Identity verification method and apparatus, electronic device, and computer program product

Abstract

The application discloses an identity verification method and device, electronic equipment and computer program product. It relates to the field of artificial intelligence, and the method comprises the following steps: collecting the optical characteristics and character information of the identity certificate of a user, determining a first quality index value of the identity information based on the optical characteristics and character information, and determining the risk level of the identity information; obtaining standard identity information, determining a second quality index value based on the standard identity information and the character information, and determining that the identity verification fails when the first quality index value is different from the second quality index value; when the first quality index value is the same as the second quality index value, performing data consistency verification through the risk level and a target blockchain; when the consistency verification fails, determining that the identity verification fails, and when the consistency verification passes, determining that the identity verification succeeds. Through the application, the problem of inaccurate recognition when identity verification is performed through optical character recognition in the related art is solved.

Description

Technical Field

[0001] This application relates to the field of artificial intelligence, and more specifically, to an authentication method, device, electronic device, and computer program product. Background Technology

[0002] Traditional identity verification methods, such as static passwords and personal identification numbers, are vulnerable to attacks like credential stuffing and phishing / social engineering due to their simplistic mechanisms. This makes them ill-equipped to handle increasingly complex and sophisticated cyber threats, leading to frequent identity theft, account misuse, and data breaches. To enhance security, biometric identification technology has become the mainstream authentication method. However, in less-than-ideal environments such as insufficient lighting, blurry images, aging devices, changes in facial expressions, worn fingerprints, or wearing gloves, accuracy drops significantly, resulting in false rejections or false releases. Identity verification methods combining optical character recognition (OCR) technology have been widely adopted in the financial, government, and internet service sectors.

[0003] However, the optical character recognition scheme used for identity verification in related technologies is highly sensitive to image quality. When the document is damaged, tilted, reflective, has low resolution, or is shot at a poor angle, character recognition is prone to errors and omissions, resulting in failure to extract or misreading of key fields such as name, ID number, and validity period, which in turn leads to verification failure or misjudgment.

[0004] There is currently no effective solution to the problem of inaccurate identification when using optical character recognition for identity verification in related technologies. Summary of the Invention

[0005] The main objective of this application is to provide an authentication method, device, electronic device, and computer program product to solve the problem of inaccurate identification when performing authentication through optical character recognition in related technologies.

[0006] To achieve the above objectives, according to one aspect of this application, an identity verification method is provided. The method includes: acquiring optical features and character information of a user's identity document through optical character recognition; determining a first quality index value of the user's identity information based on the optical features and character information, and determining a risk level of the user's identity information; obtaining the user's standard identity information from a preset database; determining a second quality index value of the user's identity information based on the standard identity information and character information; determining that the user's identity verification has failed if the first quality index value differs from the second quality index value; performing data consistency verification through the risk level and a target blockchain if the first quality index value and the second quality index value are the same, wherein the target blockchain contains the user's identity node, behavior node, and verification node each time the identity information is verified; determining that the user's identity verification has failed if the consistency verification fails, and determining that the user's identity verification has succeeded if the consistency verification passes.

[0007] Optionally, after confirming successful user authentication, the method further includes: upon detecting user transaction behavior, collecting the user's historical behavior data, wherein the historical behavior data includes the user's historical transaction operations and historical authentication operations; inputting the user's identity information and historical behavior data into a target model, processing to obtain the user's predicted operation behavior within a target period, wherein the predicted operation behavior includes predicted transaction operations and predicted authentication operations, the target model includes a gated recurrent unit layer, a fully connected network layer, and an output layer, the gated recurrent unit layer processes the historical behavior data and outputs a first hidden state vector, the fully connected network layer processes the user's identity information and outputs a second hidden state vector, merging the first hidden state vector and the second hidden state vector to obtain a merged state vector, the output layer outputting the predicted operation behavior based on the state vector; and verifying the user's transaction behavior based on the predicted operation behavior.

[0008] Optionally, verifying a user's transaction behavior based on predicted operational behavior includes: obtaining the user's actual transaction operation behavior and actual identity verification operation behavior within the target period; determining whether the predicted transaction operation behavior and the actual transaction operation behavior are the same, and determining whether the predicted identity verification operation behavior and the actual identity verification operation behavior are the same; if the predicted transaction operation behavior and the actual transaction operation behavior are the same, and the predicted identity verification operation behavior and the actual identity verification operation behavior are the same, then the transaction behavior verification is determined to be successful; if the predicted transaction operation behavior and the actual transaction operation behavior are different, or if the predicted identity verification operation behavior and the actual identity verification operation behavior are different, then the transaction behavior verification is determined to be unsuccessful.

[0009] Optionally, determining the first quality index value of the user's identity information based on optical features and character information includes: determining character contrast, edge sharpness, and image noise level in the optical features; determining a sharpness index value based on at least one of character contrast, edge sharpness, and image noise level; detecting whether there are missing field contents in the character information, obtaining a field missing detection result, and determining a completeness index value based on the field missing detection result; calculating the similarity between the character information and standard identity information, and determining the similarity as an accuracy index value; and inputting the sharpness index value, completeness index value, and accuracy index value into a preset function to obtain the first quality index value.

[0010] Optionally, determining the risk level of a user's identity information includes: obtaining the user's average quality indicator value and standard deviation over a historical period; calculating the difference between the first quality indicator value and the average quality indicator value; calculating the ratio of the difference to the standard deviation to obtain the standardized first quality indicator value; calculating the absolute value of the standardized first quality indicator value to obtain the target value; determining the risk level of the user's identity information as level one when the target value is less than or equal to a first threshold; determining the risk level of the user's identity information as level two when the target value is greater than or equal to the first threshold but less than a second threshold; and determining the risk level of the user's identity information as level three when the target value is greater than or equal to the second threshold. The first threshold is less than the second threshold, the risk level of level one is lower than that of level two, and the risk level of level two is lower than that of level three.

[0011] Optionally, determining the second quality index value of a user's identity information based on standard identity information and character information includes: mapping character information into N first numerical feature vectors through a preset character encoding method, determining N second numerical feature vectors corresponding to the standard identity information, where N is a positive integer; calculating the distance parameters between the first and second numerical feature vectors of the same type of identity information to obtain N distance parameters; and calculating the average of the N distance parameters to obtain the second quality index value.

[0012] Optionally, data consistency verification based on risk level and target blockchain includes: In the case of risk level 1, calling the current-time identity nodes and behavior nodes from the target blockchain, verifying the consistency of identity information in the current-time identity nodes with standard identity information, and verifying the consistency of operational behavior in the current-time behavior nodes with historical operational behavior; In the case of risk level 2, calling a preset number of identity nodes and behavior nodes from the target blockchain, and performing cross-validation based on the preset number of identity nodes and behavior nodes, wherein the cross-validation compares the consistency of identity information in all identity nodes and the consistency of operational behavior in all behavior nodes; In the case of risk level 3, calling all identity nodes and behavior nodes from the target blockchain, and performing cross-validation based on all identity nodes and behavior nodes in the target blockchain.

[0013] To achieve the above objectives, according to another aspect of this application, an identity verification device is provided. The device includes: a collection unit, configured to collect optical features and character information of a user's identity document through optical character recognition, determine a first quality index value of the user's identity information based on the optical features and character information, and determine the risk level of the user's identity information; an acquisition unit, configured to acquire the user's standard identity information from a preset database, determine a second quality index value of the user's identity information based on the standard identity information and character information, and determine that the user's identity verification has failed if the first quality index value differs from the second quality index value; a verification unit, configured to perform data consistency verification through the risk level and a target blockchain if the first quality index value and the second quality index value are the same, wherein the target blockchain contains the user's identity node, behavior node, and verification node each time the identity information is verified; and a first determination unit, configured to determine that the user's identity verification has failed if the consistency verification fails, and determine that the user's identity verification has succeeded if the consistency verification passes.

[0014] In this embodiment, optical features and character information of the user's identity document are collected through optical character recognition. A first quality index value for the user's identity information is determined based on these features and character information, and the risk level of the user's identity information is also determined. Standard identity information of the user is obtained from a preset database. A second quality index value for the user's identity information is determined based on the standard identity information and character information. If the first quality index value differs from the second quality index value, the user's identity verification is deemed to have failed. If the first quality index value and the second quality index value are the same, data consistency verification is performed using the risk level and a target blockchain. The target blockchain contains the user's identity node, behavior node, and verification node for each identity verification. In the case of consistency... If the identity verification fails, the user's identity verification is determined to have failed. If the consistency verification passes, the user's identity verification is determined to have succeeded. The quality of the ID card image is evaluated by quality index values ​​determined based on optical features and character information. Identity verification is performed by comparing the first quality index value with the second quality index value, effectively identifying blurry, damaged, or counterfeit documents. Data consistency verification is performed based on the risk level of the identity information and the target blockchain, ensuring the privacy and immutability of sensitive data. This reduces the false judgment rate caused by image quality, thereby improving the accuracy of identification during identity verification through optical character recognition and solving the technical problem of inaccurate identification when using optical character recognition for identity verification. Attached Figure Description

[0015] The accompanying drawings, which form part of this application, are used to provide a further understanding of this application. The illustrative embodiments and descriptions of this application are used to explain this application and do not constitute an undue limitation of this application. In the drawings:

[0016] Figure 1 A hardware structure block diagram of a computer terminal (or mobile device) for implementing an authentication method is shown.

[0017] Figure 2 This is a flowchart of an authentication method provided according to an embodiment of this application;

[0018] Figure 3 This is a flowchart of an optional authentication method provided according to an embodiment of this application;

[0019] Figure 4 This is a schematic diagram illustrating the number of authentication attempts for different risk levels according to embodiments of this application;

[0020] Figure 5 This is a schematic diagram illustrating the prediction accuracy of a deep learning model provided in the embodiments of this application;

[0021] Figure 6 This is a schematic diagram illustrating the transaction success rate provided in the embodiments of this application;

[0022] Figure 7 This is a schematic diagram illustrating the identity verification success rate and transaction verification success rate according to the embodiments of this application;

[0023] Figure 8 This is a schematic diagram of an authentication device provided according to an embodiment of this application;

[0024] Figure 9 This is a structural block diagram of an electronic device according to an embodiment of this application. Detailed Implementation

[0025] To enable those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present application, and not all embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative effort should fall within the scope of protection of the present application.

[0026] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of this application described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0027] It should be noted that the information collected in this application (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for display, data used for analysis, etc.) are information and data authorized by the user or fully authorized by all parties. Furthermore, the collection, storage, use, processing, transmission, provision, disclosure, and application of this data all comply with relevant laws, regulations, and standards, necessary confidentiality measures have been taken, and they do not violate public order and good morals. Corresponding operation entry points are provided for users to choose to authorize or refuse. For example, interfaces are set up between this system and relevant users or organizations, providing users with corresponding operation entry points for them to choose to agree to or refuse automated decision results; if the user chooses to refuse, the process proceeds to the expert decision-making stage.

[0028] Example 1

[0029] According to an embodiment of this application, an authentication method embodiment is also provided. It should be noted that the steps shown in the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions. Although a logical order is shown in the flowchart, in some cases, the steps shown or described may be executed in a different order than that shown here.

[0030] The method embodiment provided in Embodiment 1 of this application can be executed on a mobile terminal, computer terminal, or similar computing device. Figure 1 A hardware structure block diagram of a computer terminal (or mobile device) for implementing an authentication method is shown. Figure 1 As shown, the computer terminal 10 (or mobile device) may include one or more processors 102 (shown as 102a, 102b, ..., 102n in the figure) 102 (processor 102 may include, but is not limited to, processing devices such as MCU (Microcontroller Unit) or FPGA (Field-Programmable Gate Array), memory 104 for storing data, and transmission device 106 for communication functions. In addition, it may also include: a display, input / output interfaces (I / O interfaces), a USB (Universal Serial Bus) port (which may be included as one of the ports of a BUS (Business Bus), a network interface, a power supply, and / or a camera. Those skilled in the art will understand that... Figure 1 The structure shown is for illustrative purposes only and does not limit the structure of the aforementioned electronic device. For example, computer terminal 10 may also include... Figure 1 The more or fewer components shown, or having the same Figure 1 The different configurations shown.

[0031] It should be noted that the aforementioned one or more processors 102 and / or other data processing circuits are generally referred to herein as "data processing circuits". These data processing circuits may be embodied, in whole or in part, in software, hardware, firmware, or any other combination thereof. Furthermore, the data processing circuits may be a single, independent processing module, or may be integrated, in whole or in part, into any other element within the computer terminal 10 (or mobile device). As involved in the embodiments of this application, the data processing circuits serve as a processor control mechanism (e.g., selection of a variable resistor termination path connected to an interface).

[0032] The memory 104 can be used to store software programs and modules of application software, such as the program instructions / data storage device corresponding to the authentication method in this embodiment. The processor 102 executes various functional applications and data processing by running the software programs and modules stored in the memory 104, thereby implementing the aforementioned authentication method. The memory 104 may include high-speed random access memory, and may also include non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some instances, the memory 104 may further include memory remotely located relative to the processor 102, and these remote memories can be connected to the computer terminal 10 via a network. Examples of such networks include, but are not limited to, the Internet, corporate intranets, local area networks, mobile communication networks, and combinations thereof.

[0033] The transmission device 106 is used to receive or send data via a network. Specific examples of the network described above may include a wireless network provided by the communication provider of the computer terminal 10. In one example, the transmission device 106 includes a Network Interface Controller (NIC), which can connect to other network devices via a base station to communicate with the Internet. In another example, the transmission device 106 may be a Radio Frequency (RF) module, used for wireless communication with the Internet.

[0034] The display may be, for example, a touchscreen LCD display that allows the user to interact with the user interface of the computer terminal 10 (or mobile device).

[0035] In the above operating environment, this application provides an authentication method. Figure 2 This is a flowchart of the authentication method provided according to the embodiments of this application, such as... Figure 2 As shown, the method includes:

[0036] Step S201: Collect the optical features and character information of the user's identity document through optical character recognition, determine the first quality index value of the user's identity information based on the optical features and character information, and determine the risk level of the user's identity information.

[0037] In step S201, optical features may include physical imaging quality parameters such as character edge sharpness, background and text contrast, image noise level, and portrait clarity. Character information can be specific characters identified in the identity document, including identity information such as name and ID number. Based on these optical features and extracted character information, the identity information is quality-assessed, and a first quality index value is calculated. This value is a quantitative assessment of the authenticity and information integrity of the original document image, identifying low-quality, high-risk submitted documents at the source, eliminating abnormal input caused by camera shake, obstruction, printing distortion, or forged splicing, avoiding subsequent processes wasting invalid data, thereby improving the overall robustness and efficiency of verification. Simultaneously, the risk level of the identity information is determined based on the quality index values ​​from the user's historical identity verification process.

[0038] Step S202: Obtain the user's standard identity information from the preset database, determine the second quality index value of the user's identity information based on the standard identity information and character information, and determine the user's identity verification failure if the first quality index value and the second quality index value are different.

[0039] In step S202, the preset database can be a database where financial institutions store user information, and the standard identity information can be verified real identity data. By mapping the extracted character information to a numerical feature vector according to a preset character encoding standard, and then comparing it with the numerical vector of the standard identity information using Euclidean distance, the second quality index value is determined. This transforms semantic recognition into feature similarity evaluation in mathematical space, enabling more accurate detection of subtle anomalies such as character replacement, number misalignment, or forgery and tampering. The second quality index value reflects the credibility of the content's authenticity. When there is a significant difference between the first quality index value (image optical quality) and the second quality index value (content structure consistency), for example, if the image is clear but the characters do not match the standard library, or if the image is blurry but the characters perfectly match, a contradiction is determined, and identity verification fails.

[0040] Step S203: If the first quality indicator value and the second quality indicator value are the same, data consistency verification is performed through risk level and target blockchain. The target blockchain contains the identity node, behavior node and verification node of the user each time the identity information is verified.

[0041] In step S203, if the first quality indicator value is the same as the second quality indicator value, it indicates that the user-submitted document has no obvious abnormalities at the image and content levels, and a higher-level trusted collaborative verification is triggered. By collecting the user's historical behavior data, a target blockchain is established using the user's identity information and historical behavior data. Based on the risk level, the target blockchain adaptively selects user identity information and historical behavior data to participate in data consistency verification. For example, in high-risk situations, all identity nodes, behavior nodes, and verification nodes are invoked; in low-risk situations, only the current behavior node is activated.

[0042] It's important to note that the target blockchain is not static storage, but rather a distributed ledger constructed from user historical behavior data (login device, IP address, transaction frequency) and identity digests. Each node stores hash-encrypted, immutable data blocks, ensuring that any modifications are recorded on the chain. Based on the risk level, associated historical data blocks matching the current verification time and behavioral context are adaptively extracted from this chain. The current operation is compared to ensure it maintains logical consistency with the user's long-term behavioral patterns and identity registration information in terms of time, space, and semantics. This upgrades single-point verification to trusted historical behavior tracing, determining whether the current operation conforms to the user's consistent behavioral trajectory, thereby effectively defending against advanced attacks such as impersonation and account theft.

[0043] Step S204: If the consistency verification fails, determine that the user's authentication has failed; if the consistency verification passes, determine that the user's authentication has succeeded.

[0044] In step S204, a failure to pass the consistency verification indicates that even if the document image is clear and the content matches, the current operation conflicts with the user's historical behavior patterns or identity chain trajectory recorded by the system. For example, a user who has been logging in locally for a long time suddenly makes a large transfer from another location, or the behavior node shows that the user usually uses device A but used device B, which has never been registered before. This is judged as high-risk and suspicious behavior, and the process is immediately interrupted to prevent potential risks. Conversely, if the behavior trajectory, identity digest, and blockchain node data matching the risk level are highly consistent, it proves that the current operation is a legitimate action initiated by the user in a situation that meets expectations, and the user's identity verification is confirmed to be successful.

[0045] The identity verification method provided in this application collects optical features and character information of a user's identity document through optical character recognition (OCR). Based on the optical features and character information, it determines a first quality index value of the user's identity information and the risk level of the user's identity information. It then retrieves the user's standard identity information from a preset database and determines a second quality index value of the user's identity information based on the standard identity information and character information. If the first quality index value differs from the second quality index value, the user's identity verification fails. If the first quality index value and the second quality index value are the same, data consistency verification is performed using the risk level and a target blockchain. The target blockchain contains the user's identity node, behavior node, and verification node for each identity verification. If the consistency verification fails, the user's identity verification is determined to have failed; if the consistency verification passes, the user's identity verification is determined to have succeeded. The quality of the ID card image is evaluated based on quality index values ​​determined by optical features and character information. Identity verification is performed based on the comparison of the first quality index value and the second quality index value, effectively identifying blurry, damaged, or counterfeit documents. Data consistency verification is performed according to the risk level of the identity information and the target blockchain, ensuring the privacy and immutability of sensitive data and reducing the false judgment rate caused by image quality. This achieves the technical effect of improving the accuracy of identification during identity verification through optical character recognition, thereby solving the technical problem of inaccurate identification when using optical character recognition for identity verification.

[0046] If authentication is successful, further verification is performed when the user engages in a transaction. Optionally, in the authentication method provided in this application embodiment, after determining that the user's authentication is successful, the method further includes: collecting the user's historical behavior data when the user's transaction behavior is detected, wherein the historical behavior data includes the user's historical transaction operation behavior and historical authentication operation behavior; inputting the user's identity information and historical behavior data into a target model, processing to obtain the user's predicted operation behavior within a target period, wherein the predicted operation behavior includes predicted transaction operation behavior and predicted authentication operation behavior, the target model includes a gated recurrent unit layer, a fully connected network layer and an output layer, the gated recurrent unit layer processes the historical behavior data and outputs a first hidden state vector, the fully connected network layer processes the user's identity information and outputs a second hidden state vector, the first hidden state vector and the second hidden state vector are merged to obtain a merged state vector, the output layer outputs the predicted operation behavior based on the state vector; and verifying the user's transaction behavior based on the predicted operation behavior.

[0047] In some embodiments, user identity information and historical behavior data are input into the target model as feature vectors for identity verification. This predicts the user's operational behavior during identity verification within a target period. Simultaneously, a user transaction identity verification strategy is designed, binding the predicted user operational behavior during identity verification within the target period with transaction information to perform secondary verification of the user's identity information during the current transaction process. The historical behavior data is not read directly from the target blockchain but is obtained as raw data from an off-chain behavior database.

[0048] For example, the identity feature vector corresponding to a user's identity information. It is a length of A fixed-length vector, obtained by mapping user identity information (such as name, ID number, etc.) through a preset character encoding standard, where each character corresponds to a numerical value, and is padded or truncated to a fixed length. Historical behavior feature vectors corresponding to historical behavior data. It is a time series matrix with the shape of × ,in The length of the historical time window (e.g., the last 30 login events). For each time step, a behavioral feature dimension is defined (including encoded features such as login device, IP, and operation type). For non-uniformly spaced events, they can be aligned to equal time steps through resampling or interpolation. The identity feature vector is then... and historical behavior feature vector By merging, a complete user feature representation can be formed. ,in, , This is the identity vector.

[0049] The target model can be a two-branch neural network model, with the following structure: The identity feature extraction branch consists of fully connected network layers: Input... After passing through two fully connected layers, each followed by a ReLU activation function and Dropout regularization, the final output is the identity hidden state vector, which is also the first hidden state vector. ,in, The identity feature dimension is used (e.g., 64-dimensional). A gated recurrent unit layer is employed to process the behavioral time series. The behavior sequence extraction branch is constructed by a gated recurrent unit layer, which can effectively capture long-term dependencies in the sequence. The hidden state dimension is set to... (e.g., 128-dimensional). The hidden state of the last time step. It serves as a global representation of the behavioral sequence and can also output the hidden state sequence at all time steps for subsequent fine-grained analysis.

[0050] To enhance the model's focus on key behaviors, an attention mechanism can be added after the gated recurrent unit layer to perform weighted aggregation of the hidden states at each time step, resulting in the behavior hidden state vector, which is also the second hidden state vector. .Will and Concatenate to form a state vector This vector is then input into an output layer, which predicts the behavior at the next k time steps. A sequence generation module is designed to do this. Starting from the initial state, a recurrent neural network is used to progressively generate the probability of behavior categories at each future time step. Or directly output a through a fully connected layer. The matrix represents the joint probability for the next k steps. Wherein, ,in, The number of preset operation categories (such as login, transfer, query, etc.). At the current time point, This becomes the next time step, and so on.

[0051] Predicting user behavior involves forecasting the most likely action a user is currently performing and comparing it in real-time with the type of transaction the user actually submitted to determine if the current behavior matches their historical behavior patterns. Predicting the sequence of actions a user might perform in the next k time steps (e.g., the next few minutes, hours, or interactions) is used to assess whether the current transaction matches the user's expected future behavior patterns. For example, if the model predicts a large transfer within the next hour, and a large transfer happens to occur, it's considered a match; conversely, if future predictions are all small queries, but a large transfer occurs, a risk warning is triggered.

[0052] The target model can be trained end-to-end, with the loss function being the cross-entropy loss of the predicted user behavior. The target model is trained using historical user data. The training samples are constructed as follows:

[0053] (1) For each user, select consecutive Behavioral data at each time step, before Steps as input sequence The next k steps (e.g., k=5) are used as labels for future actions.

[0054] (2) The current behavior label is taken from the first position. The actual operation of the steps.

[0055] (3) Identity characteristics Retrieved from the user's identity information during registration or the most recently updated information.

[0056] During training, an adaptive moment estimation optimizer was used, with an initial learning rate of 0.001, a batch size of 64, and the number of training epochs determined by early stopping on the validation set. Network hyperparameters (such as the hidden layer dimensions of gated recurrent unit layers, the layer dimensions of fully connected networks, and the attention mechanism) were determined through grid search or Bayesian optimization, with validation set accuracy as the objective. After training, the target model can output a predicted sequence of future behaviors from any user's input of their identity features and recent behavior sequence. .

[0057] For example, in real-time authentication scenarios, when a user initiates a transaction (such as a transfer, payment, or modification of key information), the current transaction information is bound to the model's prediction results to achieve dynamic and proactive verification. Key attributes are extracted from the transaction request, including transaction type (e.g., transfer), transaction amount, recipient, device information, and IP address, and encoded into a format consistent with the model's output space. The user's most recent... Step's historical behavioral data and current identity vector Input the trained target model to obtain the predicted operational behavior.

[0058] This embodiment constructs a target model, using user identity information and historical behavior data as feature vector inputs, which can predict the user's predicted operational behavior in transactions. Based on the predicted operational behavior, the user's transaction behavior is verified, achieving a high degree of integration between user identity verification and transaction verification, and reducing the probability of transaction risk behavior.

[0059] Transaction behavior is verified by comparing the predicted operation behavior with the actual operation behavior. Optionally, in the identity verification method provided in this application embodiment, verifying the user's transaction behavior based on the predicted operation behavior includes: obtaining the user's actual transaction operation behavior and actual identity verification operation behavior within a target period; determining whether the predicted transaction operation behavior and the actual transaction operation behavior are the same, and determining whether the predicted identity verification operation behavior and the actual identity verification operation behavior are the same; if the predicted transaction operation behavior and the actual transaction operation behavior are the same, and the predicted identity verification operation behavior and the actual identity verification operation behavior are the same, then the transaction behavior verification is determined to be successful; if the predicted transaction operation behavior and the actual transaction operation behavior are different, or if the predicted identity verification operation behavior and the actual identity verification operation behavior are different, then the transaction behavior verification is determined to be unsuccessful.

[0060] In some embodiments, predicting operational behavior may include predicting transaction operational behavior and predicting authentication operational behavior. The actual transaction operational behavior is compared with the category with the highest probability among the predicted transaction operational behaviors of the target model. If they match and the probability exceeds a set probability threshold (e.g., 0.8), the current actual transaction operational behavior is considered to meet expectations. Similarly, the actual authentication operational behavior is compared with the category with the highest probability among the predicted authentication operational behaviors of the target model. If they match and the probability exceeds a set probability threshold (e.g., 0.8), the current actual authentication operational behavior is considered to meet expectations.

[0061] For example, the actual transaction's time and amount are roughly matched with the predicted future behavior sequence (e.g., if a large transfer is predicted within the next hour, and the current transaction happens to be a large transfer, it's considered a match). If both match, it's deemed low-risk, the transaction verification passes, and only basic identity verification (such as password or fingerprint) is required. If the current behavior prediction doesn't match the actual transaction type, or the future prediction deviates significantly from the transaction pattern (e.g., predicted as a small purchase, but actually a large transfer), the transaction verification fails, triggering enhanced verification. This requires the user to perform additional identity authentication, such as SMS verification code, facial recognition, or answering security questions, while the abnormal event is recorded in the target blockchain's verification node. If enhanced verification fails, the transaction is terminated, the user is marked as high-risk, and the user's risk level is updated so that subsequent verifications use a stricter node selection strategy. If verification passes, the transaction information is bound to the user's identity verification result, generating an immutable verification record on the blockchain. This record includes transaction hash, verification time, risk level, and verification method used, providing a basis for subsequent auditing and risk control.

[0062] It should be noted that if enhanced verification is triggered due to a discrepancy between predicted and actual transaction behavior, and the enhanced verification fails, the transaction will be terminated, and the user's risk level will be updated to high risk. This update will be immediately written to the user's risk status record, but as mentioned earlier, it will not affect the identity verification result. Subsequent verification requests initiated by this user will directly adopt the high-risk level node selection strategy and data block generation frequency.

[0063] This embodiment compares the predicted transaction behavior output by the target model with the actual transaction behavior, effectively identifying abnormal behaviors such as forgery and theft. Even if an attacker obtains the user's identity information, they will not be able to simulate its unique behavior pattern, thus improving the accuracy and response speed of identifying complex transaction risks.

[0064] After collecting the optical features and character information of the identity document, a first quality index value is determined based on the optical features and character information. Optionally, in the identity verification method provided in this application embodiment, determining the first quality index value of the user's identity information based on the optical features and character information includes: determining the character contrast, edge sharpness, and image noise level in the optical features; determining a clarity index value based on at least one of the character contrast, edge sharpness, and image noise level; detecting whether there are missing field contents in the character information, obtaining a field missing detection result, and determining a completeness index value based on the field missing detection result; calculating the similarity between the character information and standard identity information, and determining the similarity as an accuracy index value; and inputting the clarity index value, completeness index value, and accuracy index value into a preset function to obtain the first quality index value.

[0065] In some embodiments, optical character recognition (OCR) is used to convert images of printed or handwritten text (such as scanned copies, photographs, and text within photographs) into an editable text format. Here, the optical characteristics of character information contained in the user's identity information are extracted, such as character contrast (referring to ID number, name, date of birth, etc.), edge sharpness (referring to the clarity of facial features in the image), and image noise level (referring to the personal photograph on the ID and the user's current personal photograph; the current personal photograph is taken by a photographing device and needs to meet actual edge sharpness; the higher the sharpness, the clearer the image). Based on the OCR method, the optical characteristics of character information contained in the user's identity information are extracted, and a sharpness index value is defined. Simultaneously, the integrity of the extracted character information is examined based on its optical properties, and an integrity index value is defined. Checking for completeness can include verifying whether necessary fields are missing, such as ID number and name.

[0066] To quantify the quality of the recognition results, sharpness, completeness, and accuracy are defined, with all values ​​ranging from [value range missing]. A higher value indicates better quality: Sharpness index value The sharpness of character regions in the original image is reflected and calculated by an image processing algorithm. The calculation method is as follows: obtain image blocks of character regions from an optical character recognition engine; calculate the Laplacian variance for each character block; a larger value indicates sharper image edges and higher sharpness; average the Laplacian variances of all character blocks and linearly map them to the interval [1, 100].

[0067] ;

[0068] Where max_expected is an empirical threshold (e.g., 1000), representing the maximum expected value of the Laplacian variance that may be reached under normal image quality, which can be adjusted according to the actual image resolution; This is the average Laplacian variance of all character regions in the image. The Laplacian operator is used to detect image edges; the larger the variance, the sharper the edges and the clearer the image.

[0069] Let the set of required fields be... The actual set of fields identified is ,but:

[0070] ;

[0071] If the field content is empty or obviously invalid (such as the ID number being too short), it will not be counted. This calculation ensures that none of the required fields are missing.

[0072] Accuracy index value The determination includes: comparing the character information of optical character recognition with the user's standard identity information stored in a preset database character by character, and calculating the matching degree; for each field, using edit distance to calculate the similarity between the recognized string and the standard string:

[0073] ;

[0074] in, Edit distance refers to the minimum number of editing operations (insertion, deletion, replacement) required to transform one string into another. A smaller value indicates that the two strings are more similar. and These represent the lengths of the two strings respectively. The maximum of the two values ​​is used as the normalized denominator to ensure that the similarity is not excessively affected by differences in string length. A weighted average of the similarities across all fields is then calculated (weights can be assigned according to field importance). If some fields are not present in the database (e.g., for new users), only the existing fields are calculated, and missing fields are not included in the denominator. Optionally, the optical characteristics of the character information are compared with the standard identity information stored in the user identity database to calculate the matching degree, and the accuracy index value is defined by the matching degree. Calculating the matching degree can not only filter out users with similar information, but also verify user information to determine whether false identity information has been submitted.

[0075] It should be noted that user identity databases require advanced access; in practice, access and matching require application to relevant departments or must be conducted without violating user privacy protections. (Based on clarity index values) Integrity index value and accuracy index values The first quality index value is obtained. .

[0076] Specifically, the first quality indicator value The calculation formula is as follows:

[0077] ;

[0078] in, , , These are the weighting coefficients, satisfying... The weights can be dynamically adjusted according to the application scenario. For example, in scenarios requiring high-precision verification (such as financial transactions), the accuracy weight can be increased, while in scenarios with poor image quality but requiring rapid verification (such as mobile devices), the clarity weight can be increased. The specific weight values ​​are preset by technical experts.

[0079] This embodiment constructs a standardized quality scoring system by defining clarity, integrity, and accuracy indicators for user identity information, thereby improving the accuracy and flexibility of identity information verification.

[0080] The risk level can be determined based on the quality index values ​​in the user's historical identity verification process. Optionally, in the identity verification method provided in this application embodiment, determining the risk level of the user's identity information includes: obtaining the user's average quality index value and standard deviation in a historical period; calculating the difference between the first quality index value and the average quality index value; and calculating the ratio of the difference to the standard deviation to obtain the standardized first quality index value; calculating the absolute value of the standardized first quality index value to obtain the target value; determining the risk level of the user's identity information as the first level when the target value is less than or equal to a first threshold; determining the risk level of the user's identity information as the second level when the target value is greater than or equal to the first threshold but less than a second threshold; and determining the risk level of the user's identity information as the third level when the target value is greater than or equal to the second threshold. The first threshold is less than the second threshold, the risk level of the first level is lower than that of the second level, and the risk level of the second level is lower than that of the third level.

[0081] In some embodiments, in order to eliminate scoring biases caused by different users, different devices, and different environments, the first quality index value is... Standardization is performed. The standardization benchmark uses the user's historical successful verification records. Statistics: Collect the user's most recent N successful verifications. The value of N (which can be configured according to the system, e.g., N=10) is used to calculate the average quality index value. and standard deviation If there is no historical record (new user), then the global average of all users is used. and standard deviation The standardized score, that is, the first quality indicator value after standardization, is calculated using the following formula:

[0082] ;

[0083] in, This reflects the degree of deviation of the current recognition quality from the user's historical level. A value greater than 0 indicates that it is better than the historical average, while a value less than 0 indicates that it is worse than the historical average. The absolute value of the first quality indicator after standardization is calculated to obtain the target value. Based on the magnitude of the target value, the identity information is divided into three risk levels: a target value less than or equal to the first threshold (e.g., 1) indicates that the current recognition quality is within the normal historical fluctuation range, and is recorded as low risk, i.e., level one; a target value between the first and second thresholds (e.g., 2) indicates that the quality has some anomalies and requires stronger verification, and is recorded as medium risk, i.e., level two; a target value greater than the second threshold indicates that the quality is significantly abnormal, and there may be forgery or serious recognition errors, and is recorded as high risk, i.e., level three. The values ​​1 and 2 can be adjusted according to system security requirements (e.g., changed to 1.5 and 2.5) and used in subsequent steps to select verification strategies.

[0084] This embodiment dynamically defines the risk level by the degree of deviation between the current quality indicator value and the historical average quality indicator value, thereby improving the accuracy and flexibility of identity information verification.

[0085] The second quality index value is determined by the distance parameter between the feature vectors corresponding to the standard identity information and the character information. Optionally, in the identity verification method provided in this application embodiment, determining the second quality index value of the user's identity information based on the standard identity information and the character information includes: mapping the character information into N first numerical feature vectors through a preset character encoding method, determining N second numerical feature vectors corresponding to the standard identity information, where N is a positive integer; calculating the distance parameter between the first numerical feature vector and the second numerical feature vector of the same type of identity information to obtain N distance parameters; and calculating the average value of the N distance parameters to obtain the second quality index value.

[0086] In some embodiments, a preset character encoding method is used to convert character information into a fixed-length numerical vector:

[0087] (1) Split the string of each field into characters, and map each character to the preset character encoding standard code value (0-127); for fields with insufficient length, padding characters (such as spaces, preset character encoding standard 32) are used to fill the field to the maximum length.

[0088] (2) Concatenate the preset character encoding standard code values ​​of all fields in order to form a one-dimensional identity feature vector. For example, the name Zhang San → [229, 147, 136, 228, 184, 137] (assuming UTF-8 encoding; in actual applications, a unified encoding, such as UTF-8 or GBK, is required); the ID number abcdcddeeccdcddbaf → [51, 50, 48, 49, 48, 49, 49, 57, 57, 48, 48, 49, 48, 49, 49, 50, 51, 52]. Chinese characters can be converted to Pinyin and then mapped to a preset character encoding standard. This vector will serve as one of the feature inputs for a deep learning model.

[0089] To improve the reliability of verification, the first numerical feature vector can be compared with the second numerical feature vector to calculate the vector similarity:

[0090] The distance parameter can be Euclidean distance; calculate the Euclidean distance. :

[0091] ;

[0092] in, The total length of the vector. It is the i-th first numerical eigenvector. It is the i-th second numerical feature vector; the distance is converted into a similarity score, which is also the second quality index value. :

[0093] ;

[0094] in, To determine the maximum possible distance (e.g., for a preset character encoding standard range of 0-127, the maximum difference in each dimension is 127), then... ).like If the score is below a preset threshold (e.g., 90), the current recognition result is considered inconsistent with the database information, the verification is directly deemed a failure, and the user is prompted to resubmit. The first quality index value... Compared with the second quality index value obtained based on numerical vectors If the difference between the two values ​​exceeds the tolerance range (e.g., 15), it indicates that there is an internal contradiction in the recognition result (e.g., the characters are clear but the content does not match the database), and the identity verification is determined to have failed. At this time, a second manual review can be triggered or the verification can be directly rejected.

[0095] This embodiment determines the second quality index value by constructing a distance parameter between the feature vectors corresponding to standard identity information and character information, which makes up for the semantic errors caused by optical character recognition errors or image blurring. Furthermore, it accurately identifies forgery and tampering behavior through field-level distance analysis, builds a content authenticity guarantee mechanism in the identity verification process, and enhances the defense capability against highly realistic forged documents and data tampering attacks.

[0096] Based on different risk levels, nodes in the target blockchain are selected for data consistency verification. Optionally, in the identity verification method provided in this application embodiment, data consistency verification through risk level and target blockchain includes: when the risk level is first level, calling the current time identity node and behavior node from the target blockchain, verifying the consistency of the identity information in the current time identity node with the standard identity information, and verifying the consistency of the operation behavior in the current time behavior node with the historical operation behavior; when the risk level is second level, calling a preset number of identity nodes and behavior nodes from the target blockchain, and performing cross-verification based on the preset number of identity nodes and behavior nodes, wherein the cross-verification compares the consistency of the identity information in all identity nodes and the consistency of the operation behavior in all behavior nodes; when the risk level is third level, calling all identity nodes and behavior nodes from the target blockchain, and performing cross-verification based on all identity nodes and behavior nodes in the target blockchain.

[0097] In some embodiments, the target blockchain can be built during user registration or system initialization and run continuously. Identity nodes, behavior nodes, and verification nodes exist long-term, each maintaining a chronologically generated chain of data blocks. Each time a user initiates a verification request, the target blockchain is directly invoked without needing to be rebuilt. Specifically, identity nodes are used to store cryptographic digests (such as SHA-256 hash values) of user identity information and their metadata; the original plaintext information is not uploaded to the chain, thus ensuring data verifiability while preventing privacy leaks. Behavior nodes are responsible for storing digests of users' historical behavior data, which are obtained by constructing a Merkle tree from behavior data over a period of time and taking its root hash, enabling efficient verification of the integrity of large amounts of data. Verification nodes record key information for each user verification event, including a list of participating nodes, verification results, risk levels, and timestamps, forming an immutable audit log.

[0098] Each node's data blocks are organized in a chain structure: identity node data blocks contain an identity information digest, a timestamp, and the hash value of the previous identity node data block; behavior node data blocks contain the behavior data Merkle root, a timestamp, and the hash value of the previous behavior node data block; and verification node data blocks record verification result details and are linked to the previous verification node data block. Each node's data block chain is maintained solely by that node, and the chains between nodes are independent of each other, but global consistency is guaranteed through a distributed consensus mechanism. This allows each node to store only data relevant to its own type, reducing storage and synchronization overhead; data blocks are linked within the same node, facilitating rapid verification of data integrity (tampering with any data block will cause all subsequent hash values ​​to mismatch); and data block chains for different node types can be generated at different frequencies (e.g., low frequency for identity nodes, high frequency for behavior nodes), adapting to dynamic risk adjustment.

[0099] It should be noted that although each type of node forms an independent chain, the entire network ensures the time order and global perspective of data blocks across nodes through a consensus protocol. Figure 1 This creates a unified distributed ledger. The generation frequency of data blocks is dynamically adjusted based on node type and user risk level: identity node data blocks can be generated when user identity information changes or at fixed periods (such as monthly); the generation of behavior node data blocks is linked to risk level, with high-risk users' behavior node data blocks being generated at a lower frequency than low-risk users'. This allows for more intensive monitoring of abnormal behavior; verification node data blocks are generated instantly upon each verification event, where... It can be set to 2-3 times, depending on actual needs. By distributing user identity information and historical behavior data across different nodes, each node stores only a portion of the information, ensuring security and privacy when acquiring this data.

[0100] It's important to note that the blockchain only stores cryptographic digests (such as hash values ​​and Merkle roots) of user identity information and behavioral data to ensure data integrity and immutability. The original, detailed behavioral data (such as login time, device information, and transaction records) is stored off-chain in a distributed database or local data warehouse for efficient querying and deep learning model access. When data integrity needs to be verified, the digest of the original data can be recalculated and compared with the on-chain digest.

[0101] The risk level is used to dynamically determine the number and type of nodes required for the current verification. This mechanism ensures high security while minimizing the waste of system resources. For example, at the first level, only the current identity node and current behavior node (i.e., the main node determined by the user ID hash mapping) directly associated with the user are selected for verification. A behavior-first verification order is adopted, verifying the behavior node first. If the behavior pattern is normal, it passes quickly. Identity node verification is only triggered when the behavior is abnormal, thereby significantly reducing latency.

[0102] Level Two: Approximately half of the identity nodes and half of the behavior nodes are randomly selected for cross-validation, using an identity-priority order. Identity nodes are verified first to ensure user authenticity, and behavior nodes are verified only after successful identity verification. If identity verification fails, the user is rejected directly to avoid subsequent resource consumption. Node selection is based on a pseudo-random seed using the user ID to ensure that the selection results are repeated and evenly distributed when the same user is verified multiple times.

[0103] Level 3: Forces full cross-validation of all identity nodes and all behavior nodes in the network, following the same identity priority order, to ensure the reliability of the verification results with the highest strength.

[0104] During cross-validation, each node independently compares its stored data digest with the information currently submitted by the user, signs the comparison result, and sends it to the validator node. After collecting a sufficient number of responses, the validator node determines whether the validation passes according to the preset consensus rules (such as majority consensus) and generates a validation data block containing the complete validation record, which is then uploaded to the blockchain.

[0105] It should be noted that the risk level is tightly coupled with the blockchain data generation and verification strategy to form an adaptive closed loop. Whenever a user's risk level is updated, the new level takes effect immediately and affects two aspects of behavior: (1) The background task adjusts the generation cycle of subsequent behavior node data blocks according to the latest risk level. The behavior node data of high-risk users is packaged and uploaded to the chain more frequently, enabling the system to capture subtle changes in behavior patterns earlier and provide higher time resolution data support for subsequent verification. (2) When a user initiates an identity verification request next time, the corresponding node selection rule will be executed directly according to the latest risk level. If the risk level changes again during the verification process (e.g., due to fluctuations in quality score), the changed level will be applied to the next verification after the current verification ends, ensuring that the system always makes decisions based on the latest situation.

[0106] This embodiment constructs a distributed blockchain based on user identity information and historical behavior data. The user's identity information and behavior data are distributed across different nodes, and the nodes participating in the verification are adaptively selected according to the risk level, which further improves the security and reliability of identity verification. The dynamic adjustment mechanism not only improves the system's response speed to sudden risks, but also optimizes the resource utilization rate under long-term operation. Low-risk users enjoy a smooth experience brought by lightweight verification, while high-risk users receive strict protection, achieving a balance between security and efficiency.

[0107] According to another embodiment of this application, an optional authentication method is also provided. Figure 3 This is a flowchart of an optional authentication method provided according to an embodiment of this application, such as... Figure 3 As shown, the method includes:

[0108] Step S1: Collect the user's identity information during the identity verification process. Use optical character recognition (OCR) to score the quality of the identity information and define the risk level. Then, convert the character information in the identity information into numerical information to obtain new identity information. Score the quality of the new identity information. If the quality scores of the current identity information and the new identity information are different, the current identity information is incorrect and the identity verification fails. Otherwise, continue execution.

[0109] Step S2: Collect users' historical behavior data. Establish a distributed blockchain using users' identity information and historical behavior data. The distributed blockchain adaptively selects user identity information and historical behavior data to participate in verification based on the defined risk level.

[0110] Step S3: Establish a deep learning model. Input the user identity information and historical behavior data as feature vectors for identity verification into the deep learning model to predict the user's operational behavior in the current and future identity verification. At the same time, design a user transaction identity verification strategy. Bind the predicted user's operational behavior in the current and future identity verification with the transaction information to perform secondary verification of the user's identity information in the current and future transaction processes.

[0111] The optional identity verification method in this embodiment improves the accuracy and flexibility of identity verification by constructing a standardized quality scoring system and dynamically defining risk levels based on the deviation between the standardized score and the historical average score. It also constructs a distributed blockchain based on user identity information and historical behavior data, distributing user identity information and behavior data across different nodes and adaptively selecting nodes to participate in verification according to risk levels, further enhancing the security and reliability of the application system. Furthermore, by constructing a deep learning model that uses user identity information and behavior data as feature vector input, it can predict user behavior in the present and future and design corresponding transaction identity verification strategies, achieving a high degree of integration between user identity verification and transaction verification, thus reducing the probability of risky transactions.

[0112] The accuracy and security of the selected authentication methods in the authentication process were verified through simulation experiments. The experiments used... The database used data from a sample of registered users, whose identity information included name, ID number, address, phone number, and email address, was also collected. Additionally, historical behavioral data from the past year, such as login time, device information, visited pages, and transaction types, was also collected. Implementation process:

[0113] In the first phase of the simulation experiment, text information was extracted from the user-submitted identity image files to generate initial identity information data; then, optical character recognition software was used to score the quality of this data, including a sharpness index value (…). ), integrity index value ( ) and accuracy index values ​​( The quality scoring system is based on standardized processing, and a standardized quality score is calculated using a formula. Based on this, the characters in the identity information are encoded using a preset character encoding standard code to generate a new feature vector, and a new quality score is generated by comparison. );if and If there are significant differences, the identity information is marked as abnormal, and the user is required to re-verify.

[0114] In the second phase of the simulation experiment, a distributed blockchain system was constructed based on users' historical behavior data. Users' identity information and behavior data were distributed across different blockchain nodes, which were divided into identity nodes, behavior nodes, and verification nodes. The system adaptively selected nodes to participate in verification based on risk level. For users with high risk level, the system selected all nodes for cross-verification; for users with medium risk level, it selected some nodes for verification; and for users with low risk level, it selected only the current node for verification.

[0115] In the third stage of the simulation experiment, a deep learning model was established. The model adopts a structure combining long short-term memory networks and fully connected networks. It takes user identity information and historical behavior data as input to predict the user's operational behavior in the current and future authentication process. In addition, a transaction authentication strategy was designed in the experiment. The model prediction results are bound to the user's transaction information and a second verification is performed to ensure the security of the transaction. Table 1 compares the verification accuracy and false positive rate of the optional authentication methods with those of related technologies, based on the number of user samples and the simulation process.

[0116] Table 1

[0117]

[0118] As shown in Table 1, this embodiment achieves an accuracy rate of 90.1% and a false positive rate of 9.9% in terms of historical behavior data accuracy. Compared with related technologies, the distributed blockchain system constructed in this embodiment enhances data security and reliability, especially in high-risk situations, where cross-validation further reduces the possibility of false positives. Finally, in the comprehensive risk level assessment, this embodiment achieves an accuracy rate of 89.4% and a false positive rate that decreases from 18.5% to 10.6%. This result demonstrates that deep analysis of user identity information and historical behavior data using deep learning models can not only effectively improve the accuracy of verification but also dynamically adjust risk assessments, reducing the risk caused by misjudgments due to single features. Especially in complex application scenarios, this embodiment exhibits stronger adaptability and anti-interference capabilities.

[0119] also, Figure 4 This is a schematic diagram illustrating the number of authentication attempts for different risk levels according to embodiments of this application, such as... Figure 4 As shown, in In the random sampling of users, the risk level defined in this embodiment can be dynamically adjusted (i.e., low-risk verification frequency is low, saving resources, and high-risk verification frequency is high, ensuring security), and excessive resource consumption is avoided, reducing the burden on the actual application system. Figure 5 This is a schematic diagram illustrating the prediction accuracy of a deep learning model according to an embodiment of this application, such as... Figure 5 As shown, the deep learning model in this embodiment can accurately predict user behavior with an accuracy rate of 100%. By designing corresponding transaction authentication strategies in the model, the transaction success rate can be increased under different transaction types. Figure 6 This is a schematic diagram of the transaction success rate provided in the embodiments of this application, such as... Figure 6 As shown, the high degree of integration between user authentication and transaction verification reduces the probability of risky transactions. Figure 7 This is a schematic diagram illustrating the identity verification success rate and transaction verification success rate according to embodiments of this application, as shown below. Figure 7 As shown, high-accuracy user authentication enhances transaction security, and the combination of authentication and transaction verification reduces risky transaction behaviors. This improves the system's accuracy and security, enhances its operability and reliability in practical applications, and has broad application prospects.

[0120] It should be noted that the steps shown in the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions, and although a logical order is shown in the flowchart, in some cases the steps shown or described may be executed in a different order than that shown here.

[0121] Example 2

[0122] This application also provides an authentication device. It should be noted that the authentication device of this application can be used to execute the authentication method provided in this application. The authentication device provided in this application is described below.

[0123] According to an embodiment of this application, an apparatus for implementing the above-described authentication method is also provided. Figure 8 This is a schematic diagram of an authentication device provided according to an embodiment of this application, such as... Figure 8 As shown, the device includes:

[0124] The acquisition unit 801 is used to acquire the optical features and character information of the user's identity document through optical character recognition, determine the first quality index value of the user's identity information based on the optical features and character information, and determine the risk level of the user's identity information.

[0125] The acquisition unit 802 is used to acquire the user's standard identity information from a preset database, determine the second quality index value of the user's identity information based on the standard identity information and character information, and determine that the user's identity verification has failed if the first quality index value is different from the second quality index value.

[0126] Verification unit 803 is used to verify data consistency through risk level and target blockchain when the first quality indicator value and the second quality indicator value are the same. The target blockchain contains the identity node, behavior node and verification node of the user each time the identity information is verified.

[0127] The first determining unit 804 is used to determine that the user's authentication has failed if the consistency verification fails, and to determine that the user's authentication has succeeded if the consistency verification passes.

[0128] The identity verification device provided in this application embodiment includes: a collection unit 801 that collects optical features and character information of a user's identity document through optical character recognition, determines a first quality index value of the user's identity information based on the optical features and character information, and determines the risk level of the user's identity information; an acquisition unit 802 that acquires the user's standard identity information from a preset database, determines a second quality index value of the user's identity information based on the standard identity information and character information, and determines that the user's identity verification has failed if the first quality index value and the second quality index value are different; and a verification unit 803 that, if the first quality index value and the second quality index value are the same, performs data consistency verification through the risk level and a target blockchain, wherein the target blockchain contains the user's identity node and row information each time the identity information is verified. The system includes nodes and verification nodes; the first determining unit 804 determines user authentication failure if consistency verification fails, and determines user authentication success if consistency verification passes. It evaluates the quality of the ID card image based on quality index values ​​determined by optical features and character information, and performs authentication based on a comparison of the first and second quality index values. This effectively identifies blurry, damaged, or counterfeit documents. Data consistency verification is performed based on the risk level of the identity information and the target blockchain, ensuring the privacy and immutability of sensitive data. This reduces the false judgment rate caused by image quality, thereby improving the accuracy of identification during identity verification via optical character recognition and solving the technical problem of inaccurate identification during identity verification via optical character recognition.

[0129] Optionally, in the identity verification device provided in this application embodiment, the device further includes: a historical behavior data acquisition unit, used to acquire the user's historical behavior data when the user's transaction behavior is detected, wherein the historical behavior data includes the user's historical transaction operation behavior and historical identity verification operation behavior; an input unit, used to input the user's identity information and historical behavior data into a target model, process it to obtain the user's predicted operation behavior within a target period, wherein the predicted operation behavior includes predicted transaction operation behavior and predicted identity verification operation behavior, the target model includes a gated recurrent unit layer, a fully connected network layer and an output layer, the gated recurrent unit layer processes the historical behavior data and outputs a first hidden state vector, the fully connected network layer processes the user's identity information and outputs a second hidden state vector, the first hidden state vector and the second hidden state vector are merged to obtain a merged state vector, and the output layer outputs the predicted operation behavior based on the state vector; and a transaction behavior verification unit, used to verify the user's transaction behavior based on the predicted operation behavior.

[0130] Optionally, in the identity verification device provided in this application embodiment, the transaction behavior verification unit includes: a first acquisition module, used to acquire the user's actual transaction operation behavior and actual identity verification operation behavior within a target period; a judgment module, used to determine whether the predicted transaction operation behavior and the actual transaction operation behavior are the same, and to determine whether the predicted identity verification operation behavior and the actual identity verification operation behavior are the same; a first determination module, used to determine that the transaction behavior verification is successful when the predicted transaction operation behavior and the actual transaction operation behavior are the same, and the predicted identity verification operation behavior and the actual identity verification operation behavior are the same; and a second determination module, used to determine that the transaction behavior verification is unsuccessful when the predicted transaction operation behavior and the actual transaction operation behavior are different, or when the predicted identity verification operation behavior and the actual identity verification operation behavior are different.

[0131] Optionally, in the identity verification device provided in this application embodiment, the acquisition unit 801 includes: a third determining module, used to determine character contrast, edge sharpness, and image noise level in optical features, and determine a sharpness index value based on at least one of character contrast, edge sharpness, and image noise level; a detection module, used to detect whether there are missing field contents in the character information, obtain a field missing detection result, and determine an integrity index value based on the field missing detection result; a first calculation module, used to calculate the similarity between the character information and standard identity information, and determine the similarity as an accuracy index value; and an input module, used to input the sharpness index value, integrity index value, and accuracy index value into a preset function to obtain a first quality index value.

[0132] Optionally, in the identity verification device provided in this application embodiment, the acquisition unit 801 includes: a second acquisition module, used to acquire the user's average quality index value and standard deviation within a historical period, calculate the difference between the first quality index value and the average quality index value, and calculate the ratio of the difference to the standard deviation to obtain the standardized first quality index value; a second calculation module, used to calculate the absolute value of the standardized first quality index value to obtain a target value; and a fourth determination module, used to determine the risk level of the user's identity information as a first level when the target value is less than or equal to a first threshold, determine the risk level of the user's identity information as a second level when the target value is greater than or equal to the first threshold but less than a second threshold, and determine the risk level of the user's identity information as a third level when the target value is greater than or equal to the second threshold, wherein the first threshold is less than the second threshold, the risk level of the first level is lower than the second level, and the risk level of the second level is lower than the third level.

[0133] Optionally, in the identity verification device provided in this application embodiment, the acquisition unit 802 includes: a mapping module, used to map character information into N first numerical feature vectors through a preset character encoding method, and determine N second numerical feature vectors corresponding to standard identity information, where N is a positive integer; a third calculation module, used to calculate the distance parameters between the first numerical feature vectors and the second numerical feature vectors of the same type of identity information, and obtain N distance parameters; and a fourth calculation module, used to calculate the average value of the N distance parameters, and obtain a second quality index value.

[0134] Optionally, in the identity verification device provided in this application embodiment, the verification unit 803 includes: a first verification module, configured to, when the risk level is first level, call the identity node and behavior node of the current time from the target blockchain, verify the consistency between the identity information in the identity node of the current time and the standard identity information, and verify the consistency between the operation behavior in the behavior node of the current time and the historical operation behavior; a second verification module, configured to, when the risk level is second level, call the preset number of identity nodes and behavior nodes from the target blockchain, and perform cross-verification based on the preset number of identity nodes and behavior nodes, wherein the cross-verification compares the consistency of the identity information in all identity nodes and the consistency of the operation behavior in all behavior nodes; and a third verification module, configured to, when the risk level is third level, call all identity nodes and behavior nodes from the target blockchain, and perform cross-verification based on all identity nodes and behavior nodes in the target blockchain.

[0135] It should be noted that the above-mentioned acquisition unit 801, acquisition unit 802, verification unit 803, and first determination unit 804 correspond to steps S201 to S204 in Embodiment 1. The four units and the corresponding steps implement the same instances and application scenarios, but are not limited to the content disclosed in Embodiment 1. It should be noted that the above-mentioned modules or units can be hardware or software components stored in memory (e.g., memory 104) and processed by one or more processors (e.g., processors 102a, 102b, ..., 102n). The above-mentioned modules or units can also be part of the device and run in the computer terminal 10 provided in Embodiment 1.

[0136] Example 3

[0137] Embodiments of this application may provide an electronic device. Figure 9 This is a structural block diagram of an electronic device according to an embodiment of this application. Figure 9 As shown, the electronic device may include: one or more ( Figure 9(Only one is shown) processor 902, memory 904, memory controller, and peripheral interface, wherein the peripheral interface is connected to the radio frequency module, audio module and display.

[0138] The memory can be used to store software programs and modules, such as the program instructions / modules corresponding to the methods and apparatus in the embodiments of this application. The processor executes various functional applications and data processing by running the software programs and modules stored in the memory, thereby implementing the above-described methods. The memory may include high-speed random access memory, and may also include non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some instances, the memory may further include memory remotely located relative to the processor, and these remote memories can be connected to the terminal via a network. Examples of such networks include, but are not limited to, the Internet, corporate intranets, local area networks, mobile communication networks, and combinations thereof.

[0139] The processor can access information and applications stored in the memory via a transmission device to execute the following steps: Acquire optical features and character information of the user's identity document using optical character recognition; determine a first quality index value for the user's identity information based on the optical features and character information, and determine the risk level of the user's identity information; retrieve the user's standard identity information from a preset database; determine a second quality index value for the user's identity information based on the standard identity information and character information; if the first quality index value differs from the second quality index value, determine that the user's identity verification has failed; if the first quality index value and the second quality index value are the same, perform data consistency verification through the risk level and a target blockchain, wherein the target blockchain contains the user's identity node, behavior node, and verification node each time the identity information is verified; if the consistency verification fails, determine that the user's identity verification has failed; if the consistency verification passes, determine that the user's identity verification has succeeded.

[0140] The processor can also access information and applications stored in the memory via a transmission device to perform the following steps: Upon detecting a user's transaction behavior, collect the user's historical behavior data, including historical transaction operations and historical identity verification operations; input the user's identity information and historical behavior data into a target model, process it to obtain the user's predicted operational behavior within a target period, including predicted transaction operations and predicted identity verification operations. The target model includes a gated recurrent unit layer, a fully connected network layer, and an output layer. The gated recurrent unit layer processes the historical behavior data and outputs a first hidden state vector; the fully connected network layer processes the user's identity information and outputs a second hidden state vector; the first and second hidden state vectors are merged to obtain a merged state vector; the output layer outputs the predicted operational behavior based on the state vector; and the user's transaction behavior is verified based on the predicted operational behavior.

[0141] The processor can also invoke information and applications stored in the memory via the transmission device to perform the following steps: obtaining the user's actual transaction operation behavior and actual authentication operation behavior within the target period; determining whether the predicted transaction operation behavior and the actual transaction operation behavior are the same, and determining whether the predicted authentication operation behavior and the actual authentication operation behavior are the same; if the predicted transaction operation behavior and the actual transaction operation behavior are the same, and the predicted authentication operation behavior and the actual authentication operation behavior are the same, determining that the transaction behavior verification is successful; if the predicted transaction operation behavior and the actual transaction operation behavior are different, or if the predicted authentication operation behavior and the actual authentication operation behavior are different, determining that the transaction behavior verification is unsuccessful.

[0142] The processor can also invoke information and applications stored in the memory via a transmission device to perform the following steps: determining character contrast, edge sharpness, and image noise level in optical features, and determining a sharpness index value based on at least one of character contrast, edge sharpness, and image noise level; detecting whether there are missing field contents in the character information, obtaining a field missing detection result, and determining an integrity index value based on the field missing detection result; calculating the similarity between the character information and standard identity information, and determining the similarity as an accuracy index value; inputting the sharpness index value, integrity index value, and accuracy index value into a preset function to obtain a first quality index value.

[0143] The processor can also call the information and application programs stored in the memory through the transmission device to perform the following steps: obtain the user's average quality index value and standard deviation over a historical period; calculate the difference between the first quality index value and the average quality index value; calculate the ratio of the difference to the standard deviation to obtain the standardized first quality index value; calculate the absolute value of the standardized first quality index value to obtain the target value; if the target value is less than or equal to a first threshold, determine the risk level of the user's identity information as level one; if the target value is greater than or equal to the first threshold but less than a second threshold, determine the risk level of the user's identity information as level two; if the target value is greater than or equal to the second threshold, determine the risk level of the user's identity information as level three, wherein the first threshold is less than the second threshold, the risk level of level one is lower than that of level two, and the risk level of level two is lower than that of level three.

[0144] The processor can also call the information and application stored in the memory through the transmission device to perform the following steps: map character information into N first numerical feature vectors through a preset character encoding method, determine N second numerical feature vectors corresponding to standard identity information, where N is a positive integer; calculate the distance parameters between the first and second numerical feature vectors of the same type of identity information to obtain N distance parameters; calculate the average value of the N distance parameters to obtain the second quality index value.

[0145] The processor can also invoke information and applications stored in the memory via a transmission device to perform the following steps: Under the first risk level, invoke the current-time identity nodes and behavior nodes from the target blockchain, verify the consistency of the identity information in the current-time identity nodes with the standard identity information, and verify the consistency of the operational behavior in the current-time behavior nodes with historical operational behavior; Under the second risk level, invoke a preset number of identity nodes and behavior nodes from the target blockchain, and perform cross-validation based on the preset number of identity nodes and behavior nodes, wherein the cross-validation compares the consistency of identity information in all identity nodes and the consistency of operational behavior in all behavior nodes; Under the third risk level, invoke all identity nodes and behavior nodes from the target blockchain, and perform cross-validation based on all identity nodes and behavior nodes in the target blockchain.

[0146] This application provides a scheme that uses optical character recognition to collect optical features and character information from a user's identity document; determines a first quality index value of the user's identity information based on the optical features and character information; determines the risk level of the user's identity information; retrieves the user's standard identity information from a preset database; determines a second quality index value of the user's identity information based on the standard identity information and character information; if the first quality index value differs from the second quality index value, the user's identity verification is deemed to have failed; if the first quality index value and the second quality index value are the same, data consistency verification is performed using the risk level and a target blockchain, wherein the target blockchain contains the user's identity node, behavior node, and verification node each time the user verifies their identity information; if the consistency verification fails, the user's identity verification is deemed to have failed; if the consistency verification passes, the user's identity verification is deemed to have succeeded. The quality of ID card images is evaluated by using quality index values ​​determined based on optical features and character information. Identity verification is performed by comparing the first quality index value with the second quality index value. This effectively identifies blurry, damaged, or counterfeit documents. Data consistency verification is performed based on the risk level of the identity information and the target blockchain, ensuring the privacy and immutability of sensitive data. This reduces the false judgment rate caused by image quality, thereby improving the accuracy of identification during identity verification through optical character recognition and solving the technical problem of inaccurate identification when using optical character recognition for identity verification.

[0147] Those skilled in the art will understand that Figure 9 The structure shown is for illustrative purposes only. Electronic devices can also be smartphones, tablets, handheld computers, mobile internet devices (MIDs), PADs, and other terminal devices. Figure 9 This does not limit the structure of the aforementioned electronic device. For example, electronic devices may also include components that are more... Figure 9 The more or fewer components shown (such as network interfaces, display devices, etc.), or having the same Figure 9 The different configurations shown.

[0148] Those skilled in the art will understand that all or part of the steps in the various methods of the above embodiments can be implemented by a program instructing the hardware related to the terminal device. The program can be stored in a computer-readable storage medium, which may include: flash drive, read-only memory (ROM), random access memory (RAM), disk or optical disk, etc.

[0149] Example 4

[0150] Embodiments of this application also provide a storage medium. Optionally, in this embodiment, the storage medium can be used to store the program code executed by the authentication method provided in Embodiment 1.

[0151] Optionally, in this embodiment, the storage medium may be located in any computer terminal in a group of computer terminals in a computer network, or in any mobile terminal in a group of mobile terminals.

[0152] This application also provides a computer program product that, when executed on a data processing device, is suitable for performing authentication method steps.

[0153] The sequence numbers of the embodiments in this application are for descriptive purposes only and do not represent the superiority or inferiority of the embodiments.

[0154] In the above embodiments of this application, the descriptions of each embodiment have different focuses. For parts not described in detail in a certain embodiment, please refer to the relevant descriptions of other embodiments.

[0155] In the several embodiments provided in this application, it should be understood that the disclosed technical content can be implemented in other ways. The device embodiments described above are merely illustrative; for example, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the displayed or discussed mutual coupling, direct coupling, or communication connection may be through some interfaces; the indirect coupling or communication connection between units or modules may be electrical or other forms.

[0156] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0157] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0158] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as a USB flash drive, read-only memory (ROM), random access memory (RAM), portable hard drive, magnetic disk, or optical disk.

[0159] The above description is only a preferred embodiment of this application. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the principle of this application, and these improvements and modifications should also be considered within the scope of protection of this application.

Claims

1. An authentication method, characterized in that, include: Optical character recognition is used to collect the optical features and character information of a user's identity document. Based on the optical features and character information, a first quality index value of the user's identity information is determined, and the risk level of the user's identity information is determined. The system retrieves the user's standard identity information from a preset database, determines a second quality index value for the user's identity information based on the standard identity information and the character information, and determines that the user's identity verification has failed if the first quality index value is different from the second quality index value. When the first quality indicator value is the same as the second quality indicator value, data consistency verification is performed through the risk level and the target blockchain, wherein the target blockchain contains the user's identity node, behavior node and verification node each time the user verifies identity information; If the consistency verification fails, the user's authentication is determined to have failed; if the consistency verification passes, the user's authentication is determined to have succeeded.

2. The method according to claim 1, characterized in that, After confirming successful user authentication, the method further includes: Upon detecting the user's transaction behavior, the user's historical behavior data is collected, wherein the historical behavior data includes the user's historical transaction operation behavior and historical identity verification operation behavior; The user's identity information and historical behavior data are input into the target model, and the predicted operation behavior of the user within the target period is obtained. The predicted operation behavior includes predicted transaction operation behavior and predicted identity verification operation behavior. The target model includes a gated recurrent unit layer, a fully connected network layer, and an output layer. The gated recurrent unit layer processes the historical behavior data and outputs a first hidden state vector. The fully connected network layer processes the user's identity information and outputs a second hidden state vector. The first hidden state vector and the second hidden state vector are merged to obtain a merged state vector. The output layer outputs the predicted operation behavior based on the state vector. The user's transaction behavior is verified based on the predicted operational behavior.

3. The method according to claim 2, characterized in that, Verifying the user's transaction behavior based on the predicted operational behavior includes: Obtain the user's actual transaction and identity verification behavior within the target period; Determine whether the predicted transaction operation behavior and the actual transaction operation behavior are the same, and determine whether the predicted identity verification operation behavior and the actual identity verification operation behavior are the same; If the predicted transaction behavior and the actual transaction behavior are the same, and the predicted identity verification behavior and the actual identity verification behavior are the same, then the transaction behavior is determined to be verified successfully. If the predicted transaction behavior differs from the actual transaction behavior, or if the predicted identity verification behavior differs from the actual identity verification behavior, the transaction behavior verification is determined to have failed.

4. The method according to claim 1, characterized in that, The first quality index value for determining the user's identity information based on the optical features and the character information includes: Determine the character contrast, edge sharpness, and image noise level among the optical features, and determine a sharpness index value based on at least one of the character contrast, edge sharpness, and image noise level; Detect whether there are missing field contents in the character information, obtain the field missing detection result, and determine the integrity index value based on the field missing detection result; Calculate the similarity between the character information and the standard identity information, and determine the similarity as an accuracy index value; The first quality index value is obtained by inputting the clarity index value, the integrity index value, and the accuracy index value into a preset function.

5. The method according to claim 1, characterized in that, Determining the risk level of the user's identity information includes: Obtain the average quality index value and standard deviation of the user within the historical period, calculate the difference between the first quality index value and the average quality index value, and calculate the ratio of the difference to the standard deviation to obtain the standardized first quality index value. Calculate the absolute value of the first quality index value after the standardization process to obtain the target value; If the target value is less than or equal to a first threshold, the risk level of the user's identity information is determined to be a first level. If the target value is greater than or equal to the first threshold but less than a second threshold, the risk level of the user's identity information is determined to be a second level. If the target value is greater than or equal to the second threshold, the risk level of the user's identity information is determined to be a third level. The first threshold is less than the second threshold, the risk level of the first level is lower than that of the second level, and the risk level of the second level is lower than that of the third level.

6. The method according to claim 1, characterized in that, The second quality index value for determining the user's identity information based on the standard identity information and the character information includes: The character information is mapped into N first numerical feature vectors by a preset character encoding method, and the N second numerical feature vectors corresponding to the standard identity information are determined, where N is a positive integer; Calculate the distance parameters between the first numerical feature vector and the second numerical feature vector of the same type of identity information to obtain N distance parameters; The average value of the N distance parameters is calculated to obtain the second quality index value.

7. The method according to claim 5, characterized in that, Data consistency verification through the aforementioned risk level and target blockchain includes: When the risk level is the first level, the identity node and the behavior node at the current time are called from the target blockchain to verify the consistency between the identity information in the identity node at the current time and the standard identity information, and to verify the consistency between the operation behavior in the behavior node at the current time and the historical operation behavior. When the risk level is the second level, a preset number of identity nodes and behavior nodes are called from the target blockchain, and cross-validation is performed based on the preset number of identity nodes and behavior nodes, wherein the cross-validation compares the consistency of identity information in all identity nodes and the consistency of operational behavior in all behavior nodes; When the risk level is the third level, all the identity nodes and behavior nodes are called from the target blockchain, and cross-validation is performed based on all the identity nodes and behavior nodes in the target blockchain.

8. An authentication device, characterized in that, include: The acquisition unit is used to acquire optical features and character information of a user's identity document through optical character recognition, determine a first quality index value of the user's identity information based on the optical features and the character information, and determine the risk level of the user's identity information. The acquisition unit is configured to acquire the user's standard identity information from a preset database, determine a second quality index value of the user's identity information based on the standard identity information and the character information, and determine that the user's identity verification has failed if the first quality index value is different from the second quality index value. The verification unit is used to perform data consistency verification through the risk level and the target blockchain when the first quality indicator value and the second quality indicator value are the same. The target blockchain includes the user's identity node, behavior node and verification node each time the user verifies identity information. The first determining unit is configured to determine that the user's authentication has failed if the consistency verification fails, and to determine that the user's authentication has succeeded if the consistency verification passes.

9. An electronic device, characterized in that, include: Memory, which stores executable programs; A processor for running the program, wherein the program, when running, performs the authentication method according to any one of claims 1 to 7.

10. A computer program product comprising computer instructions, characterized in that, When the computer instructions are executed by the processor, they implement the steps of the authentication method according to any one of claims 1 to 7.

Images