotted in image to be verified code.That is, traditional verification code generation method includes two steps, firstly generates and test
Code is demonstrate,proved, the safety of generated identifying code is secondly promoted.Therefore, the prior art is in the step for generating the identifying code with high security
Rapid relatively complicated, formation efficiency is relatively low.
【Invention content】
In view of this, the present invention provides a kind of method, apparatus established identifying code and generate model and generation identifying code,
To the generation step of the identifying code of simplified high security, the formation efficiency of the identifying code of high security is improved.
The present invention is that technical scheme applied to solve the technical problem is to provide a kind of side for establishing identifying code and generating model
Method, the method includes：Obtain the identifying code of preset kind；Constraint information set is obtained, and is established each in constraint information set
Correspondence between constraint information and each identifying code；Structure fights network comprising the generation for generating model and discrimination model；Root
The generation is trained to fight network according to the identifying code of acquisition and corresponding constraint information, the generation obtained using training
Generation model in confrontation network is verified code and generates model；The identifying code generates model for being generated according to constraint information
Identifying code.
According to one preferred embodiment of the present invention, the identifying code of the preset kind include Digital verification code, alphabetical identifying code,
One kind in Chinese character identifying code, picture validation code.
According to one preferred embodiment of the present invention, the constraint information set is according to the preset rule of correspondence and acquired
The type of identifying code design.
According to one preferred embodiment of the present invention, described to establish in constraint information set between each constraint information and each identifying code
Correspondence include：It establishes corresponding between each constraint information and the correct verification result of each identifying code in constraint information set
Relationship.
According to one preferred embodiment of the present invention, the life is trained according to the identifying code of acquisition and corresponding constraint information
Include at confrontation network：Using acquired identifying code as authentic specimen；P n-dimensional random variable ns and constraint information input are generated into mould
Type, the output result that generation model is obtained is as generation sample；By constraint information and its corresponding authentic specimen, generate sample
Input as discrimination model；Alternately the training generation model and discrimination model, until the generation fights network convergence.
The present invention is that technical scheme applied to solve the technical problem is to provide a kind of method generating identifying code, the side
Method includes：Sampling obtains constraint information from preset constraint information set；The constraint that P n-dimensional random variable ns are obtained with sampling is believed
The identifying code that breath input training obtains generates model, and the output result that model is generated using the identifying code obtains described sample
The corresponding identifying code of constraint information arrived.
According to one preferred embodiment of the present invention, the method further includes：Obtained identifying code is shown, and is obtained
Recognition result input by user；Compare the recognition result with the obtained constraint information that samples to show if the two is consistent
It is verified, otherwise verifies and do not pass through.
The present invention is that technical scheme applied to solve the technical problem is to provide a kind of dress established identifying code and generate model
It sets, described device includes：First acquisition unit, the identifying code for obtaining preset kind；Second acquisition unit, for obtaining about
Beam information aggregate, and establish the correspondence in constraint information set between each constraint information and each identifying code；Network struction list
Member, for building comprising the generation confrontation network for generating model and discrimination model；Network training unit, for testing according to acquisition
It demonstrate,proves code and corresponding constraint information trains the generation to fight network, in the generation confrontation network obtained using training
Generation model be verified code generate model；The identifying code generates model and is used to generate identifying code according to constraint information.
According to one preferred embodiment of the present invention, the identifying code of the preset kind include Digital verification code, alphabetical identifying code,
One kind in Chinese character identifying code, picture validation code.
According to one preferred embodiment of the present invention, the constraint information set is according to the preset rule of correspondence and acquired
The type of identifying code design.
According to one preferred embodiment of the present invention, the second acquisition unit each constraint information in establishing constraint information set
It is specific to execute when correspondence between each identifying code：Establish each constraint information and each identifying code in constraint information set
Correspondence between correct verification result.
According to one preferred embodiment of the present invention, the network training unit is according to the identifying code of acquisition and corresponding
It is specific to execute when constraint information trains the generation confrontation network：Using acquired identifying code as authentic specimen；By P dimension with
Machine variable generates model with constraint information input, and the output result that generation model is obtained is as generation sample；By constraint information
And its input of corresponding authentic specimen, generation sample as discrimination model；Alternately the training generation model and discrimination model,
Until the generation fights network convergence.
The present invention is that technical scheme applied to solve the technical problem is to provide a kind of device generating identifying code, the dress
Set including：Sampling unit obtains constraint information for being sampled from preset constraint information set；Generation unit, for tieing up P
The identifying code that the constraint information input training that stochastic variable is obtained with sampling obtains generates model, and mould is generated using the identifying code
The output result of type obtains the corresponding identifying code of constraint information for sampling and obtaining.
According to one preferred embodiment of the present invention, described device further includes authentication unit, for performing the following operations：To institute
Obtained identifying code is shown, and obtains recognition result input by user；Compare the recognition result to obtain with the sampling
Constraint information show to be verified if the two is consistent, otherwise verify and do not pass through.
As can be seen from the above technical solutions, the present invention fights net by the identifying code of acquired high security to generating
Network is trained, and to generate model using the obtained identifying code of training, just can be realized according to the constraint information inputted
The purpose for generating the identifying code of corresponding high security further simplifies the generation step of the identifying code of high security, improves
The formation efficiency of the identifying code of high security.
【Description of the drawings】
Fig. 1 is the network structure for the generation confrontation network that one embodiment of the invention provides；
Fig. 2 is the network structure for the generation model that one embodiment of the invention provides；
Fig. 3 is the network structure for the discrimination model that one embodiment of the invention provides；
Fig. 4 is the method flow diagram for the generation identifying code that one embodiment of the invention provides；
Fig. 5 is the structure drawing of device established identifying code and generate model that one embodiment of the invention provides；
Fig. 6 is the structure drawing of device for the generation identifying code that one embodiment of the invention provides；
Fig. 7 is the block diagram for the computer system/server that one embodiment of the invention provides.
【Specific implementation mode】
To make the objectives, technical solutions, and advantages of the present invention clearer, right in the following with reference to the drawings and specific embodiments
The present invention is described in detail.
The term used in embodiments of the present invention is the purpose only merely for description specific embodiment, is not intended to be limiting
The present invention.In the embodiment of the present invention and "an" of singulative used in the attached claims, " described " and "the"
It is also intended to including most forms, unless context clearly shows that other meanings.
It should be appreciated that term "and/or" used herein is only a kind of incidence relation of description affiliated partner, indicate
There may be three kinds of relationships, for example, A and/or B, can indicate：Individualism A, exists simultaneously A and B, individualism B these three
Situation.In addition, character "/" herein, it is a kind of relationship of "or" to typically represent forward-backward correlation object.
Depending on context, word as used in this " if " can be construed to " ... when " or " when ...
When " or " in response to determination " or " in response to detection ".Similarly, depend on context, phrase " if it is determined that " or " if detection
(condition or event of statement) " can be construed to " when determining " or " in response to determination " or " when the detection (condition of statement
Or event) when " or " in response to detection (condition or event of statement) ".
Identifying code according to the present invention be in the case where carrying out account login, the scenes such as payment by related application current
The identifying code that interface is generated, such as in certain Video Applications login account, generated in log-in interface by the Video Applications
Identifying code can be logged in normally after completing to the verification of the identifying code inputted.Some embodiments of the present invention are related to two
Partial content：Identifying code is established to generate the method for model and established identifying code is utilized to generate the side that model generates identifying code
Method.
It is described first to establishing the method that identifying code generates model.In the present invention, it is based on generating confrontation network
(Generative Adversarial Nets, GAN) establishes identifying code and generates model.
Specifically, it may be used and be verified code generation model as under type pre-establishes：
(1) identifying code of preset kind is obtained.
There is currently various types of identifying codes, such as Digital verification code, alphabetical identifying code, Chinese character identifying code, picture to test
Demonstrate,prove code etc..In order to generate the identifying code with high security, the identifying code of the preset kind acquired in this step be with
The identifying code of the preset kind of high security.
It is understood that the identifying code trained using different types of identifying code, which generates model, can generate correspondence
The identifying code of type.For example, the identifying code trained using Digital verification code, which generates model, can generate corresponding each number
The identifying code of word, the identifying code trained using alphabetical identifying code, which generates model, can generate the identifying code of corresponding each letter
Deng.Therefore, this step trains to obtain testing for the identifying code that can generate corresponding types by obtaining the identifying code of preset kind
It demonstrate,proves code and generates model.For convenience's sake, the present invention is illustrated using the type of acquired identifying code as Digital verification code.
(2) obtain constraint information set, and establish in constraint information set each constraint information and acquired identifying code it
Between correspondence.
In this step, constraint information set is obtained according to the identifying code of acquired preset kind, and further established
Correspondence in acquired constraint information set between each constraint information and each identifying code.Wherein, acquired constraint letter
Breath collection is combined into one group of independently incoherent information aggregate, passes through the correspondence between each constraint information and each identifying code so that
Give a certain constraint information, just can generate identifying code corresponding with given constraint information, so according to generated
The corresponding constraint information of identifying code completes the verification to identifying code.
When obtaining constraint information set, can be set according to the type of the preset rule of correspondence and acquired identifying code
Meter obtains constraint information set.After obtaining constraint information set, further establish in constraint information set each constraint information with
Correspondence between each identifying code.Wherein, it establishes corresponding between each constraint information and each identifying code in constraint information set
Relationship is the correspondence established between each constraint information and the correct verification result of each identifying code.
For example, if acquired identifying code is Digital verification code, each Digital verification code corresponds to 0~9 this 10 respectively
Constraint information set then can be designed as { 0,1,2,3,4,5,6,7,8,9 } by number according to the rule of the corresponding number of number,
Further establish the correspondence between each constraint information and the Digital verification code of corresponding each number in constraint information set.It can be with
Understand, constraint information set can also be designed as according to the rule of the corresponding number of letter a, b, c, d, e, f, g, h, i,
J }, and establish a and the Digital verification code of corresponding " 1 ", the Digital verification code of the Digital verification code of b and correspondence " 2 ", c and correspondence " 3 "
Correspondence Deng between.If acquired identifying code is alphabetical identifying code, each letter identifying code corresponds to a~z this 26 respectively
Constraint information set can be then designed as { a, b, c, d...z }, further built by letter according to the rule of the corresponding letter of letter
The correspondence made treaty in beam information aggregate between each constraint information and the alphabetical identifying code of corresponding each letter.If acquired tests
Card code is Chinese character identifying code, for example, be respectively corresponding Chinese character " then ", " special ", " strong ", " profit " Chinese character identifying code, then can be according to
The rule of the corresponding Chinese character of number, constraint information set is designed as { 1,2,3,4 }, and further establish " 1 " and corresponding " then "
Chinese character identifying code, " 2 " and corresponding " special " Chinese character identifying code, " 3 " and Chinese character identifying code, " 4 " and corresponding " profit " for corresponding to " strong "
Chinese character identifying code between correspondence.If acquired identifying code is picture validation code, for example, corresponding " banana " and " fire
Constraint information set can be then designed as { 1,2 } by the picture validation code of vehicle " according to the rule of the corresponding picture tag of number, and
It is the picture validation code of " banana ", " 2 " and picture validation code that corresponding label is " train " further to establish " 1 " and corresponding label
Between correspondence.
(3) structure fights network comprising the generation for generating model and discrimination model.
In the present invention, structure fights network comprising the generation for generating model and discrimination model, and obtained based on training
The generation model in confrontation network is generated, the identifying code for generating high security identifying code is obtained and generates model.
The network structure used in the present invention for generating confrontation network is as shown in fig. 1 comprising generates model and differentiation
Model.Wherein, the responsibility for generating model is generation and authentic specimen generation sample as similar as possible, and the responsibility of discrimination model is then
It is to distinguish authentic specimen and generation sample as far as possible.It is trained by way of generating and fighting game between model and discrimination model whole
A generation fights network so that the authenticity for generating the generation sample of model output is as high as possible so that discrimination model without
It is generation sample or authentic specimen that method, which is distinguished by the obtained output of generation model,.
Network structure to generating generation model and discrimination model that confrontation network is included in the present invention illustrates：
1) network structure of model is generated
Generate confrontation network in generate model network structure it is as shown in Figure 2, mainly include input layer, middle layer and
Output layer.
Wherein, using P n-dimensional random variable ns and each constraint information as the input for generating model, P is tieed up into random become by input layer
Amount and constraint information are combined, and input middle layer afterwards obtained united information is carried out moulding (reshape).This
In, united information is carried out moulding being specially the matrix form for being converted to united information multirow.Middle layer includes N number of by warp
The module of lamination and information constrained layer composition, wherein N are the natural number more than or equal to 1.Specifically, each mould of middle layer letter in the block
Breath restraint layer is for combining the output information of warp lamination in the module with constraint information, by obtained united information
As the input of the warp lamination of next module in middle layer, progress is recycled with this.Processing by the N number of module of middle layer obtains
After output information, obtained output information is subjected to deconvolution processing, is used as obtained handling result by output layer
The output result output of model is generated, the output result for generating model is the identifying code of corresponding inputted constraint information.
Wherein, P n-dimensional random variable ns and the united mode of constraint information progress are by input layer：Constraint information is appended to P
The tail portion of n-dimensional random variable n, will be to add the input of united information that mode forms as middle layer.Similarly, information constrained layer
It is by the output information of warp lamination and the united mode of constraint information：Constraint information is appended to the output information of warp lamination
Tail portion, will be to add the input of united information that mode forms as warp lamination in next module.By by constraint information
It is appended to the mode of the tail portion of the output information of P n-dimensional random variable ns or warp lamination, can more accurately realize constraint information
It is corresponding between the identifying code for generating model output.
2) network structure of discrimination model
Generate confrontation network in discrimination model network structure it is as shown in Figure 3, mainly include input layer, middle layer,
Full articulamentum and output layer.
Wherein, using constraint information and its corresponding authentic specimen, generation sample as the input of discrimination model, input layer will
After sample data and constraint information are combined, obtained united information is inputted into middle layer.Middle layer includes M by convolution
The module of layer and information constrained layer composition, wherein M are the natural number more than or equal to 1.Specifically, each mould of middle layer information in the block
Restraint layer for the output information of convolutional layer in the module to be combined with constraint information, using obtained united information as
The input of next module convolutional layer in middle layer recycles progress with this.Processing by M module of middle layer obtains output information
Afterwards, the handling result obtained after output information process of convolution moulding (reshape) is combined with constraint information afterwards, by gained
The united information arrived inputs full articulamentum.Full articulamentum is mapped according to obtained united information, will be reflected by output layer
The output result that result is penetrated as discrimination model exports, and it is true that the output result of discrimination model, which is inputted sample data,
The probability of sample.
It is understood that be similarly will about for the united mode of output information of constraint information and sample data or convolutional layer
Beam information adding to the tail portion of sample data or the output information of convolutional layer mode, so as to more accurately realize constraint
It is corresponding between information and the output result of discrimination model.
(4) confrontation network is generated according to the identifying code of acquisition and its training of corresponding constraint information, is obtained using training
The generation model generated in confrontation network is verified code and generates model.
The generation confrontation network being made of generation model and discrimination model is trained by the way of alternately training, when
When entire generation confrontation network convergence, then it is assumed that the training for generating confrontation network terminates, and then the generation confrontation that training is obtained
Generation model in network generates model as identifying code, generates model by the identifying code, just can obtain and be inputted
The corresponding identifying code of constraint information.
Specifically, the training process for generating confrontation network is as follows：
Using acquired identifying code as authentic specimen.Using P n-dimensional random variable ns and each constraint information as input, will generate
Generation sample of the obtained output result of model as corresponding each constraint information.Then by each constraint information and its corresponding true
Real sample generates input of the sample as discrimination model, and corresponding discrimination model and life are obtained according to the output result of discrimination model
At the loss function of model, and then according to the loss function of the loss function of discrimination model and generation model, adjustment generates model
With the parameter in the network structure of discrimination model, network convergence is fought until generating.
Specifically, the loss function of following formula computational discrimination model can be used：
In formula：Indicate the loss of discrimination model when input is authentic specimen,Indicate that input is made a living
At the loss of discrimination model when sample.
Specifically, it can be calculated separately using following formula
In formula：D_aLIndicate that the output of discrimination model, N_real indicate the number of the authentic specimen of input discrimination model
Amount, N_fake indicate the quantity of the generation sample of input discrimination model, N_real=N_fake.y(i)Indicate that sample i inputs differentiate
Desired output when model, i.e., the annotation results of each sample, if authentic specimen, then y(i)It is 1, if generating sample, then y(i)
It is 0.
Therefore, the loss function of discrimination model can be further simplified as：
Specifically, the parameter of discrimination model is adjusted according to above-mentioned loss function, the training objective of discrimination model is to minimize
Above-mentioned loss function.Optionally, during a specific implementation of the present embodiment, if the obtained loss in preset times
Function convergence, then it is assumed that the loss function minimizes；If or obtained loss function converges to preset value, then it is assumed that
Loss function minimizes；Can also be if that frequency of training is more than preset times, then it is assumed that loss function minimizes.When loss letter
When number minimizes, that is, think to complete the training of discrimination model.
Specifically, following formula can be used to calculate the loss function for generating model：
In formula：N_fake is the quantity for the generation sample for inputting discrimination model；y(i)Output is indicated to generate sample
When, the desired output of discrimination model, y herein(i)It is 1, therefore above-mentioned formula can be simplified, obtained simplified result
For：
Specifically, the parameter for generating model and discrimination model is adjusted according to above-mentioned loss function, generates the training of model
Target is to minimize above-mentioned loss function.Optionally, during a specific implementation of the present embodiment, if in preset times
Obtained loss function convergence, then it is assumed that the loss function minimizes；If or obtained loss function converges to
Preset value, then it is assumed that loss function minimizes；Can also be if that frequency of training is more than preset times, then it is assumed that loss function is most
Smallization.
When the loss function of the loss function and discrimination model that generate model minimizes, that is, think to fight net to generating
The training of network is completed, and just can will be generated the generation model fought in network and be generated model as identifying code.Using obtained
Identifying code generates model, can export high security corresponding with the constraint information inputted according to the constraint information inputted
Identifying code.
It will additionally be appreciated that when training generates confrontation network, it can also be to generating the differentiation mould fought in network
Type carries out pre-training, after making it have certain differentiation accuracy, in the discrimination model to being obtained by generation model and pre-training
The generation confrontation network of composition is trained.The process of pre-training discrimination model is：Using acquired identifying code as true sample
This, will tie up random vector and the obtained output result of constraint information as sample is generated, by each constraint by generation model according to P
Information and corresponding authentic specimen generate sample as input training discrimination model.Complete the pre- instruction to discrimination model
After white silk, then the generation confrontation network by generation model and discrimination model is trained by the way of alternately training.
Next the mode that identifying code is generated to the present invention is described, and the core concept that identifying code is generated in the present invention exists
In：The identifying code obtained using advance training is generated model and obtains the verification of high security corresponding with the constraint information inputted
Code.
Fig. 4 is the method flow diagram for the generation identifying code that one embodiment of the invention provides, as shown in Figure 4, the method
Including：
In 401, is sampled from preset constraint information set and obtain constraint information.
In this step, each constraint information corresponds to the preset kind identifying code to be generated in preset constraint information set
Correct verification as a result, and sampling obtained constraint information for generating corresponding identifying code.It has been described in front about
Correspondence between beam information and identifying code, herein without repeating.If for example, generating the verification that model generates using identifying code
When code is Digital verification code, then preset constraint information set can be include 0~9 this 10 digital information aggregates 0,1,
2,3,4,5,6,7,8,9 }, the sampling constraints information from the constraint information set is right to generate corresponding 0~9 this 10 number institutes
The identifying code answered.In this application, the number of the constraint information obtained to sampling can be one without limiting, can also
It is multiple.
In 402, the constraint information input identifying code that training obtains in advance that P n-dimensional random variable ns are obtained with sampling is generated
Model, the output result that model is generated using the identifying code obtain the corresponding identifying code of constraint information for sampling and obtaining.
In this step, using the step 401 obtained constraint information of sampling and P n-dimensional random variable ns as inputting, using pre-
The output result that the identifying code that first training obtains generates model obtains the corresponding identifying code of inputted constraint information.
It is understood that when if sampling obtains multiple constraint informations, according to the sampling order of constraint information, will verify
Code generates the output result that model obtains and is combined, to obtain corresponding to the combined authentication code of multiple sampling constraints information.Citing
For, if the obtained constraint information of sampling is respectively (0,3,2,5), by identifying code generate model generates respectively correspondence 0,3,2,
5 Digital verification code, and the Digital verification code generated is combined according to sampling order, obtain corresponding 4 digital numbers
Word identifying code.
After generating identifying code, the identifying code generated is shown, the identifying code shown is known by user
Not；After the recognition result for obtaining user again, it is compared with the constraint information that sampling obtains according to recognition result input by user,
If the two is consistent, it is verified, otherwise authentication failed.
For example, sampling obtains 4 constraint informations from constraint information set, if respectively 0,2,4,7, about by 4
Beam information input respectively identifying code generate model, according still further to sampling order by identifying code generate the obtained output result of model into
Row combination, obtains the Digital verification code of corresponding 4 constraint informations, and user inputs after obtained Digital verification code is identified
The recognition result inputted is obtained constraint information with sampling and is compared by recognition result, if recognition result input by user is
(0247), it is consistent that constraint information is obtained with sampling, then shows to be verified, otherwise show authentication failed.
Fig. 5 is the structure drawing of device established identifying code and generate model that one embodiment of the invention provides, as shown in Figure 5,
Described device includes：First acquisition unit 51, second acquisition unit 52, network struction unit 53 and network training unit 54.
First acquisition unit 51, the identifying code for obtaining preset kind.
There is currently various types of identifying codes, such as Digital verification code, alphabetical identifying code, Chinese character identifying code, picture to test
Demonstrate,prove code etc..In order to generate the identifying code with high security, the verification of the preset kind acquired in first acquisition unit 51
Code is the identifying code of the preset kind with high security.
It is understood that the identifying code trained using different types of identifying code, which generates model, can generate correspondence
The identifying code of type.For example, the identifying code trained using Digital verification code, which generates model, can generate corresponding each number
The identifying code of word, the identifying code trained using alphabetical identifying code, which generates model, can generate the identifying code of corresponding each letter
Deng.Therefore, first acquisition unit 51 can generate testing for corresponding types by obtaining the identifying code of preset kind to train to obtain
The identifying code for demonstrate,proving code generates model.
Second acquisition unit 52, for obtaining constraint information set, and establish in constraint information set each constraint information with
Correspondence between acquired identifying code.
The identifying code of preset kind of the second acquisition unit 52 acquired in first acquisition unit 51 obtains constraint information
Set, and further establish the correspondence in acquired constraint information set between each constraint information and each identifying code.Its
In, the constraint information collection acquired in second acquisition unit 52 is combined into one group of independently incoherent information aggregate, is believed by each constraint
Correspondence between breath and each identifying code so that give a certain constraint information, just can generate and given constraint information
Corresponding identifying code, and then according to constraint information corresponding with the identifying code generated, complete the verification to identifying code.
Second acquisition unit 52, can be according to the preset rule of correspondence and acquired when obtaining constraint information set
The type of identifying code designs to obtain constraint information set.Second acquisition unit 52 is further built after obtaining constraint information set
The correspondence made treaty in beam information aggregate between each constraint information and each identifying code.Wherein, second acquisition unit 52 is being established
Correspondence in constraint information set between each constraint information and each identifying code is to establish each constraint information and each identifying code
Correct verification result between correspondence.
Network struction unit 53, for building comprising the generation confrontation network for generating model and discrimination model.
The structure of network struction unit 53 fights network comprising the generation for generating model and discrimination model, to based on trained
To generation confrontation network in generation model, obtain for generate high security identifying code identifying code generate model.
Network struction unit 53 builds the network knot for generating generation model and discrimination model that confrontation network is included respectively
Structure.
1) structure generates the network structure of model
Network struction unit 53 builds the main generation model for including input layer, middle layer and output layer.Wherein, for life
It is had been described above at each layer structure of model, herein without repeating.
2) network structure of discrimination model is built
Network struction unit 53 builds the main discrimination model including input layer, middle layer, full articulamentum and output layer.Its
In, each layer structure of discrimination model is had been described above, herein without repeating.
Network training unit 54, for generating confrontation net according to the identifying code of acquisition and its training of corresponding constraint information
Network, the generation obtained using training are fought the generation model in network and are verified code generation model.
Network training unit 54 by generation model and sentences the structure of network struction unit 53 by the way of alternately training
The generation confrontation network that other model is constituted is trained, and when entire generation confrontation network convergence, network training unit 54 is thought
The training for generating confrontation network terminates, and then the generation model in the generation confrontation network that training obtains is generated as identifying code
Model generates model by the identifying code, just can obtain identifying code corresponding with the constraint information inputted.
Specifically, the process of the training of network training unit 54 generation confrontation network is as follows：
Network training unit 54 is using acquired identifying code as authentic specimen.P is tieed up random become by network training unit 54
Amount, as inputting, will generate the obtained output result of model as the generation sample of corresponding each constraint information with each constraint information
This.Then network training unit 54 is using each constraint information and its corresponding authentic specimen, generation sample as the defeated of discrimination model
Enter, corresponding discrimination model is obtained according to the output result of discrimination model and generates the loss function of model, and then network training list
Member 54 is according to the loss function of the loss function and generation model of discrimination model, the network knot of adjustment generation model and discrimination model
Parameter in structure, until generating confrontation network convergence.Discrimination model and the loss function for generating model are had been described above,
Herein without repeating.
When the loss function of the loss function and discrimination model that generate model minimizes, network training unit 54 is recognized
To be completed to the training for generating confrontation network, it just can will generate the generation model fought in network and generate mould as identifying code
Type.Model is generated using obtained identifying code, the constraint information that can be exported and be inputted according to the constraint information inputted
The identifying code of corresponding high security.
It will additionally be appreciated that when network training unit 54 trains and generates confrontation network, can also be fought to generating
Discrimination model in network carries out pre-training, after making it have certain differentiation accuracy, to by generation model and pre-training
The generation confrontation network of obtained discrimination model composition is trained.The process of 54 pre-training discrimination model of network training unit
For：Using acquired identifying code as authentic specimen, random vector will be tieed up according to P by generation model and constraint information is obtained
Result is exported as sample is generated, each constraint information and corresponding authentic specimen, generation sample are sentenced as input training
Other model.After completing to the pre-training of discrimination model, then to by generation model and discrimination model by the way of alternately training
Generation confrontation network be trained.
Fig. 6 is the structure drawing of device for the generation identifying code that one embodiment of the invention provides, as shown in Figure 6, described device
Including：Sampling unit 61, generation unit 62 and authentication unit 63.
Sampling unit 61 obtains constraint information for being sampled from preset constraint information set.
Each constraint information corresponds to the correct verification for the preset kind identifying code to be generated in preset constraint information set
As a result, and sampling unit 61 samples obtained constraint information for generating corresponding identifying code.It has been described in front about
Correspondence between beam information and identifying code, herein without repeating.In this application, obtained pact is sampled to sampling unit 61
The number of beam information can be one, or multiple without limiting.
Generation unit 62, by the constraint information input identifying code that training obtains in advance that P n-dimensional random variable ns are obtained with sampling
Model is generated, the output result that model is generated using the identifying code obtains the corresponding verification of constraint information for sampling and obtaining
Code.
Sampling unit 61 is sampled obtained constraint information and P n-dimensional random variable ns as input, utilized by generation unit 62
The output result that the identifying code that training obtains in advance generates model obtains the corresponding identifying code of inputted constraint information.
It is understood that when if the sampling of sampling unit 61 obtains multiple constraint informations, generation unit 62 is according to constraint
The output result that identifying code generation model obtains is combined by the sampling order of information, to obtain corresponding to multiple sampling constraints
The combined authentication code of information.
Authentication unit 63, for being verified to the identifying code generated.
After generation unit 62 generates identifying code, authentication unit 63 is shown the identifying code generated, and user is to institute
The identifying code of displaying inputs recognition result after being identified；It is defeated according to user after authentication unit 63 obtains the recognition result of user
The recognition result entered is compared with the constraint information that sampling obtains, if the two is consistent, is verified, otherwise authentication failed.
As shown in fig. 7, computer system/server 012 is showed in the form of universal computing device.Computer system/clothes
The component of business device 012 can include but is not limited to：One or more processor or processing unit 016, system storage
028, the bus 018 of connection different system component (including system storage 028 and processing unit 016).
Bus 018 indicates one or more in a few class bus structures, including memory bus or Memory Controller,
Peripheral bus, graphics acceleration port, processor or the local bus using the arbitrary bus structures in a variety of bus structures.It lifts
For example, these architectures include but not limited to industry standard architecture (ISA) bus, microchannel architecture (MAC)
Bus, enhanced isa bus, Video Electronics Standards Association (VESA) local bus and peripheral component interconnection (PCI) bus.
Computer system/server 012 typically comprises a variety of computer system readable media.These media can be appointed
The usable medium what can be accessed by computer system/server 012, including volatile and non-volatile media, movably
With immovable medium.
System storage 028 may include the computer system readable media of form of volatile memory, such as deposit at random
Access to memory (RAM) 030 and/or cache memory 032.Computer system/server 012 may further include other
Removable/nonremovable, volatile/non-volatile computer system storage medium.Only as an example, storage system 034 can
For reading and writing immovable, non-volatile magnetic media (Fig. 7 do not show, commonly referred to as " hard disk drive ").Although in Fig. 7
It is not shown, can provide for the disc driver to moving non-volatile magnetic disk (such as " floppy disk ") read-write, and pair can
The CD drive that mobile anonvolatile optical disk (such as CD-ROM, DVD-ROM or other optical mediums) is read and write.In these situations
Under, each driver can be connected by one or more data media interfaces with bus 018.Memory 028 may include
There is one group of (for example, at least one) program module, these program modules to be configured at least one program product, the program product
To execute the function of various embodiments of the present invention.
Program/utility 040 with one group of (at least one) program module 042, can be stored in such as memory
In 028, such program module 042 includes --- but being not limited to --- operating system, one or more application program, other
Program module and program data may include the realization of network environment in each or certain combination in these examples.Journey
Sequence module 042 usually executes function and/or method in embodiment described in the invention.
Computer system/server 012 can also with one or more external equipments 014 (such as keyboard, sensing equipment,
Display 024 etc.) communication, in the present invention, computer system/server 012 is communicated with outside radar equipment, can also be with
One or more enable a user to the equipment interacted with the computer system/server 012 communication, and/or with make the meter
Any equipment that calculation machine systems/servers 012 can be communicated with one or more of the other computing device (such as network interface card, modulation
Demodulator etc.) communication.This communication can be carried out by input/output (I/O) interface 022.Also, computer system/clothes
Being engaged in device 012 can also be by network adapter 020 and one or more network (such as LAN (LAN), wide area network (WAN)
And/or public network, such as internet) communication.As shown, network adapter 020 by bus 018 and computer system/
Other modules of server 012 communicate.It should be understood that although not shown in the drawings, computer system/server 012 can be combined
Using other hardware and/or software module, including but not limited to：Microcode, device driver, redundant processing unit, external magnetic
Dish driving array, RAID system, tape drive and data backup storage system etc..
Processing unit 016 is stored in program in system storage 028 by operation, to perform various functions using with
And data processing, such as realize a kind of method established identifying code and generate model, may include：
Obtain the identifying code of preset kind；
Constraint information set is obtained, and establishes the corresponding pass in constraint information set between each constraint information and each identifying code
System；
Structure fights network comprising the generation for generating model and discrimination model；
It trains the generation to fight network according to the identifying code of acquisition and corresponding constraint information, is obtained using training
The generation confrontation network in generation model be verified code generate model.
It can also realize a kind of method generating identifying code, may include：
Sampling obtains constraint information from preset constraint information set；
The identifying code that the constraint information input training that P n-dimensional random variable ns are obtained with sampling obtains is generated into model, utilizes institute
The output result for stating identifying code generation model obtains the corresponding identifying code of constraint information for sampling and obtaining.
Above-mentioned computer program can be set in computer storage media, i.e., the computer storage media is encoded with
Computer program, the program by one or more computers when being executed so that one or more computers execute in the present invention
State method flow shown in embodiment and/or device operation.For example, the method stream executed by said one or multiple processors
Journey may include：
Obtain the identifying code of preset kind；
Constraint information set is obtained, and establishes the corresponding pass in constraint information set between each constraint information and each identifying code
System；
Structure fights network comprising the generation for generating model and discrimination model；
It trains the generation to fight network according to the identifying code of acquisition and corresponding constraint information, is obtained using training
The generation confrontation network in generation model be verified code generate model.
Can also include：
Sampling obtains constraint information from preset constraint information set；
The identifying code that the constraint information input training that P n-dimensional random variable ns are obtained with sampling obtains is generated into model, utilizes institute
The output result for stating identifying code generation model obtains the corresponding identifying code of constraint information for sampling and obtaining.
With time, the development of technology, medium meaning is more and more extensive, and the route of transmission of computer program is no longer limited by
Tangible medium, can also directly be downloaded from network etc..The arbitrary combination of one or more computer-readable media may be used.
Computer-readable medium can be computer-readable signal media or computer readable storage medium.Computer-readable storage medium
Matter for example may be-but not limited to-system, device or the device of electricity, magnetic, optical, electromagnetic, infrared ray or semiconductor, or
The arbitrary above combination of person.The more specific example (non exhaustive list) of computer readable storage medium includes：There are one tools
Or the electrical connections of multiple conducting wires, portable computer diskette, hard disk, random access memory (RAM), read-only memory (ROM),
Erasable programmable read only memory (EPROM or flash memory), optical fiber, portable compact disc read-only memory (CD-ROM), light
Memory device, magnetic memory device or above-mentioned any appropriate combination.In this document, computer readable storage medium can
With to be any include or the tangible medium of storage program, the program can be commanded execution system, device or device use or
Person is in connection.
Computer-readable signal media may include in a base band or as the data-signal that a carrier wave part is propagated,
Wherein carry computer-readable program code.Diversified forms may be used in the data-signal of this propagation, including --- but
It is not limited to --- electromagnetic signal, optical signal or above-mentioned any appropriate combination.Computer-readable signal media can also be
Any computer-readable medium other than computer readable storage medium, which can send, propagate or
Transmission for by instruction execution system, device either device use or program in connection.
The program code for including on computer-readable medium can transmit with any suitable medium, including --- but it is unlimited
In --- wireless, electric wire, optical cable, RF etc. or above-mentioned any appropriate combination.
It can be write with one or more programming languages or combinations thereof for executing the computer that operates of the present invention
Program code, described program design language include object oriented program language-such as Java, Smalltalk, C++,
Further include conventional procedural programming language-such as " C " language or similar programming language.Program code can be with
It fully executes, partly execute on the user computer on the user computer, being executed as an independent software package, portion
Divide and partly executes or executed on a remote computer or server completely on the remote computer on the user computer.?
Be related in the situation of remote computer, remote computer can pass through the network of any kind --- including LAN (LAN) or
Wide area network (WAN) is connected to subscriber computer, or, it may be connected to outer computer (such as provided using Internet service
Quotient is connected by internet).
Using technical solution provided by the present invention, model is generated by the identifying code that training obtains in advance, can be realized
According to constraint information generate high security identifying code purpose, to simplify high security identifying code generation step,
Improve the formation efficiency of the identifying code of high security.
In several embodiments provided by the present invention, it should be understood that disclosed system, device and method can be with
It realizes by another way.For example, the apparatus embodiments described above are merely exemplary, for example, the unit
It divides, only a kind of division of logic function, formula that in actual implementation, there may be another division manner.
The unit illustrated as separating component may or may not be physically separated, aobvious as unit
The component shown may or may not be physical unit, you can be located at a place, or may be distributed over multiple
In network element.Some or all of unit therein can be selected according to the actual needs to realize the mesh of this embodiment scheme
's.
In addition, each functional unit in each embodiment of the present invention can be integrated in a processing unit, it can also
It is that each unit physically exists alone, it can also be during two or more units be integrated in one unit.Above-mentioned integrated list
The form that hardware had both may be used in member is realized, can also be realized in the form of hardware adds SFU software functional unit.
The above-mentioned integrated unit being realized in the form of SFU software functional unit can be stored in one and computer-readable deposit
In storage media.Above-mentioned SFU software functional unit is stored in a storage medium, including some instructions are used so that a computer
It is each that equipment (can be personal computer, server or the network equipment etc.) or processor (processor) execute the present invention
The part steps of embodiment the method.And storage medium above-mentioned includes：USB flash disk, mobile hard disk, read-only memory (Read-
Only Memory, ROM), random access memory (Random Access Memory, RAM), magnetic disc or CD etc. it is various
The medium of program code can be stored.
The foregoing is merely illustrative of the preferred embodiments of the present invention, is not intended to limit the invention, all essences in the present invention
With within principle, any modification, equivalent substitution, improvement and etc. done should be included within the scope of protection of the invention god.
Claims (16)
1. a kind of method established identifying code and generate model, which is characterized in that the method includes：
Obtain the identifying code of preset kind；
Constraint information set is obtained, and establishes the correspondence in constraint information set between each constraint information and each identifying code；
Structure fights network comprising the generation for generating model and discrimination model；
The generation is trained to fight network according to the identifying code of acquisition and corresponding constraint information, the institute obtained using training
It states the generation model generated in confrontation network and is verified code generation model；
The identifying code generates model and is used to generate identifying code according to constraint information.
2. according to the method described in claim 1, it is characterized in that, the identifying code of the preset kind include Digital verification code,
One kind in alphabetical identifying code, Chinese character identifying code, picture validation code.
3. according to the method described in claim 1, it is characterized in that, the constraint information set is according to the preset rule of correspondence
And the type of acquired identifying code designs.
4. according to the method described in claim 1, it is characterized in that, it is described establish in constraint information set each constraint information with it is each
Correspondence between identifying code includes：
Establish the correspondence between each constraint information and the correct verification result of each identifying code in constraint information set.
5. according to the method described in claim 1, it is characterized in that, according to the identifying code of acquisition and corresponding constraint information
The training generation fights network：
Using acquired identifying code as authentic specimen；
P n-dimensional random variable ns and constraint information input are generated into model, the output result that generation model is obtained is as generation sample；
Using constraint information and its corresponding authentic specimen, sample is generated as the input of discrimination model；
Alternately the training generation model and discrimination model, until the generation fights network convergence.
6. a kind of method generating identifying code, which is characterized in that the method includes：
Sampling obtains constraint information from preset constraint information set；
The identifying code that the constraint information input training that P n-dimensional random variable ns are obtained with sampling obtains is generated into model, is tested using described
The output result that card code generates model obtains the corresponding identifying code of constraint information for sampling and obtaining.
7. according to the method described in claim 6, it is characterized in that, the method further includes：
Obtained identifying code is shown, and obtains recognition result input by user；
Compare the recognition result to show to be verified, otherwise test if the two is consistent with the obtained constraint information that samples
Card does not pass through.
8. a kind of device established identifying code and generate model, which is characterized in that described device includes：
First acquisition unit, the identifying code for obtaining preset kind；
Second acquisition unit for obtaining constraint information set, and establishes each constraint information and each verification in constraint information set
Correspondence between code；
Network struction unit, for building comprising the generation confrontation network for generating model and discrimination model；
Network training unit, for training the generation to fight net according to the identifying code of acquisition and corresponding constraint information
Network, the generation obtained using training are fought the generation model in network and are verified code generation model；
The identifying code generates model and is used to generate identifying code according to constraint information.
9. device according to claim 8, which is characterized in that the identifying code of the preset kind include Digital verification code,
One kind in alphabetical identifying code, Chinese character identifying code, picture validation code.
10. device according to claim 8, which is characterized in that the constraint information set is according to preset corresponding rule
Then and the type of acquired identifying code designs.
11. device according to claim 8, which is characterized in that the second acquisition unit is establishing constraint information set
In correspondence between each constraint information and each identifying code when, it is specific to execute：
Establish the correspondence between each constraint information and the correct verification result of each identifying code in constraint information set.
12. device according to claim 8, which is characterized in that the network training unit is in the identifying code according to acquisition
And corresponding constraint information is when training the generations confrontation network, specific execution：
Using acquired identifying code as authentic specimen；
P n-dimensional random variable ns and constraint information input are generated into model, the output result that generation model is obtained is as generation sample；
Using constraint information and its corresponding authentic specimen, sample is generated as the input of discrimination model；
Alternately the training generation model and discrimination model, until the generation fights network convergence.
13. a kind of device generating identifying code, which is characterized in that described device includes：
Sampling unit obtains constraint information for being sampled from preset constraint information set；
Generation unit, the identifying code that the constraint information input training for obtaining P n-dimensional random variable ns with sampling obtains generate mould
Type, the output result that model is generated using the identifying code obtain the corresponding identifying code of constraint information for sampling and obtaining.
14. device according to claim 13, which is characterized in that described device further includes authentication unit, is used to execute
It operates below：
Obtained identifying code is shown, and obtains recognition result input by user；
Compare the recognition result to show to be verified, otherwise test if the two is consistent with the obtained constraint information that samples
Card does not pass through.
15. a kind of computer equipment, including memory, processor and it is stored on the memory and can be on the processor
The computer program of operation, which is characterized in that the processor is realized when executing described program as any in claim 1~7
Method described in.
16. a kind of computer readable storage medium, is stored thereon with computer program, which is characterized in that described program is handled
Such as method according to any one of claims 1 to 7 is realized when device executes.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810479656.4A CN108763915A (en) | 2018-05-18 | 2018-05-18 | Identifying code is established to generate model and generate the method, apparatus of identifying code |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810479656.4A CN108763915A (en) | 2018-05-18 | 2018-05-18 | Identifying code is established to generate model and generate the method, apparatus of identifying code |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN108763915A true CN108763915A (en) | 2018-11-06 |
Family
ID=64007292
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810479656.4A Pending CN108763915A (en) | 2018-05-18 | 2018-05-18 | Identifying code is established to generate model and generate the method, apparatus of identifying code |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108763915A (en) |
Cited By (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109815658A (en) * | 2018-12-14 | 2019-05-28 | 平安科技（深圳）有限公司 | A kind of verification method and device, computer equipment and computer storage medium |
| CN110009057A (en) * | 2019-04-16 | 2019-07-12 | 四川大学 | A kind of graphical verification code recognition methods based on deep learning |
| CN110210204A (en) * | 2019-05-30 | 2019-09-06 | 网易（杭州）网络有限公司 | Verification code generation method and device, storage medium and electronic equipment |
| CN110399712A (en) * | 2019-07-31 | 2019-11-01 | 网易（杭州）网络有限公司 | Validation-cross method, apparatus, medium and calculating equipment based on identifying code |
| CN111353140A (en) * | 2018-12-24 | 2020-06-30 | 阿里巴巴集团控股有限公司 | Verification code generation and display method, device and system |
| CN111460426A (en) * | 2020-04-02 | 2020-07-28 | 武汉大学 | Anti-evolution framework based anti-deep learning text verification code generation system and method |
| CN111488422A (en) * | 2019-01-25 | 2020-08-04 | 深信服科技股份有限公司 | Incremental method and device for structured data sample, electronic equipment and medium |
| CN111611767A (en) * | 2020-05-21 | 2020-09-01 | 北京百度网讯科技有限公司 | Verification method and device |
| CN111783064A (en) * | 2020-06-30 | 2020-10-16 | 平安国际智慧城市科技股份有限公司 | Method and device for generating graphic verification code, computer equipment and storage medium |
| CN112104457A (en) * | 2020-08-28 | 2020-12-18 | 苏州云葫芦信息科技有限公司 | Method and system for generating verification code for converting digital to Chinese character type |
| CN115001771A (en) * | 2022-05-25 | 2022-09-02 | 武汉极意网络科技有限公司 | Verification code defense method, system, equipment and storage medium based on automatic updating |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20100077210A1 (en) * | 2008-09-24 | 2010-03-25 | Yahoo! Inc | Captcha image generation |
| CN103400107A (en) * | 2013-07-05 | 2013-11-20 | 百度在线网络技术（北京）有限公司 | Method and device for generating dynamic verification code picture, and verification method and device |
| CN107392973A (en) * | 2017-06-06 | 2017-11-24 | 中国科学院自动化研究所 | Pixel-level handwritten Chinese character automatic generation method, storage device, processing unit |
| CN107967475A (en) * | 2017-11-16 | 2018-04-27 | 广州探迹科技有限公司 | A kind of method for recognizing verification code based on window sliding and convolutional neural networks |
-
2018
- 2018-05-18 CN CN201810479656.4A patent/CN108763915A/en active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20100077210A1 (en) * | 2008-09-24 | 2010-03-25 | Yahoo! Inc | Captcha image generation |
| CN103400107A (en) * | 2013-07-05 | 2013-11-20 | 百度在线网络技术（北京）有限公司 | Method and device for generating dynamic verification code picture, and verification method and device |
| CN107392973A (en) * | 2017-06-06 | 2017-11-24 | 中国科学院自动化研究所 | Pixel-level handwritten Chinese character automatic generation method, storage device, processing unit |
| CN107967475A (en) * | 2017-11-16 | 2018-04-27 | 广州探迹科技有限公司 | A kind of method for recognizing verification code based on window sliding and convolutional neural networks |
Non-Patent Citations (2)
| Title |
|---|
| GUIXIN YE 等: "Yet Another Text Captcha Solver:A Generative Adversarial Network Based Approach", 《PROCEEDINGS OF THE 2018 ACM SIGSAC CONFERENCE ON COMPUTER AND COMMUNICATIONS SECURITY》 * |
| HYUN KWON 等: "CAPTCHA Image Generation Systems Using Generative Adversarial Networks", 《IEICE TRANSACTIONS ON INFORMATION AND SYSTEMS》 * |
Cited By (15)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109815658A (en) * | 2018-12-14 | 2019-05-28 | 平安科技（深圳）有限公司 | A kind of verification method and device, computer equipment and computer storage medium |
| CN111353140A (en) * | 2018-12-24 | 2020-06-30 | 阿里巴巴集团控股有限公司 | Verification code generation and display method, device and system |
| CN111353140B (en) * | 2018-12-24 | 2024-03-22 | 阿里巴巴集团控股有限公司 | Verification code generation and display method, device and system |
| CN111488422A (en) * | 2019-01-25 | 2020-08-04 | 深信服科技股份有限公司 | Incremental method and device for structured data sample, electronic equipment and medium |
| CN110009057A (en) * | 2019-04-16 | 2019-07-12 | 四川大学 | A kind of graphical verification code recognition methods based on deep learning |
| CN110210204A (en) * | 2019-05-30 | 2019-09-06 | 网易（杭州）网络有限公司 | Verification code generation method and device, storage medium and electronic equipment |
| CN110399712A (en) * | 2019-07-31 | 2019-11-01 | 网易（杭州）网络有限公司 | Validation-cross method, apparatus, medium and calculating equipment based on identifying code |
| CN111460426A (en) * | 2020-04-02 | 2020-07-28 | 武汉大学 | Anti-evolution framework based anti-deep learning text verification code generation system and method |
| CN111611767A (en) * | 2020-05-21 | 2020-09-01 | 北京百度网讯科技有限公司 | Verification method and device |
| CN111611767B (en) * | 2020-05-21 | 2023-04-25 | 北京百度网讯科技有限公司 | Verification method and device |
| CN111783064A (en) * | 2020-06-30 | 2020-10-16 | 平安国际智慧城市科技股份有限公司 | Method and device for generating graphic verification code, computer equipment and storage medium |
| CN112104457A (en) * | 2020-08-28 | 2020-12-18 | 苏州云葫芦信息科技有限公司 | Method and system for generating verification code for converting digital to Chinese character type |
| CN112104457B (en) * | 2020-08-28 | 2022-06-17 | 苏州云葫芦信息科技有限公司 | Method and system for generating verification code for converting numbers into Chinese character types |
| CN115001771A (en) * | 2022-05-25 | 2022-09-02 | 武汉极意网络科技有限公司 | Verification code defense method, system, equipment and storage medium based on automatic updating |
| CN115001771B (en) * | 2022-05-25 | 2024-01-26 | 武汉极意网络科技有限公司 | Verification code defending method, system, equipment and storage medium based on automatic updating |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108763915A (en) | Identifying code is established to generate model and generate the method, apparatus of identifying code | |
| CN109147810B (en) | Establish the method, apparatus, equipment and computer storage medium of speech enhan-cement network | |
| CN108171257B (en) | Fine granularity image recognition model training and recognition methods, device and storage medium | |
| US20230041233A1 (en) | Image recognition method and apparatus, computing device, and computer-readable storage medium | |
| CN109902659B (en) | Method and apparatus for processing human body image | |
| US10817707B2 (en) | Attack sample generating method and apparatus, device and storage medium | |
| CN108052577A (en) | A kind of generic text content mining method, apparatus, server and storage medium | |
| CN107545241A (en) | Neural network model is trained and biopsy method, device and storage medium | |
| CN109815156A (en) | Displaying test method, device, equipment and the storage medium of visual element in the page | |
| CN110163257A (en) | Method, apparatus, equipment and the computer storage medium of drawing-out structure information | |
| CN109583501A (en) | Picture classification, the generation method of Classification and Identification model, device, equipment and medium | |
| CN110245348A (en) | A kind of intension recognizing method and system | |
| CN110378346A (en) | Establish the method, apparatus, equipment and computer storage medium of Text region model | |
| CN109599095A (en) | A kind of mask method of voice data, device, equipment and computer storage medium | |
| CN110363810A (en) | Establish the method, apparatus, equipment and computer storage medium of image detection model | |
| CN109523611A (en) | Identifying code Picture Generation Method and device | |
| CN106713370B (en) | A kind of identity identifying method, server and mobile terminal | |
| CN110232340A (en) | Establish the method, apparatus of video classification model and visual classification | |
| CN110276023A (en) | POI changes event discovery method, apparatus, calculates equipment and medium | |
| CN108654091A (en) | Method, medium, device and computing device for verification of practising fraud in game | |
| CN107908641A (en) | A kind of method and system for obtaining picture labeled data | |
| CN110347872A (en) | Video cover image extracting method and device, storage medium and electronic equipment | |
| CN108229535A (en) | Relate to yellow image audit method, apparatus, computer equipment and storage medium | |
| CN110378095A (en) | Validation-cross method, apparatus, medium and calculating equipment based on identifying code | |
| CN108428451A (en) | Sound control method, electronic equipment and speech control system |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20181106 |
|
| RJ01 | Rejection of invention patent application after publication |