CN116894242A

CN116894242A - Identification method and device of track verification code, electronic equipment and storage medium

Info

Publication number: CN116894242A
Application number: CN202310855800.0A
Authority: CN
Inventors: 董奕
Original assignee: Beijing Baidu Netcom Science and Technology Co Ltd
Current assignee: Beijing Baidu Netcom Science and Technology Co Ltd
Priority date: 2023-07-12
Filing date: 2023-07-12
Publication date: 2023-10-17

Abstract

The disclosure provides a method for identifying a track verification code, a device for identifying the track verification code, electronic equipment and a non-transient computer readable storage medium, and relates to the technical field of data, in particular to the technical field of deep learning, wherein the specific implementation scheme is as follows: acquiring track verification code drawing data, and acquiring a natural person drawing index value according to the track verification code drawing data; when the natural person drawing index value meets the natural person operation primary screening condition, extracting at least one biological behavior characteristic from the track verification code drawing data; and verifying whether the track verification code drawing data are generated by natural human operation according to the biological behavior characteristics. According to the technical scheme, the verification code input by natural human operation and the verification code simulated by the robot can be accurately and efficiently distinguished, the accuracy of distinguishing the track verification code is improved, and the safety of the network environment protected by verification behaviors is ensured.

Description

Identification method and device of track verification code, electronic equipment and storage medium

Technical Field

The disclosure relates to the technical field of data, in particular to the technical field of deep learning, and specifically relates to a track verification code identification method, a track verification code identification device, electronic equipment and a non-transitory computer readable storage medium.

Background

The verification code is used as a biological authentication technology, and is widely applied to various man-machine verification products in order to meet the requirement of the current network environment for identity authentication.

The verification code is realized by initiating a specific verification question, judging a natural person if the answer is correct, and judging a robot if the answer is wrong, wherein the important design concept is that the human is easy to answer and the machine is difficult to answer. The verification code is self-developed, so that the verification code is continuously struggled with a malicious black-product attacker, for example, for the graphic verification code, the black-product attacker obtains limited verification pictures through multiple accesses, and the picture characters are recognized to be forcedly cracked by combining OCR (Optical Character Recognition ) technology; the speech verification code can also be broken through the increasingly powerful speech recognition technology. The technology of verification codes and black-producing attackers is also continually upgraded in continuous combat.

Therefore, how to accurately and efficiently verify whether the verification code to be identified is generated by natural human operation is an important problem to be solved at present.

Disclosure of Invention

The present disclosure provides a method, apparatus, device and storage medium for identifying a track verification code, an identification device of the track verification code, an electronic device and a non-transitory computer readable storage medium.

According to an aspect of the present disclosure, there is provided a method for identifying a track verification code, including:

acquiring track verification code drawing data, and acquiring a natural person drawing index value according to the track verification code drawing data;

when the natural person drawing index value meets the natural person operation preliminary screening condition, extracting at least one biological behavior characteristic from the track verification code drawing data;

and verifying whether the track verification code drawing data are generated by natural human operation according to the biological behavior characteristics.

According to another aspect of the present disclosure, there is provided an identification apparatus for a track verification code, including:

the index value acquisition module is used for acquiring track verification code drawing data and acquiring natural person drawing index values according to the track verification code drawing data;

the biological characteristic extraction module is used for extracting at least one biological behavior characteristic from the track verification code drawing data when the natural person drawing index value is determined to meet the natural person operation primary screening condition;

and the natural person operation verification module is used for verifying whether the track verification code drawing data are generated by natural person operation according to the biological behavior characteristics.

According to another aspect of the present disclosure, there is also provided an electronic apparatus including:

at least one processor; and

a memory communicatively coupled to the at least one processor; wherein,,

the memory stores instructions executable by the at least one processor to enable the at least one processor to perform a method of identifying a trajectory verification code as in any one of the embodiments of the present disclosure.

According to another aspect of the present disclosure, there is also provided a non-transitory computer-readable storage medium storing computer instructions for causing a computer to perform the method of identifying a trajectory verification code according to any one of the embodiments of the present disclosure.

It should be understood that the description in this section is not intended to identify key or critical features of the embodiments of the disclosure, nor is it intended to be used to limit the scope of the disclosure. Other features of the present disclosure will become apparent from the following specification.

Drawings

The drawings are for a better understanding of the present solution and are not to be construed as limiting the present disclosure. Wherein:

fig. 1 is a schematic diagram of a method for identifying a track verification code according to an embodiment of the disclosure;

FIG. 2 is a schematic diagram of a captcha input interface provided in accordance with an embodiment of the present disclosure;

FIG. 3 is a schematic diagram of another method for identifying a track verification code according to an embodiment of the present disclosure;

FIG. 4 is a schematic diagram of a method of identifying a trajectory verification code provided in accordance with an embodiment of the present disclosure;

FIG. 5 is a block diagram of an identification device for track verification codes according to an embodiment of the present disclosure;

fig. 6 is a block diagram of an electronic device for implementing a method of identifying a track verification code in accordance with an embodiment of the present disclosure.

Detailed Description

Exemplary embodiments of the present disclosure are described below in conjunction with the accompanying drawings, which include various details of the embodiments of the present disclosure to facilitate understanding, and should be considered as merely exemplary. Accordingly, one of ordinary skill in the art will recognize that various changes and modifications of the embodiments described herein can be made without departing from the scope and spirit of the present disclosure. Also, descriptions of well-known functions and constructions are omitted in the following description for clarity and conciseness.

Fig. 1 is a flowchart of a method for identifying a track verification code according to an embodiment of the present disclosure. The method and the device can be applied to the situation that whether the track verification code is generated by natural human operation or not is effectively recognized. The method may be performed by a track verification code recognition device, which may be implemented in hardware and/or software, and may be typically integrated in a server.

Correspondingly, as shown in fig. 1, the method for identifying the track verification code provided by the embodiment of the disclosure includes the following specific steps:

s110, acquiring track verification code drawing data, and acquiring a natural person drawing index value according to the track verification code drawing data.

In an optional application scenario of this embodiment, after the user (actual human or robot-simulated human) completes the account login operation for the set website, the user triggers the verification code input mechanism. At this time, the website server provides a track verification code input interface for the user, and the user inputs track verification code drawing data to perform natural human operation or robot operation detection.

The track verification code input interface is used for indicating a user to input a curve of a set track. The track verification code drawing data are drawing data formed when a user draws the curve in a mode of mouse dragging click or touch screen sliding touch.

Specifically, the track verification code drawing data may include coordinates of each track point in the curve track, drawing time of the coordinates of each track point, and the like. Further, the track verification code drawing data may further include motion description information, for example, if the user draws the curve by means of dragging and clicking with a mouse, the motion description information may further include speed information, acceleration information and the like of the mouse when drawing each track point. For another example, if the user draws the curve by sliding the touch screen, the motion description information may further include a touch pressure when each track point is generated by touching, which is not limited in this embodiment.

The natural person drawing index value is an attribute value which is inherent to the natural person in the curve drawing process and can be rapidly distinguished from the robot. In a specific example, the natural person drawing index value may be a curve drawing speed, a distribution of touch pressure when each locus point is generated by touching, a left hand or right hand attribute which is possessed when drawing a curve, a tendency to start drawing from the upper part or from the lower part when drawing a curve, a tendency to start drawing from the left side or from the right side, or the like. It will be understood that, by deep mining of various human behavior characteristics when a natural person draws a curve, a person skilled in the art may select the natural person drawing index value, which is not limited in this embodiment.

In this embodiment, one or more natural person drawing index values may be directly extracted from the trajectory verification code drawing data, and one or more natural person drawing index values may also be calculated according to the trajectory verification code drawing data and the trajectory shape and distribution of the curve to be drawn by the user.

S120, when the natural person drawing index value meets the natural person operation preliminary screening condition, extracting at least one biological behavior characteristic from the track verification code drawing data.

In this embodiment, in order to efficiently identify the track verification code, a two-stage identification process is introduced. The first stage is based on the natural person drawing index value that the data acquisition was drawn to the orbit verification code for carry out the preliminary screening of natural person operation, only just after the preliminary screening passes, can carry out second stage identification process, the second stage is based on draw at least one biological behavior characteristic in the data is drawn to the orbit verification code, is used for carrying out the accurate discernment of natural person operation.

It will be appreciated that the complexity of the second stage identification may be much higher than the first stage identification.

The natural person operation preliminary screening condition can be understood as an index value range interval in which the index value is supposed to fall when the track verification code drawing data is generated by natural person operation. The natural person operating prescreening condition may be generated by collecting historical trajectory operation data of one or more natural persons, or may be preset based on actual experience, which is not limited in this embodiment.

When the natural person drawing index value meets the natural person operation preliminary screening condition, the condition that the trajectory verification code drawing data are generated for the robot simulation cannot be clearly eliminated through simple one-stage identification is explained, and at the moment, the biological behavior characteristics in the trajectory verification code drawing data need to be continuously extracted for two-stage identification.

It can be understood that if it is determined that the natural person drawing index value does not satisfy the natural person operation primary screening condition, it may be directly determined that the trajectory verification code drawing data is not generated by the natural person operation, and the verification code verification process of this time is not passed.

Among these, biological behavioral characteristics can be understood as behavioral characteristics that can effectively distinguish between natural human and robotic operations. The biological behavior feature can be an explicit feature directly extracted from the trace verification code drawing data, or can be an implicit feature obtained by processing the trace verification code drawing data at least once.

And S130, verifying whether the track verification code drawing data are generated by natural human operation according to the biological behavior characteristics.

In this embodiment, the one or more biological behavior features may be input into a pre-trained neural network model, and the neural network model outputs a recognition result that is generated by whether the trajectory verification code drawing data is natural human operation; or, the one or more biological behavior characteristics can be input into one or more pre-built fitting formulas, and whether the trajectory verification code drawing data are generated by natural human operation or not is verified according to the calculation result of the one or more fitting formulas; or, the one or more biological behavior features may be input to an expert rule module, and the expert rule module outputs an identification result that whether the trajectory verification code drawing data is generated by natural human operation, or may verify whether the trajectory verification code drawing data is generated by natural human operation according to an integrated identification result of the centralized verification manner, or the like, which is not limited in this embodiment.

Similarly, if the track verification code drawing data is verified to be generated by natural human operation according to the biological behavior characteristics, the verification code verification process is not passed.

According to the technical scheme, the natural person drawing index value is obtained by obtaining the track verification code drawing data and according to the track verification code drawing data; when the natural person drawing index value meets the natural person operation preliminary screening condition, extracting at least one biological behavior characteristic from the track verification code drawing data; according to the biological behavior characteristics, the technical means of verifying whether the track verification code drawing data are generated by natural human operation is introduced, a two-stage verification code identification scheme with different implementation complexity is introduced, the verification code input by the natural human operation and the verification code simulated by the robot can be accurately and efficiently distinguished, the accuracy of distinguishing the track verification code is improved, and the safety of the network environment protected by the verification behavior is ensured.

On the basis of the above embodiments, obtaining the track verification code drawing data may include:

responding to a verification code acquisition request, and providing a verification code input interface, wherein the verification code input interface comprises a background image and a track guide curve attached to the background image; and acquiring the track verification code drawing data through the verification code input interface.

Alternatively, a schematic diagram of an alternative verification code input interface is shown in FIG. 2. As shown in FIG. 2, the verification code input interface includes a background image and a track guiding curve, and simultaneously has guiding text for prompting a user to draw a track verification code according to the track guiding curve. And when the user draws the curve along the track guiding curve in the verification code input interface, track verification code drawing data can be correspondingly generated.

It should be noted that the track verification code proposed in this alternative embodiment is a new type of verification code, different from the existing sliding verification code.

In this alternative embodiment, a new track verification code is proposed, which is no longer a horizontal, vertical input sliding track, but a curve of various unpredictable types. Through the arrangement, the difficulty of generating the track verification codes by the robot simulation is greatly increased, and furthermore, the index values are drawn for the natural people acquired by the track verification codes, so that the natural people or the robot can be more effectively distinguished.

Based on the above embodiments, the obtaining the natural person drawing index value according to the trajectory verification code drawing data may include at least one of the following:

Determining a track drawing speed according to the curve length of the track guiding curve and the track drawing time extracted from the track verification code drawing data;

extracting the drawing force of a set drawing position from the drawing data of the track verification code;

and determining the direction tendency according to the space distribution type of the track guiding curve and the track drawing direction extracted from the track verification code drawing data.

It will be appreciated that a natural person should have a track drawing speed within a reasonable data range when drawing a track verification code with reference to the track guide curve, and that the drawing speed will also vary from person to person. Therefore, the trajectory drawing speed can be taken as a natural person drawing index value.

Correspondingly, the number of pixels occupied by the track guiding curve can be obtained in the verification code input interface, and the curve length of the track guiding curve can be calculated based on the preset pixel size. And then, the track drawing time can be conveniently calculated by acquiring the drawing time of the initial drawing point and the drawing time of the final drawing point from the track verification code drawing data, and the result of dividing the curve length by the track drawing time can be calculated as the track drawing speed.

Similarly, when a natural person draws the track verification code in a sliding touch manner of the touch screen by referring to the track guide curve, the drawing force of the natural person should be within a reasonable data range, and the drawing force is different from person to person. Therefore, the drawing force can also be used as a natural person drawing index value.

Specifically, when the user generates track verification code drawing data in a sliding touch manner through the touch screen, the touch screen can acquire touch pressure values at one or more touch positions through the pressure sensor array. Correspondingly, the drawing force of the set drawing position can be extracted from the collection value of the pressure sensor array recorded in the track verification code drawing data. The set position may be a position where a start-stop drawing point is located, a position where a turning point of a drawing track is located, and the like.

In this alternative embodiment, the specific drawing position is selected, mainly considering that different natural people have different touch habits, the drawing force at the specific drawing position is generally different, for example, the position of the initial drawing point is also different. Therefore, different natural person operation primary screening conditions can be further set for different natural persons, so that the identification accuracy of the track verification code is further improved.

Further, when different natural persons draw track guiding curves with different spatial distribution types, the track drawing directions adopted by the different natural persons are generally different. The spatial distribution type may include: horizontal large-span distribution, vertical large-span distribution, and horizontal and vertical large-span distribution.

In a specific example, if the span (X-axis coordinate difference) of the start point and the end point in the horizontal direction in one trajectory guide curve is far greater than the span (Y-axis coordinate difference) in the vertical direction, the spatial distribution type of the trajectory guide curve is determined to be a horizontal large-span distribution; if the spans of the starting point and the ending point in the track guiding curve in the horizontal direction are far smaller than the spans in the vertical direction, determining that the spatial distribution type of the track guiding curve is vertical large-span distribution; if the horizontal spans of the start point and the end point in one track guiding curve are similar to the vertical spans, the spatial distribution type of the track guiding curve is determined to be horizontal and vertical large-span distributions.

As described above, when different natural persons draw track guiding curves of different spatial distribution types, the track drawing directions adopted by the different natural persons are generally different, for example, for the spatial distribution types distributed in a large span in the horizontal direction, some natural persons tend to draw from left to right, and some natural persons tend to draw from right to left; alternatively, for the spatial distribution type of the vertical large-span distribution, some natural persons tend to draw from top to bottom, and some natural persons tend to draw from bottom to top. Based on this, the rendering direction tendency can be regarded as a natural person rendering index value that varies from person to person.

Through the arrangement, the attribute values which can be used for obviously distinguishing the natural person from the robot can be effectively mined and used as the drawing index values of the natural person. And further, the effective execution of the primary screening process of natural person operation can be ensured in the first stage identification based on the natural person drawing index value obtained by the track verification code drawing data. The method can rapidly and conveniently screen out the track verification code drawing data which obviously does not meet the condition of natural human operation primary screening, and effectively reduce the identification time of the track verification code while ensuring the identification accuracy of the track verification code.

Fig. 3 is a flowchart of another method for identifying a track verification code according to an embodiment of the present disclosure. The present embodiment is refined based on the above embodiments. In this embodiment, the operation of determining that the natural person drawing index value satisfies the natural person operation preliminary screening condition is refined.

Correspondingly, as shown in fig. 3, the method for identifying the track verification code provided by the embodiment of the disclosure includes the following specific steps:

s310, acquiring track verification code drawing data, and acquiring a natural person drawing index value according to the track verification code drawing data.

S320, acquiring the identity of the client providing the track verification code drawing data.

The identity of the client may be understood as identification information for uniquely identifying the identity of the track verification code drawing data provider. For example, the identity of the client may be login information used when the user logs in to the website server through the website client, or may be a device identity of a terminal device installed on the website client.

S330, detecting whether historical operation track data matched with the identity is stored or not: if yes, executing S340; otherwise, S350 is performed.

The history operation trajectory data may be understood as trajectory data formed when the user performs the screen operation by setting the client. The historical operation track data can comprise various data generated correspondingly when the mouse is dragged and clicked, various data generated correspondingly when the touch screen is slid and touched, and the like.

In this embodiment, the server may periodically store historical operation track data of each login client for a last period of time (for example, 7 days, 2 weeks, or one month, etc.), and accurately describe track drawing habits of different natural persons through the historical operation track data.

S340, generating a target drawing index value range matched with the natural person drawing index value according to the historical operation track data, and executing S360.

In this embodiment, different types of target drawing index value ranges may be generated correspondingly using locally stored historical operation trajectory data for different types of natural person drawing index values.

As described above, the natural person drawing index value may include a trajectory drawing speed, a drawing force for setting a drawing position, and a drawing direction tendency. Correspondingly, after each data item for calculating the drawing index value of each natural person is extracted from the historical operation track data, the average trend value of the drawing index value of each type of natural person is calculated based on each data item, and a target drawing index value range matched with each natural person drawing index value respectively is obtained by setting a preset data fluctuation range.

S350, inquiring the crowd attribute label matched with the identity, and executing S370.

In this embodiment, if the historical operation track data matched with the identity is not stored, the historical operation track data of other clients similar to the user attribute of the client may be considered to determine the target drawing index value range respectively matched with each natural person drawing index value.

Among other things, crowd attribute tags may be understood as tags that describe common attributes of a class of people. In a specific example, the crowd attribute tag may be: "30-40 years old", "male" and "software developer", etc.

And S360, when the natural person drawing index value is verified to be in the target drawing index value range, determining that the natural person drawing index value meets a natural person operation primary screening condition, and executing S3100.

And S370, obtaining similar historical operation track data of a plurality of similar clients matched with the clients according to the crowd attribute tags.

After the crowd attribute tag of the client is obtained, even if the historical operation track data of the client cannot be obtained, the similar historical operation track data of a plurality of similar clients matched with the client can be obtained to serve as approximate substitutes of the historical operation track data of the client.

It should be noted that, in the embodiments of the present disclosure, corresponding historical operation track data is obtained only after the authorization of the user is obtained in a reasonable and legal manner. Meanwhile, the data required to be collected when the crowd attribute labels are established for all the clients are also obtained by reasonable and legal modes and after the authorization of the user.

S380, generating a reference drawing index value range matched with the natural person drawing index value according to the similar historical operation track data.

S390, when the natural person drawing index value is verified to be in the reference drawing index value range, determining that the natural person drawing index value meets a natural person operation primary screening condition, and executing S3100.

S3100, extracting at least one biological behavior characteristic from the track verification code drawing data.

S3110, according to the biological behavior characteristics, verifying whether the track verification code drawing data are generated by natural human operation.

According to the technical scheme, the historical operation track data of the client providing the track verification code drawing data or the historical operation track data of the similar clients with the same attributes as the client population are used for generating the target drawing index value range matched with the natural person drawing index value to judge the natural person operation primary screening condition, so that the track verification code drawing data which obviously do not meet the natural person operation primary screening condition can be accurately screened out, the identification accuracy of the track verification code is ensured, and the identification time of the track verification code is further reduced.

On the basis of the above embodiments, the method may further include: if the track verification code drawing data does not meet the natural human operation preliminary screening condition or the input track data is verified to be generated for natural human operation according to the biological behavior characteristics, adding a robot tendency identification for a client providing the track verification code drawing data;

intercepting a target verification code input interface to be provided for a target client in response to a verification code acquisition request of the target client added with the robot tendency identification;

detecting the number of turning points of a track guiding curve in the target verification code input interface;

if the number of turning points does not exceed the preset number threshold, requesting to generate a new target verification code input interface until the number of turning points of the track guide curve in the new target verification code input interface exceeds the number threshold;

and providing the new target verification code input interface for the target client to input the track verification code.

In this alternative embodiment, once a specific client is detected to have entered the track verification code by way of robotic simulation, a robotic propensity identification is added to the client. Correspondingly, when identifying the verification code, the embodiments of the disclosure can be divided into two cases, and one case is that the client side sending the verification code obtaining request does not carry the robot tendency identification, and the track verification code identification can be performed on the track verification code drawing data provided by the client side according to the implementation manner of the embodiments. Another is that the client sending the verification code acquisition request (i.e., the target client described above) carries the robot-tendency identification, which indicates that the target client has attempted to crack the track verification code. At this time, a more complex track guiding curve needs to be provided for the target client, so that the purpose of the setting is that when a natural person inputs a complex track verification code, the more obvious the characteristic of the nature attribute carried by the inputted track verification code drawing data is, and further, the higher the distinction between the natural person drawing index value and the biological behavior characteristic obtained in the track verification code drawing data and the track verification code drawing data simulated by the robot is, and further, the higher the track verification code identification accuracy is, and the misjudgment rate can be correspondingly reduced.

In this optional embodiment, the number of turning points included in the track guiding curve is used as a constraint condition for the complexity of the track guiding curve, and only when the number of turning points included in the track guiding curve exceeds a preset number threshold, the track guiding curve is provided to the target client for inputting the track verification code drawing data.

Of course, it will be appreciated that the complexity of the trajectory guidance curve may be required by other constraints, such as the length of the curve, the horizontal span or the vertical span of the curve, or a combination of the above constraints, which is not limited by the present embodiment.

Fig. 4 is a flowchart of a method for identifying a trajectory verification code according to an embodiment of the present disclosure. The present embodiment is refined based on the above embodiments. In this embodiment, the operations of extracting at least one biological behavior feature from the trajectory verification code drawing data and verifying whether the trajectory verification code drawing data is generated by natural human operation according to the biological behavior feature are refined.

Correspondingly, as shown in fig. 4, the method for identifying the track verification code provided by the embodiment of the disclosure includes the following specific steps:

S410, acquiring track verification code drawing data, and acquiring a natural person drawing index value according to the track verification code drawing data.

S420, splitting the track verification code drawing data into a plurality of data sequences of data types according to the data types of all data items in the track verification code drawing data when the natural person drawing index value meets the natural person operation preliminary screening condition.

Wherein the data types include: coordinate values of coordinate axes are set in the position coordinate system, and coordinate values of motion parameters are set in the motion coordinate system. Specifically, the data types may include: x-axis coordinates, Y-axis coordinates, X-axis acceleration, Y-axis acceleration, Z-axis acceleration, X-axis angular acceleration, Y-axis angular acceleration, Z-axis angular acceleration, and the like.

In a specific example, the track verification code drawing data includes: mouse track data of a user, data collected by an acceleration sensor on the mouse and data collected by a gyroscope sensor on the mouse.

The mouse track data format of the user is [ [ xu1, yu1, tu1], [ xu2, yu2, tu2], … ], wherein: xu is the abscissa of the track point; yu is the ordinate of the track point; tu is the trace point timestamp.

The data format reported by the acceleration sensor of the user is [ [ ax1, ay1, az1, t1], [ ax2, ay2, az2, t2], … ], wherein: ax1 is the acceleration of the x axis of the recording point; ay1 is the acceleration of the y axis of the recording point; az1 is the recorded point z-axis acceleration; t1 is a record point timestamp.

The data format reported by the user's gyro sensor is [ [ wx1, wy1, wz1, t1], [ wx2, wy2, wz2, t2], … ], wherein: wx1 is the angular acceleration of the x-axis of the recording point, wy1 is the angular acceleration of the y-axis of the recording point; wz1 is the angular acceleration of the z-axis of the recording point; t1 is a record point timestamp.

And drawing the data types of all data items in the data according to the track verification code, and sequencing the data types and the time stamps to obtain a plurality of data sequences with the following types:

x= { xu1, xu2, … }, sequence y= { yu1, yu2, … }, ax= { AX1, AX2, … }, … wz= { WZ1, WZ2 … }.

S430, performing first-order differential processing and second-order differential processing on the data sequences of each data type respectively to obtain first-order differential data sequences and second-order differential data sequences of different data types.

S440, respectively extracting at least one statistical feature from each data sequence, each first-order differential data sequence and each second-order differential data sequence.

Wherein the statistical characteristic comprises at least one of: maximum, minimum, average, standard deviation, peak and deviation.

Through the arrangement, various biological behavior characteristics closely related to the natural person attribute are deeply excavated, and effective data guarantee is provided for the follow-up accurate track verification code identification.

S450, determining the data sequence, the first-order differential data sequence, the second-order differential data sequence and the at least one statistical characteristic as the biological behavior characteristic.

By extracting the original sequence and the first-order and second-order differential sequences of the trajectory data and calculating the statistical features, feature vectors reflecting the biological behaviors of the user can be obtained.

S460, inputting the biological behavior characteristics into a natural human operation recognition model, and acquiring whether the track verification code drawing data is a first recognition result generated by natural human operation or not.

Optionally, inputting the biological behavior feature into a natural person operation recognition model, and obtaining whether the trajectory verification code drawing data is a first recognition result generated by natural person operation may include:

respectively inputting the biological behavior characteristics into a first sub-model and a second sub-model in the natural human operation recognition model;

The first sub-model is constructed based on a gradient lifting tree algorithm, and the second sub-model is constructed based on a gradient lifting decision tree algorithm;

generating a first classification result matched with the biological behavior characteristics through the first sub-model, and inputting the first classification result into a weighted integration sub-model in the natural human operation recognition model;

outputting a second classification result matched with the biological behavior characteristics through the second sub-model, and inputting the first classification result into the weighted integration sub-model in the natural human operation recognition model;

and carrying out weighting processing on the first classification result and the second classification result through the weighting integration submodel, and outputting a first recognition result whether the track verification code drawing data are generated by natural human operation or not.

In embodiments of the present disclosure, machine learning models may be employed to classify and predict extracted features. The training data set comprises track data collected by a real user and track data collected by the simulation robot, and is accurately marked. By training the machine learning model using this training dataset, real user and robot behavior can be accurately identified and distinguished.

In the disclosed embodiments, XGBoost and LightGBM are specifically selected as machine learning models to classify and predict extracted features.

XGBoost model: XGBoost (eXtreme Gradient Boosting, limit gradient boosting) is a machine learning algorithm based on gradient boosting trees. It improves the performance of the overall model by iteratively training a plurality of weak classifiers, each iteration optimizing a loss function. The present embodiment will use the XGBoost model to build a biological behavior recognition classifier.

LightGBM model: the LightGBM is a machine learning algorithm based on a gradient lifting decision tree, and has high efficiency performance and low memory consumption. The method adopts a decision tree algorithm based on a histogram and a splitting mode based on Leaf-wise, and can dynamically select optimal characteristics and splitting points in the training process. The present embodiment will use the LightGBM model as an alternative to biological behavior recognition.

The two models are trained on feature vectors, and learn how to accurately classify natural human and robot behaviors by optimizing objective functions. Finally, a weighted average method is used for synthesizing the prediction results of the two models, and an integrated result, namely a first recognition result, can be efficiently and accurately output.

Based on the above embodiments, the natural human operation recognition model is obtained by training in a cold start offline training combined with on-line timing increment updating. That is, training of the natural person operation recognition model is divided into two stages: cold start offline training and online timing delta training updates.

1. Cold start offline training:

the cold start offline training phase aims to build an initial natural human operation recognition model. This stage involves the following steps:

1) And (3) data collection: biological behavior data generated by real users and machine scripts are collected manually. These data are used to construct training and validation sets.

2) Data preprocessing: preprocessing the collected biological behavior data, marking the wrong sample, and collecting the wrong sample for removal.

3) Characteristic engineering: the calculation of the biological behavior characteristics as described in sections S410-S450 is performed on the preprocessed data.

4) Model selection and training: in the cold start offline training phase, appropriate machine learning models (e.g., XGBoost and LightGBM) are selected and model training is performed using a training set. The performance of the model is optimized by adjusting the super parameters of the model and adopting the technical means of cross verification and the like.

5) Model evaluation: and evaluating the trained natural human operation recognition model by using a verification set, wherein the indexes comprise an accuracy rate, a recall rate, an F1 value and the like, so as to ensure the performance and generalization capability of the natural human operation recognition model.

6) Model preservation: the trained natural human operation recognition model is saved for subsequent online application and timing incremental training updates.

2. On-line timed incremental training update:

to track changes in user behavior and accommodate new attack patterns, embodiments of the present disclosure introduce a mechanism for online timing delta training updates. The mechanism involves the following steps:

1) And (3) data acquisition: new training samples are periodically collected from the biological behavior characteristics of an online natural person (real user), and include mouse track data, sensor data and verification of whether the training samples pass or not. These samples are used for incremental training updates.

2) Resampling data: in order to prevent the problem of model overfitting caused by uneven distribution of the online data labels (imbalance of the proportion of verification code passing and verification code not passing), resampling the newly added samples ensures that the proportion of the samples passing verification and the samples not passing verification in the sampled training samples is 1:1.

3) Characteristic engineering: and performing feature engineering on the preprocessed incremental training data to extract the most distinguishable and representative features. This step may employ the same feature engineering method as the cold start offline training phase.

4) Model updating: the incremental training data is used to update an existing natural person operation recognition model. The updating method can adopt an online learning algorithm, such as incremental training, online gradient descent and the like. By updating the model with incremental data, the natural human operation recognition model can be adapted to changes in user behavior and new attack patterns.

5) Model evaluation: the updated model is evaluated using the validation set to ensure that the performance and generalization ability of the natural human operational recognition model remain within acceptable limits. If the evaluation result is not ideal, parameters of the model can be adjusted or other strategies can be adopted for further optimization.

6) Model preservation: the updated natural human operation recognition model is saved for the next online application and timing increment training update.

Through the combination of cold start offline training and online timing increment training updating, the natural human operation recognition model of the embodiment of the disclosure can continuously learn and adapt to the change of user behaviors, and the accuracy and the safety of a verification code system are improved.

S470, inputting the biological behavior characteristics into an expert rule module, and acquiring whether the track verification code drawing data is a second recognition result generated by natural human operation.

In this embodiment, in addition to the first recognition result obtained by using the natural human operation recognition model, an expert rule module is further introduced, and a natural human operation judgment result for the biological behavior feature is output in parallel with the natural human operation recognition model.

Among the advantages of expert rules are:

the interpretation is strong: the expert rules module is capable of formulating a series of rules based on the knowledge and experience of the domain expert, which rules can be interpreted and understood. In contrast, the natural human operational recognition model may be a black box model, whose internal decision process is difficult to interpret.

Knowledge control right: expert rules modules allow domain experts to directly participate in formulating rules, which they can formulate based on their own knowledge and judgment. This makes the rule formulation process more flexible, and can be adjusted and optimized as needed.

Prediction accuracy: expert rules can accurately judge the abnormal behavior of the trace organism under specific situations by using experience and knowledge of field experts. In some fields or particular scenarios, expert rules may be more accurate than natural human operation recognition models.

Expert rules are configured in a policy engine of the verification code through rule expressions, support conventional logic mixing, including and, or, greater than, less than, equal to, greater than, equal to, and less than or equal to.

S480, inputting the first identification result and the second identification result into a voting model, and acquiring whether the track verification code drawing data is an integration result generated by natural human operation.

In this embodiment, the accuracy and reliability of identifying the track verification code are improved by combining machine learning and expert rules and introducing an integrated learning mode.

Specifically, a voting model is designed, the recognition result of a natural human operation recognition model and the recognition result of an expert rule module are taken as input x, and whether the track verification code passes the verification is taken as input y. By training these input data we can build a voting model that can integrate the decisions of the various modules.

In this embodiment, an LR (linear regression ) L2-norm model may be selected as the basis for the voting model. This is because regularization of L2-norm ensures that the weights of the individual modules are distributed as evenly as possible, thereby preventing excessive weight bias from causing the output of a certain model to be completely ineffective. In addition, the linear regression model is adopted to effectively model the relation between the input and the output, so that the influence degree of each module on the final judgment result can be better understood.

The technical scheme of the embodiment of the disclosure can fully exert the advantages of machine learning and expert rules through an integrated learning method, and optimize the conclusion output of the machine learning and expert rules. The method has good robustness and can cope with the change of different scenes and data, thereby improving the accuracy and reliability of verification code judgment.

Meanwhile, the identification method of the track verification code provided by the embodiments of the present disclosure has no direct use experience change for all verification code users. For normal users, they can still easily pass the verification of the track verification code without additional operations or inputs. However, for users who use automatic scripts to cheat, even though they correctly recognize and draw tracks through the scripts, the two-stage track verification recognition mechanism can timely find anomalies of their biological behaviors, so that the users are prevented from bypassing the track verification code, and the overall protection capability of the track verification code is effectively improved.

As an implementation of the above method for identifying each track verification code, the present disclosure further provides an optional embodiment of an execution device for implementing the above method for identifying each track verification code.

Fig. 5 is a block diagram of an identification device for a track verification code according to an embodiment of the present disclosure. As shown in fig. 5, the apparatus includes: an index value acquisition module 510, a biometric feature extraction module 520, and a natural person operation verification module 530.

The index value obtaining module 510 is configured to obtain trajectory verification code drawing data, and obtain a natural person drawing index value according to the trajectory verification code drawing data;

the biological feature extraction module 520 is configured to extract at least one biological behavior feature from the trajectory verification code drawing data when it is determined that the natural person drawing index value meets a natural person operation preliminary screening condition;

and a natural person operation verification module 530, configured to verify whether the trajectory verification code drawing data is generated by natural person operation according to the biological behavior feature.

Based on the foregoing embodiments, the index value obtaining module is specifically configured to:

responding to a verification code acquisition request, and providing a verification code input interface, wherein the verification code input interface comprises a background image and a track guide curve attached to the background image;

and acquiring the track verification code drawing data through the verification code input interface.

On the basis of the above embodiments, the index value obtaining module is further configured to perform at least one of the following:

On the basis of the above embodiments, the biometric extraction module is specifically configured to:

acquiring an identity of a client providing the track verification code drawing data;

detecting whether historical operation track data matched with the identity mark is stored or not;

If yes, generating a target drawing index value range matched with the natural person drawing index value according to the historical operation track data;

and when the natural person drawing index value is verified to be in the target drawing index value range, determining that the natural person drawing index value meets a natural person operation primary screening condition.

On the basis of the above embodiments, the biometric extraction module is further specifically configured to:

after detecting whether historical operation track data matched with the identity is stored or not, if not, inquiring a crowd attribute tag matched with the identity;

obtaining similar historical operation track data of a plurality of similar clients matched with the clients according to the crowd attribute tags;

generating a reference drawing index value range matched with the natural person drawing index value according to the similar historical operation track data;

and when the natural person drawing index value is verified to be in the range of the reference drawing index value, determining that the natural person drawing index value meets a natural person operation primary screening condition.

On the basis of the above embodiments, the biological feature extraction module is further configured to:

splitting the track verification code drawing data into a plurality of data sequences of data types according to the data types of all data items in the track verification code drawing data;

Wherein the data types include: coordinate values of coordinate axes are set under a position coordinate system, and coordinate values of motion parameters are set in the coordinate axes are set under a motion coordinate system;

respectively carrying out first-order differential processing and second-order differential processing on the data sequences of each data type to obtain first-order differential data sequences and second-order differential data sequences of different data types;

respectively extracting at least one statistical feature from each data sequence, each first-order differential data sequence and each second-order differential data sequence;

determining the data sequence, the first-order differential data sequence, the second-order differential data sequence, and the at least one statistical feature as the biological behavior feature.

On the basis of the above embodiments, the natural person operating the verification module includes:

the first recognition result acquisition unit is used for inputting the biological behavior characteristics into a natural person operation recognition model and acquiring whether the track verification code drawing data is a first recognition result generated by natural person operation or not;

the second recognition result acquisition unit is used for inputting the biological behavior characteristics into the expert rule module and acquiring whether the track verification code drawing data are second recognition results generated by natural human operation or not;

And the integrated result acquisition unit is used for inputting the first identification result and the second identification result into a voting model and acquiring whether the track verification code drawing data is an integrated result generated by natural human operation.

On the basis of the foregoing embodiments, the first recognition result obtaining unit is specifically configured to:

Based on the above embodiments, the natural human operation recognition model is obtained by training in a mode of combining cold start offline training and on-line timing increment updating.

On the basis of the above embodiments, the robot-tendency client processing module is further configured to:

if the track verification code drawing data does not meet the natural human operation preliminary screening condition or the input track data is verified to be generated for natural human operation according to the biological behavior characteristics, adding a robot tendency identification for a client providing the track verification code drawing data;

The product can execute the method provided by any embodiment of the disclosure, and has the corresponding functional modules and beneficial effects of executing the method.

In the technical scheme of the disclosure, the related processes of collecting, storing, using, processing, transmitting, providing, disclosing and the like of the personal information of the user accord with the regulations of related laws and regulations, and the public order colloquial is not violated.

According to embodiments of the present disclosure, the present disclosure also provides an electronic device, a readable storage medium and a computer program product.

Fig. 6 illustrates a schematic block diagram of an example electronic device 600 that may be used to implement embodiments of the present disclosure. Electronic devices are intended to represent various forms of digital computers, such as laptops, desktops, workstations, personal digital assistants, servers, blade servers, mainframes, and other appropriate computers. The electronic device may also represent various forms of mobile devices, such as personal digital processing, cellular telephones, smartphones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions, are meant to be exemplary only, and are not meant to limit implementations of the disclosure described and/or claimed herein.

As shown in fig. 6, the apparatus 600 includes a computing unit 601 that can perform various appropriate actions and processes according to a computer program stored in a Read Only Memory (ROM) 602 or a computer program loaded from a storage unit 608 into a Random Access Memory (RAM) 603. In the RAM 603, various programs and data required for the operation of the device 600 may also be stored. The computing unit 601, ROM 602, and RAM 603 are connected to each other by a bus 604. An input/output (I/O) interface 605 is also connected to bus 604.

Various components in the device 600 are connected to the I/O interface 605, including: an input unit 606 such as a keyboard, mouse, etc.; an output unit 607 such as various types of displays, speakers, and the like; a storage unit 608, such as a magnetic disk, optical disk, or the like; and a communication unit 609 such as a network card, modem, wireless communication transceiver, etc. The communication unit 609 allows the device 600 to exchange information/data with other devices via a computer network, such as the internet, and/or various telecommunication networks.

The computing unit 601 may be a variety of general and/or special purpose processing components having processing and computing capabilities. Some examples of computing unit 601 include, but are not limited to, a Central Processing Unit (CPU), a Graphics Processing Unit (GPU), various specialized Artificial Intelligence (AI) computing chips, various computing units running machine learning model algorithms, a Digital Signal Processor (DSP), and any suitable processor, controller, microcontroller, etc. The computing unit 601 performs the respective methods and processes described above, for example, the identification method of the trajectory verification code described in the embodiments of the present disclosure. For example, in some embodiments, the method of identifying a track verification code described by embodiments of the disclosure may be implemented as a computer software program tangibly embodied on a machine-readable medium, such as storage unit 608. In some embodiments, part or all of the computer program may be loaded and/or installed onto the device 600 via the ROM 602 and/or the communication unit 609. When the computer program is loaded into the RAM 603 and executed by the computing unit 601, one or more steps of the track verification code identification method described in the embodiments of the present disclosure described above may be performed. Alternatively, in other embodiments, the computing unit 601 may be configured to perform the track verification code identification method described by embodiments of the present disclosure in any other suitable manner (e.g., by means of firmware).

Namely: acquiring track verification code drawing data, and acquiring a natural person drawing index value according to the track verification code drawing data;

Various implementations of the systems and techniques described here above can be implemented in digital electronic circuitry, integrated circuit systems, field Programmable Gate Arrays (FPGAs), application Specific Integrated Circuits (ASICs), application Specific Standard Products (ASSPs), systems On Chip (SOCs), complex Programmable Logic Devices (CPLDs), computer hardware, firmware, software, and/or combinations thereof. These various embodiments may include: implemented in one or more computer programs, the one or more computer programs may be executed and/or interpreted on a programmable system including at least one programmable processor, which may be a special purpose or general-purpose programmable processor, that may receive data and instructions from, and transmit data and instructions to, a storage system, at least one input device, and at least one output device.

Program code for carrying out methods of the present disclosure may be written in any combination of one or more programming languages. These program code may be provided to a processor or controller of a general purpose computer, special purpose computer, or other programmable data processing apparatus such that the program code, when executed by the processor or controller, causes the functions/operations specified in the flowchart and/or block diagram to be implemented. The program code may execute entirely on the machine, partly on the machine, as a stand-alone software package, partly on the machine and partly on a remote machine or entirely on the remote machine or server.

In the context of this disclosure, a machine-readable medium may be a tangible medium that can contain, or store a program for use by or in connection with an instruction execution system, apparatus, or device. The machine-readable medium may be a machine-readable signal medium or a machine-readable storage medium. The machine-readable medium may include, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any suitable combination of the foregoing. More specific examples of a machine-readable storage medium would include an electrical connection based on one or more wires, a portable computer diskette, a hard disk, a Random Access Memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

To provide for interaction with a user, the systems and techniques described here can be implemented on a computer having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to a user; and a keyboard and pointing device (e.g., a mouse or trackball) by which a user can provide input to the computer. Other kinds of devices may also be used to provide for interaction with a user; for example, feedback provided to the user may be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user may be received in any form, including acoustic input, speech input, or tactile input.

The systems and techniques described here can be implemented in a computing system that includes a background component (e.g., as a data server), or that includes a middleware component (e.g., an application server), or that includes a front-end component (e.g., a user computer having a graphical user interface or a web browser through which a user can interact with an implementation of the systems and techniques described here), or any combination of such background, middleware, or front-end components. The components of the system can be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include: local Area Networks (LANs), wide Area Networks (WANs), and the internet.

The computer system may include a client and a server. The client and server are typically remote from each other and typically interact through a communication network. The relationship of client and server arises by virtue of computer programs running on the respective computers and having a client-server relationship to each other. The server can be a cloud server, also called a cloud computing server or a cloud host, and is a host product in a cloud computing service system, so that the defects of high management difficulty and weak service expansibility in the traditional physical hosts and VPS service are overcome. The server may also be a server of a distributed system or a server that incorporates a blockchain.

Artificial intelligence is the discipline of studying the process of making a computer mimic certain mental processes and intelligent behaviors (e.g., learning, reasoning, thinking, planning, etc.) of a person, both hardware-level and software-level techniques. Artificial intelligence hardware technologies generally include technologies such as sensors, dedicated artificial intelligence chips, cloud computing, distributed storage, big data processing, and the like; the artificial intelligent software technology mainly comprises a computer vision technology, a voice recognition technology, a natural language processing technology, a machine learning/deep learning technology, a big data processing technology, a knowledge graph technology and the like.

Cloud computing (cloud computing) refers to a technical system that a shared physical or virtual resource pool which is elastically extensible is accessed through a network, resources can comprise servers, operating systems, networks, software, applications, storage devices and the like, and resources can be deployed and managed in an on-demand and self-service mode. Through cloud computing technology, high-efficiency and powerful data processing capability can be provided for technical application such as artificial intelligence and blockchain, and model training.

It should be appreciated that various forms of the flows shown above may be used to reorder, add, or delete steps. For example, the steps recited in the present disclosure may be performed in parallel, sequentially, or in a different order, provided that the desired results of the technical solutions provided by the present disclosure are achieved, and are not limited herein.

The above detailed description should not be taken as limiting the scope of the present disclosure. It will be apparent to those skilled in the art that various modifications, combinations, sub-combinations and alternatives are possible, depending on design requirements and other factors. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present disclosure are intended to be included within the scope of the present disclosure.

Claims

1. A method of identifying a track verification code, comprising:

2. The method of claim 1, wherein obtaining trajectory verification code drawing data comprises:

3. The method of claim 2, wherein obtaining a natural person painting index value from the trajectory verification code painting data comprises at least one of:

4. The method of claim 1, wherein determining that the natural person painting index value meets a natural person operating primary screening condition comprises:

5. The method of claim 4, further comprising, after detecting whether historical operational trajectory data matching the identity is stored:

if not, inquiring the crowd attribute label matched with the identity;

6. The method of any of claims 1-5, wherein extracting at least one biological behavioral feature in the trajectory verification code rendering data comprises:

7. The method of any of claims 1-5, wherein verifying whether the trajectory verification code rendering data is generated by natural human manipulation based on the biological behavioral characteristics comprises:

inputting the biological behavior characteristics into a natural human operation recognition model, and acquiring whether the track verification code drawing data is a first recognition result generated by natural human operation or not;

inputting the biological behavior characteristics into an expert rule module, and acquiring whether the track verification code drawing data is a second identification result generated by natural human operation or not;

and inputting the first identification result and the second identification result into a voting model, and acquiring whether the track verification code drawing data is an integrated result generated by natural human operation.

8. The method of claim 7, wherein inputting the biological behavior feature into a natural person operation recognition model, obtaining whether the trajectory verification code drawing data is a first recognition result generated by a natural person operation, comprises:

9. The method of claim 7, wherein the natural human operation recognition model is trained using cold-start offline training in combination with on-line timing delta updates.

10. The method of claim 2, further comprising:

11. An identification device for a track verification code, comprising:

12. The apparatus of claim 11, wherein the index value acquisition module is specifically configured to:

13. The apparatus of claim 12, wherein the index value acquisition module is further configured to perform at least one of:

14. The apparatus of claim 11, wherein the biometric extraction module is specifically configured to:

15. The apparatus of claim 14, the biometric extraction module further specifically configured to:

16. The apparatus of any of claims 11-15, wherein, at the biometric extraction module, further configured to:

17. The apparatus of any of claims 11-15, wherein the natural person operates the authentication module, comprising:

18. The apparatus of claim 17, wherein the first recognition result acquisition unit is specifically configured to:

19. The apparatus of claim 17, wherein the natural human operation recognition model is trained using cold-start offline training in combination with on-line timing delta updates.

20. The apparatus of claim 12, further comprising a robotic-prone client processing module to:

21. An electronic device, comprising:

at least one processor; and

a memory communicatively coupled to the at least one processor; wherein,,

the memory stores instructions executable by the at least one processor to enable the at least one processor to perform the method of any one of claims 1-10.

22. A non-transitory computer readable storage medium storing computer instructions for causing a computer to perform the method of any one of claims 1-10.

23. A computer program product comprising a computer program which, when executed by a processor, implements the steps of the method of any of claims 1-10.