Face scanning payment data transmission method and system

Abstract

The invention provides a face-scanning payment data transmission method and system, and relates to the technical field of computers, and the method comprises the steps: obtaining payment data and scene data; performing face feature extraction processing according to the user face image data to obtain user face feature data; performing feature association encryption processing according to the user face feature data, the payment request data and the equipment identification data to obtain an equipment binding encrypted data packet; based on payment position data dynamic analysis equipment, binding the regional importance of the encrypted data packet, and obtaining a position block-based data packet; performing decompression and action verification processing according to the data packet based on the position block to obtain a recombined data set passing the action verification; and performing payment verification and feedback processing according to the recombined data set to obtain a final payment verification result. According to the invention, through combination of face feature extraction and payment timestamp verification, the accuracy and robustness of identity verification are significantly improved, and the problems of identity counterfeiting and data tampering are effectively avoided.

Description

Technical Field

[0001] The present invention relates to the field of computer technology, and in particular to a face-scanning payment data transmission method and system thereof. Background Art

[0002] With the rapid development of mobile payment, face payment, as a convenient and secure payment method, has gradually become an important part of the modern payment field. At present, face payment technology mainly relies on face recognition algorithms for identity authentication and ensures the security of payment information by transmitting encrypted data packets. However, most of the existing face payment systems have some security risks and operational complexity. Traditional technologies often rely only on static encryption methods and basic image processing algorithms, which are easily affected by network delays, disordered transmission order, and identity information forgery. The emergence of these problems not only affects the payment experience, but also the security and accuracy of payment requests cannot be fully guaranteed in high-concurrency or complex network environments. The solutions of existing technologies usually rely on a single identity authentication or data encryption method, but lack comprehensive verification of multiple dimensions such as time, location, and transmission order in payment scenarios, resulting in the inability to effectively detect and prevent security risks such as identity theft, transaction tampering, and replay attacks in some cases. Summary of the invention

[0003] The purpose of the present invention is to provide a face-scanning payment data transmission method and system to improve the above problems. In order to achieve the above purpose, the technical solution adopted by the present invention is as follows:

[0004] In a first aspect, the present application provides a face-scanning payment data transmission method, comprising:

[0005] Acquire payment data and scenario data, wherein the payment data includes user facial image data, payment request data and device identification data, and the scenario data includes payment timestamp data, payment location data and transaction action data;

[0006] Performing facial feature extraction processing according to the user facial image data, extracting the user facial features and verifying the time validity of the image data in combination with the payment timestamp data, to obtain user facial feature data that passes the timing verification;

[0007] Perform feature association encryption processing according to the user facial feature data, the payment request data and the device identification data, generate a unique device key through the device identification data to encrypt the user facial feature data, and bind the encrypted data to the payment request data to obtain a device-bound encrypted data packet;

[0008] Dynamically analyzing the regional importance of the device-bound encrypted data packet based on the payment location data, marking the transmission priority of data blocks of different importance and attaching location-related metadata to obtain data packets based on location blocks;

[0009] Decompressing and verifying the action according to the position-based block data packet, decompressing the data packet and verifying the integrity of the data block transmission sequence according to the transaction action data, decrypting the data block using the device key, and obtaining a reorganized data set that passes the action verification;

[0010] Payment verification and feedback processing are performed according to the reorganized data set to obtain a final payment verification result.

[0011] In a second aspect, the present application also provides a face-scanning payment data transmission system, including:

[0012] An acquisition module, used to acquire payment data and scenario data, wherein the payment data includes user facial image data, payment request data and device identification data, and the scenario data includes payment timestamp data, payment location data and transaction action data;

[0013] An extraction module, used to perform facial feature extraction processing based on the user facial image data, extract the user facial features, and verify the time validity of the image data in combination with the payment timestamp data to obtain the user facial feature data that passes the timing verification;

[0014] An encryption module is used to perform feature association encryption processing according to the user facial feature data, the payment request data and the device identification data, generate a unique device key through the device identification data to encrypt the user facial feature data, and bind the encrypted data with the payment request data to obtain a device-bound encrypted data packet;

[0015] A marking module dynamically analyzes the regional importance of the device-bound encrypted data packet based on the payment location data, marks the transmission priority of data blocks of different importance and attaches location-related metadata to obtain a data packet based on location blocks;

[0016] A decompression module, used to perform decompression and action verification processing according to the data packet based on the position block, decompress the data packet and verify the integrity of the data block transmission sequence according to the transaction action data, and use the device key to decrypt the data block to obtain a reorganized data set that passes the action verification;

[0017] The verification module is used to perform payment verification and feedback processing according to the reorganized data set to obtain a final payment verification result.

[0018] The beneficial effects of the present invention are:

[0019] The present invention significantly improves the accuracy and robustness of identity authentication by combining facial feature extraction with payment timestamp verification, and effectively avoids identity forgery and data tampering problems; it uses payment location data, transmission priority tags and transmission sequence verification models to ensure sequential consistency and legitimacy during data transmission, thereby avoiding payment anomalies caused by network delays or transmission errors. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required for use in the embodiments are briefly introduced below. It should be understood that the following drawings only show certain embodiments of the present invention and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without creative work.

[0021] Figure 1 This is a flow chart of the face-swiping payment data transmission method in the present invention;

[0022] Figure 2 This is a schematic diagram of the structure of the face-scanning payment data transmission system in the present invention;

[0023] Figure 3 This is a schematic diagram of the structure of the face-scanning payment data transmission device in the present invention. DETAILED DESCRIPTION

[0024] In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, rather than all of the embodiments. The components of the embodiments of the present invention generally described and shown in the drawings here can be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present invention provided in the drawings is not intended to limit the scope of the claimed invention, but merely represents selected embodiments of the present invention. Based on the embodiments in the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0025] It should be noted that similar reference numerals and letters represent similar items in the following drawings, so once an item is defined in one drawing, it does not need to be further defined and explained in the subsequent drawings. At the same time, in the description of the present invention, the terms "first", "second", etc. are only used to distinguish the description and cannot be understood as indicating or implying relative importance.

[0026] Embodiment 1:

[0027] This embodiment provides a face-scanning payment data transmission method. Figure 1 , which includes steps S100 to S600.

[0028] Step S100, obtaining payment data and scenario data, wherein the payment data includes user facial image data, payment request data and device identification data, and the scenario data includes payment timestamp data, payment location data and transaction action data;

[0029] It should be noted that in the payment data part, the user's facial image data is collected in real time through high-precision camera equipment, usually combined with liveness detection technology (such as detection based on infrared sensing or depth cameras) to prevent attackers from using photos or videos to deceive the payment system. The payment request data comes directly from the user's payment behavior, such as order amount, payment method, etc., and is transmitted through a unified payment interface protocol specification to ensure the consistency of interaction with the payment gateway. The device identification data uses the device's unique hardware identification (such as IMEI, MAC address or device fingerprint). The characteristic of this data is that it cannot be tampered with, thus providing a basis for subsequent device binding encryption.

[0030] In the scenario data part, the acquisition of payment timestamp data is not only used to record the transaction time, but also to verify the time validity of the data in subsequent steps to ensure the legitimacy of the payment behavior. The accuracy of the timestamp is required to be high, usually in milliseconds, to avoid data verification failure due to time errors. Payment location data is achieved through GPS module or Wi-Fi hotspot positioning, and some payment systems will also combine multi-source positioning technology (such as cellular network positioning) to improve positioning accuracy. Location data is crucial in the subsequent regional priority calculation, which reflects the geographical restrictions of the payment scenario. Transaction action data is the dynamic capture of user interaction behaviors during the payment process, such as gesture confirmation, click operations, or device movement trajectories. These data can be collected through sensors (such as accelerometers or touch screens) to verify the behavioral consistency of payment operations and prevent unauthorized operations.

[0031] Step S200: Perform facial feature extraction processing based on the user's facial image data, extract the user's facial features, and verify the time validity of the image data in combination with the payment timestamp data to obtain the user's facial feature data that passes the timing verification;

[0032] In the process of facial feature extraction, the user's facial image data must first be preprocessed, including noise removal, illumination compensation, and face alignment, to ensure that the feature extraction is not interfered by the external environment. Feature extraction uses a deep learning model (such as ResNet or MobileNet architecture based on convolutional neural networks) to extract high-dimensional embedded features of key facial areas through multi-layer feature mapping. These features are standardized high-dimensional vectors with good discriminability and compactness, which are convenient for subsequent calculations. The introduction of payment timestamps enables this step to have dynamic verification capabilities in the time dimension. Timestamp verification in payment scenarios is usually performed by comparing the deviation between the image data acquisition time and the payment request time. When the image acquisition time exceeds the preset time window, the system will mark it as invalid and reject the payment request. This mechanism effectively prevents security risks such as replay attacks, in which attackers may use acquired legitimate images to bypass face recognition. The technical implementation of time validity verification adopts an event-driven mode, by attaching a timestamp to each image and digitally signing it to prevent time data from being tampered with. Combining facial feature extraction with timestamp verification not only enhances the real-time performance of feature extraction, but also binds payment behavior to time characteristics, solving the problem of traditional face-swiping payment that relies solely on static features and is easily deceived.

[0033] Step S300: Perform feature association encryption processing based on the user's facial feature data, the payment request data and the device identification data, generate a unique device key through the device identification data to encrypt the user's facial feature data, and bind the encrypted data to the payment request data to obtain a device-bound encrypted data packet;

[0034] It is understandable that the introduction of device identification data makes each encryption key generation bound to the current device, realizing the dynamic and uniqueness of the key, greatly enhancing the security of encryption. In addition, binding the encrypted user facial feature data with the payment request data not only protects the user's biometric privacy, but also ensures the strong correlation between payment behavior and user identity through a structured data format. The high efficiency of symmetric encryption and the optimization of block encryption strategies also make the encryption process suitable for the performance requirements of real-time payment scenarios.

[0035] Step S400: dynamically analyzing the regional importance of the device-bound encrypted data packet based on the payment location data, marking the transmission priority of data blocks of different importance and attaching location-related metadata to obtain data packets based on location blocks;

[0036] It should be noted that this step introduces payment location data to dynamically analyze the importance of data packets, and uses the regional priority matrix and distance decay model to accurately evaluate the transmission importance of different data blocks. Combined with the additional operation of location-related metadata, the data packet has regional dynamic characteristics and can adapt to the payment scenario requirements of different geographical locations. The priority marking strategy optimizes data transmission efficiency through resource allocation, while improving the security and reliability of payment data. Through the implementation of this step, the transmission of data packets has been significantly optimized, and high-priority data blocks (such as facial feature data) can be transmitted first, thereby speeding up the response speed of payment verification. The additional metadata provides geographic location context support for subsequent verification and decompression processing, effectively preventing location fraud or cross-regional attacks.

[0037] Step S500: decompress and verify the data packets based on the location blocks, decompress the data packets and verify the integrity of the data block transmission sequence based on the transaction action data, decrypt the data blocks using the device key, and obtain a reorganized data set that passes the action verification;

[0038] This step combines the dual processing of decompression and action verification, and ensures the transmission integrity of high-priority data through a block decoding strategy; the dynamic time warping algorithm adapts to slight offsets in the transaction action time sequence and enhances the robustness of transmission order verification. In addition, block-by-block decryption combined with device keys ensures the security and uniqueness of the data and prevents the data from being illegally intercepted or tampered with during transmission. Through decompression processing, the original data block is efficiently restored, while retaining the block and priority information, and optimizing the data transmission structure. The action verification process effectively prevents tampering or loss of the data transmission order and ensures data integrity.

[0039] Step S600: Perform payment verification and feedback processing according to the reorganized data set to obtain a final payment verification result.

[0040] This step establishes a three-dimensional, multi-dimensional payment verification system by matching high-dimensional vectors of facial features and combining multi-layer verification of payment scenario characteristics (such as time, location, and action). Methods such as location verification and time verification enhance the ability to detect forged data or cross-border payment behaviors. The introduction of scenario data such as time and location adds dynamism and flexibility to payment verification, especially in complex scenarios (such as remote payment or mobile payment), showing significant protection capabilities; the optimization of the payment feedback mechanism ensures that users can obtain clear reasons for failure and guidance in failure scenarios, improving the user experience.

[0041] Embodiment 2:

[0042] The difference between this embodiment and Embodiment 1 is that step S200 includes steps S210 to S240.

[0043] Step S210, performing feature point detection processing according to the user's facial image data, extracting key facial regions based on a multi-scale pyramid feature extraction method, and obtaining candidate facial feature region data;

[0044] In the feature point detection processing stage, the user's facial image data is first preprocessed, including grayscale, illumination compensation and noise removal, to improve the robustness of detection. Then, a feature point detection algorithm (such as a key point detection network based on deep learning, such as HRNet or MTCNN) is used to locate the key points of the facial area. Key points include eyebrows, eyes, nose tips, corners of the mouth and other points with obvious geometric features. The accurate positioning of these points is crucial for the subsequent cropping and feature extraction of facial areas. The feature extraction method based on multi-scale pyramid further optimizes the extraction process of key facial areas. The core idea of ​​the multi-scale pyramid method is to process the original image with multi-scale resolution through a pyramid structure to capture different levels of facial features. The high-resolution layer is used to extract facial details, such as wrinkles and pores, while the low-resolution layer is used to capture global structural features, such as facial contours and the relative position relationship of facial features. Through layer-by-layer scanning and fusion, the system can strike a balance between spatial resolution and feature details to ensure that the extracted candidate facial areas are complete and discriminative. The extracted candidate facial feature region data is the basis for subsequent high-dimensional feature encoding and is stored in the form of cropped images or feature point coordinates of the candidate region. Confidence scores are attached to filter out areas that may be affected by occlusion or low-quality images.

[0045] Step S220, performing feature encoding processing according to the candidate facial feature region data, generating a standardized facial feature vector through deep embedding representation, and obtaining preliminary facial feature vector data;

[0046] The feature encoding process takes the candidate facial feature region data as input. First, the input data is normalized to ensure that the input data meets the preset requirements of the deep model. Subsequently, a pre-trained deep neural network (such as FaceNet, ArcFace or ResNet-50) is used to extract high-dimensional embedding of the feature region. This type of network extracts local features of the face through multiple layers of convolution and pooling operations, and then maps these features to a fixed-dimensional feature space through a fully connected layer. The generated embedding representation is a set of standardized high-dimensional vectors with fixed dimensions, which are then normalized so that the vectors are distributed on the unit sphere. This normalization process ensures the direct comparability of feature vectors of different images in the feature space. In the process of deep embedding representation, the model learns highly discriminative and robust feature expression capabilities through a large amount of training data (including facial images of various ages, genders, races and environmental changes). Specifically, these feature vectors retain the uniqueness of the face (such as facial features, texture features), while having high anti-interference ability to interference factors such as illumination changes, angle changes and occlusion. The generated preliminary facial feature vector data is not only discriminative, but also retains a certain degree of compactness and versatility, which is convenient for subsequent storage and comparison.

[0047] Step S230: Perform time matching processing according to the preliminary facial feature vector data, filter valid image frames through payment timestamp data, and obtain preliminary facial feature vector data with valid time matching;

[0048] Specifically, the above screening process can eliminate image frames with too long delays (caused by transmission or processing delays) and historical feature vector data that may be used by attackers to replay attacks. In order to improve processing efficiency, the input feature vectors are sorted by timestamp, and the binary search method is used to quickly find feature data that meets the time window, thereby optimizing the screening speed. In addition to the hard screening of the time window, it can also be combined with the time interval smoothing of the frame, and the multi-frame feature vectors can be fused by weighted averaging to further improve the quality of the time matching feature vector.

[0049] Step S240: perform noise verification processing based on the preliminary facial feature vector data with effective time matching, calculate the cosine similarity between the feature vector and the standard distribution based on the distribution of the preliminary feature vector in the feature space, eliminate abnormal features that exceed the set threshold, and obtain user facial feature data that passes the timing verification.

[0050] It should be noted that this step can accurately remove abnormal features that deviate greatly from the standard distribution through feature space verification based on cosine similarity, thereby improving the overall quality of facial feature data.

[0051] Embodiment 3:

[0052] The difference between this embodiment and embodiment 1 or 2 is that step S300 includes steps S310 to S340.

[0053] Step S310: performing key generation processing according to the device identification data, and obtaining a device key by performing dynamic key calculation using a preset random factor;

[0054] First, the device identification data is regarded as a basic input that cannot be tampered with and has uniqueness. The identification data of each device is fixed at the hardware level, which can provide device-level exclusivity for encryption. By using the device identification data, the payment system can ensure that the generated keys are isolated from each other between different devices. Even if an attacker obtains the key of a certain device, it cannot be applied to other devices. Secondly, the introduction of random factors is an important means to achieve the dynamic nature of the key. The random factor is usually generated by dynamic data in the payment scenario, such as the payment timestamp, the current operating parameters of the device, or a randomly generated salt value. The role of the random factor is to dynamically perturb the static device identification data, so that the key generated for each payment operation will be different even on the same device. This can effectively prevent replay attacks or cracking based on static keys. In the key generation process, the device identification data and the random factor are combined by a specific calculation method. This process includes obfuscation and encryption processing to ensure that the generated key has high entropy, that is, the content of the key is sufficiently random and difficult to predict. Preferably, the calculation method used is a hybrid method of hash calculation or cryptographic security, which can fully utilize the uniqueness of the device identification data and ensure that the key changes dynamically under the perturbation of the random factor.

[0055] Step S320: Encode and encrypt the user's facial feature data using the device key, and obtain encrypted user facial feature data by processing different regions of the feature data one by one;

[0056] In this step, the user's facial feature data is usually processed into a high-dimensional feature vector, which represents the core expression of the user's facial features. To ensure security, the feature vector is first processed in blocks, and the entire high-dimensional vector is divided into multiple logical regions, each of which contains a set of dimensions or values ​​of specific feature points. The purpose of this block processing method is to refine the encryption process, adjust the encryption strategy according to the importance of the block, and improve the protection strength of specific areas.

[0057] Step S330: Binding the encrypted user facial feature data with the payment request data, and obtaining a binding data packet by constructing a JSON data structure and serializing it;

[0058] It is understandable that JSON (JavaScript Object Notation) is a lightweight data exchange format, and its readability and parsing efficiency are very suitable for the needs of efficient data transmission in payment scenarios. In the process of building the binding data package, the system will organize the encrypted user facial feature data and the payment request data into a multi-layer nested JSON structure. After the JSON data structure is built, it is converted into a byte stream or string format through serialization operations for easy network transmission or storage. The serialization process encodes the nested data structure into a standard string representation while retaining its complete hierarchy and field information. Through the nested design of the JSON data structure, the encrypted user facial feature data and the payment request data are bound one-to-one, ensuring a strong correlation between the user identity and the payment behavior, and eliminating the risk of identity forgery.

[0059] Step S340: perform integrity verification processing based on the intermediate data packet and device identification data, generate a verification value by using a hash algorithm and encapsulate it, verify the source authenticity and transmission integrity of the data packet in combination with the uniqueness of the device identification, and encapsulate it to obtain a device-bound encrypted data packet.

[0060] The characteristics of hash values ​​are unidirectional and collision-resistant. Even if any bit of data in the intermediate data packet is tampered with, the generated hash value will change significantly, so that data tampering can be accurately detected. The device identification data is used as an additional input and participates in the hash calculation together with the intermediate data packet content. The generated hash value not only reflects the integrity of the data packet, but also binds to a specific device, ensuring that the data packet can only be generated and verified on a specific device.

[0061] Further, step S400 includes step S410 to step S440.

[0062] Step S410: Match the payment location data with a preset geographic priority matrix, and obtain the regional weight by calculating the importance of different locations to the payment data;

[0063] Specifically, this step first analyzes the distance between the current payment location and the high-priority location, and uses distance decay to incorporate proximity into the weight calculation to ensure that payment locations close to high-priority areas receive higher weights. At the same time, historical payment weights are combined to measure the activity of these areas in past transactions to further improve the accuracy of the calculation results. In addition, a risk scoring mechanism is introduced to dynamically adjust the final weight by analyzing the security level of the current payment location (such as fraud rate or abnormal transaction frequency), so that higher priority protection can be assigned to high-risk areas. The generation of regional weights not only reflects the importance of geographic location to payment behavior, but also integrates historical data and dynamic security, providing a strong basis for subsequent data priority marking and blocking strategies. The calculation formula for regional weights is:

[0064]

[0065] Where (x, y) represents the latitude and longitude coordinates of the current payment location; W region (x, y) represents the regional weight of the current payment location (x, y); k represents the location point number in the geographic priority matrix; m represents the total number of all known priority locations in the geographic priority matrix; R k represents the basic priority score of the k-th known position; d((x,y),(x k ,y k )) indicates the current payment position (x, y) and the kth high priority position (x k ,y k ) geographical distance; σ represents the hyperparameter that controls the distance decay rate; H k represents the historical payment weight of the kth high priority position; l represents the index variable used for normalization calculation; p represents the total number of reference points involved in normalization; α represents the weight adjustment factor of the risk score; S(x,y) represents the risk score of the current payment position (x,y).

[0066] Step S420: parse the device-bound encrypted data packet in blocks based on the regional weight, and obtain an importance score for each data block by quantitatively calculating the importance of each data block;

[0067] It should be noted that the basis of logical segmentation is mainly based on the characteristics of the payment scenario and the functional requirements of the data: the device-bound encrypted data packet usually contains encrypted user facial feature data, payment request data, device metadata and auxiliary information (such as timestamp, location metadata). The first step of segmentation parsing is to identify the functional role of the data block in the payment process. For example, facial feature data is the core of identity authentication and determines the legitimacy of payment, so it is defined as a high priority; payment request data directly reflects the specific content and amount of the transaction and is a key link in transaction verification, so its priority is lower than facial data; device metadata and auxiliary information mainly provide support for payment integrity verification and traceability, and are of lower priority in the process. Regional weights provide a basis for dynamic adjustment of data block priorities based on the impact of payment locations on data security and real-time performance. For example, in high-priority areas, the weight of core data blocks (such as facial feature data) will be further increased, while low-priority areas may reduce protection requirements for device metadata. The segmentation logic not only reflects the role relationship of data in the payment process, but also dynamically adjusts the weight in combination with payment scenarios to achieve priority transmission and processing of high-value data, ensuring a balance between security and efficiency. This block parsing and logical distribution mechanism greatly improves the face-scanning payment system's ability to adapt to scene changes, while optimizing the resource allocation strategy for data transmission.

[0068] Step S430: Perform transmission priority marking processing according to the importance score, by sorting the data blocks from high to low according to the importance score and assigning priority tags to obtain a set of data blocks with transmission priority tags;

[0069] Specifically, first, the data blocks are sorted in descending order according to the importance score to ensure that the most critical data blocks (such as encrypted user facial feature data and payment request data) are ranked first. These data directly determine the core links of user identity authentication and transaction verification. Secondly, priority tags are assigned to the sorted data blocks. The granularity of the tags is divided into high priority, medium priority and low priority according to system requirements, or more refined multi-level priority tags are used. The priority tag allocation logic is not only based on the importance score, but also needs to be combined with the real-time requirements of the payment scenario. For example, in a payment scenario with dense traffic, high-priority data blocks need to be transmitted immediately to ensure payment response time, while low-priority data blocks can be delayed or transmitted in batches. The final set of data blocks with priority tags forms an orderly transmission queue to ensure that the most critical data can be transmitted first when the network bandwidth is limited or the scenario pressure is high. This priority tagging mechanism significantly improves the response speed and security assurance capabilities in complex scenarios through dynamic resource allocation and intelligent transmission strategies.

[0070] Step S440: embed payment location metadata and priority tags for each data block based on the data block set, and encapsulate them into a final data packet.

[0071] It is understandable that the final data packet is composed of multiple data blocks containing metadata and priority tags, which are arranged in the data packet in order of priority to form a multi-dimensional, structured transmission unit. This encapsulation method ensures that the payment system can give priority to high-priority data in different network environments and optimize resource utilization efficiency. At the same time, by embedding payment location metadata, it strengthens the relevance and dynamic nature of the payment scenario, laying a safe and efficient foundation for the decomposition, decryption and processing of data packets at the receiving end.

[0072] Embodiment 4:

[0073] The only difference between this embodiment and any one of embodiments 1-3 is that step S500 includes steps S510 to S530.

[0074] Step S510: decompress the data packet based on the position block, extract the position-related metadata and priority tag in the data packet to perform the decompression operation, and use the block decoding algorithm to restore the compressed data block by block to obtain the decompressed data packet;

[0075] Specifically, the first step of the decompression operation is to extract the location-related metadata and priority tags in the data packet. The location-related metadata includes the geographic coordinates of the payment location, regional weight information, and location information for data organization, such as the offset position or block index of the data block in the compressed packet. The priority tag indicates the order of decompression for the data block, which can restore the high-priority data blocks first, ensuring that key data (such as facial feature data and payment request data) can be quickly used for verification. This design is particularly important in complex payment scenarios. For example, when the network transmission conditions are poor, only high-priority data blocks can be decompressed to meet the immediacy requirements. Subsequently, the decompression process uses a block decoding algorithm to process the data packet block by block. The block decoding algorithm uses each data block as the smallest unit and recovers according to the compression strategy and the structure within the block. For example, if the front-end compression uses a block compression method based on wavelet transform or entropy coding, the decompression process needs to reversely restore the data block by block. During decoding, the priority tag guides the sequential decompression step by step, while ensuring that the index information in the location metadata accurately points to the storage location of each data block in the compressed packet, thereby improving the decompression efficiency and reducing the computational overhead. After decompression is completed, a complete decompressed data packet is generated, whose internal structure is consistent with the original data packet before compression, but it is not decrypted. The design of the decompression process has two major advantages: on the one hand, through priority tags and block processing, resource utilization can be optimized, and important data can still be quickly obtained when transmission bandwidth or computing power is limited; on the other hand, location-related metadata ensures the structural consistency and integrity of the decompressed data, so that data will not be lost or reassembled during the restoration process.

[0076] Step S520: Perform data block transmission sequence verification processing according to the decompressed data packet and the transaction action data, extract the time series information and compare it with the priority label and build a transmission sequence verification model to obtain a decompressed data packet that passes the verification;

[0077] The transmission order verification model identifies abnormal behaviors during transmission, such as disordered, missing, or redundant data blocks, by dynamically comparing time series and priority tags. The logic of the verification model includes: checking whether the timestamp of the data block is consistent with the time series range of the transaction action data, whether the priority tag is consistent with the decompression process, and whether there are skipped or repeated data blocks. For data blocks with deviations, the model will mark them as warnings or errors based on the degree and importance of the deviation. For example, the loss of high-priority data blocks will be considered a serious problem, while the misalignment of low-priority blocks may allow a certain range of tolerance. After verification, a verified decompressed data packet is generated, in which only data blocks that meet the time series and priority requirements are retained and reorganized into a complete set of logical order. Through this process, abnormal data blocks can be eliminated to ensure the consistency and integrity of the final data packet.

[0078] Step S530: Decryption processing is performed according to the verified decompressed data packets, the decrypted data blocks are reordered according to the priority order and the sequence information in the position-related metadata, and the data integrity is verified using a checksum to obtain a reconstructed data set.

[0079] During the decryption phase, each data block is decrypted one by one using the previously generated device key. The encryption characteristics of each data block are different, so the decryption needs to be processed block by block to ensure accuracy. The device key is dynamically generated and uniquely bound to the device, thereby ensuring the security of the decryption operation and preventing external attackers from using tampered keys or unauthorized devices to decrypt data blocks. Preferably, the decryption process uses an efficient and secure symmetric encryption algorithm (such as AES-256) to ensure the efficiency of decryption in real-time payment scenarios. The decrypted data blocks need to be reordered according to the priority order and the sequence information in the location-associated metadata. The core of this process is to ensure that the decrypted data can be restored to the logical order before transmission. The priority order determines that the key data blocks are in the front, ensuring that these core data can be quickly put into payment verification. The location-associated metadata (such as the starting offset or original sequence number of the data block) is used to accurately restore the position of the data block in the complete data structure. Through reordering, the scattered decrypted data blocks can be reorganized into a consistent and ordered reorganized data set. In order to ensure that the data has not been tampered with or lost during the decryption and sorting process, a checksum is used to verify the integrity of each data block. Preferably, the checksum is generated by a hash algorithm when the data block is generated, and is attached to the data block during transmission. The decrypted data block will recalculate its hash value and compare it with the original checksum. If the check succeeds, the data block is considered complete and credible; if the check fails, the data block is marked as abnormal and a compensation mechanism or error handling is triggered.

[0080] After decryption, reordering and integrity verification, all verified data blocks are integrated into a reconstructed data set. This data set is complete and ordered, including encrypted user facial feature data, payment request data and other related metadata, and restores the data structure before transmission. The reconstructed data set provides a reliable and structured input for the subsequent payment verification step.

[0081] Further, step S520 includes step S521 to step S524.

[0082] Step S521: Perform model building processing according to the decompressed data packets, extract the data block sequence based on the priority label and embed the transmission offset value to obtain a transmission sequence model sorted by priority;

[0083] Specifically, during the transmission process, each data block may experience a certain delay or offset due to network conditions or other external factors. The transmission offset value is used to record the transmission time or position changes of these data blocks. For example, a data block may be transmitted first due to its high priority, but its actual arrival time may deviate slightly from the expected time. The transmission offset value is calculated by decompressing the metadata in the data packet (such as timestamp or data packet offset index) and embedded in the transmission sequence model to accurately reflect the actual transmission order of the data block. The constructed transmission sequence model describes the transmission order and time characteristics of the data block in a serialized form. The structure of the transmission sequence model usually includes the following key information: data block identifier, priority label, transmission offset value and sequence order. The data block identifier is used to mark the unique number or type of the data block; the priority label is used to indicate the importance and transmission priority of the data block; the transmission offset value reflects the degree of deviation between the actual transmission time or order of the data block and the expected one; the sequence order sorts the data blocks according to the priority and transmission offset value to form a logical transmission order. The transmission sequence model not only improves the perception of the transmission order of data blocks, but also provides solid support for real-time verification in complex payment scenarios.

[0084] Step S522: Perform model building processing according to the transaction action data, build a mapping model based on the time series by parsing the operation sequence and timestamp information in the transaction action data, and adjust the model structure in combination with the time constraint parameters of the payment scenario to obtain a time series model;

[0085] The time series mapping model is centered on the operation sequence and timestamps, and organizes transaction action data according to the timeline. The model not only describes the relative time intervals of each operation, but also establishes the correlation between operations through mapping logic. By mapping the timestamps and sequence of operations, the model can dynamically reflect the logical process of payment behavior. The time constraint parameters of the payment scenario are an important basis for model optimization, including the maximum allowable delay or the total duration limit of the payment behavior. The time constraint parameters can dynamically adjust the structure of the time series model. For example, if the timestamp of an operation exceeds the preset time range, the model will mark the operation as abnormal and re-evaluate the integrity of the time series. Through this adjustment, the time series model can better meet the real-time requirements of the payment scenario. The time series model constructs a complete transaction action track through the analysis of timestamps and operation sequences, ensuring the sequential consistency of all key operations in the payment process, which helps prevent payment anomalies caused by incorrect sequence or missing operations.

[0086] Step S523, performing matching processing according to the transmission sequence model and the time series model, calculating the matching scores of the two models through a dynamic sequence alignment algorithm, and obtaining a sequential verification model;

[0087] It should be noted that the transmission sequence model records the priority and transmission offset of each data block, while the time series model is based on the timestamp and order of user operations. Therefore, it is necessary to use a dynamic sequence alignment algorithm to calculate the matching score of each data block, evaluate their deviation from the expected timestamp, and dynamically adjust the matching accuracy according to the requirements of the payment scenario. If the deviation of the data block exceeds a certain range, it is marked as an exception to ensure that every link in the payment process proceeds smoothly according to the time requirements. In this way, not only can the integrity and sequential consistency of the payment data in transmission be ensured, but also the real-time and tamper-proof capabilities of the payment can be improved, ensuring the smoothness and security of the user experience. The calculation formula for the matching score is:

[0088]

[0089] Among them, MatchScore represents the final matching score; i represents the sequence number of the data block in the transmission model; j represents the sequence number of the time point in the time series model; W time,i represents the time importance weight of the i-th data block in the payment scenario; T (j) represents the jth time point in the time series model; O i Indicates the offset value of the i-th data block in the transmission model.

[0090] Step S524: adjust and screen the order of the data blocks in the decompressed data packet according to the sequence verification model, verify the sequence consistency of each data block, remove abnormal data blocks and adjust the order to obtain a decompressed data packet that passes the verification.

[0091] It should be noted that in the face-scanning payment scenario, the data blocks may be out of order due to network delays, transmission errors or other factors, so it is necessary to ensure that the data can still be processed in the correct order after decompression. Specifically, first verify the matching of the transmission order of each data block with the time series according to the sequence verification model. If it is found that the order of some data blocks does not meet expectations, adjust the order of the data blocks according to the matching score and offset value to ensure that they are arranged in the correct logic. At the same time, the abnormal data blocks in the decompressed data packet will be screened, and those data blocks that cannot match the transmission order will be eliminated.

[0092] Further, step S600 includes step S610 to step S630.

[0093] Step S610, obtaining a face template library stored in the system, and performing face feature matching processing on the user face feature data in the reorganized data set and the face template library, and obtaining a user face matching result by calculating the similarity between the feature vector of the user face feature data and the user face template vector pre-stored in the template library;

[0094] In this step, the encrypted and decrypted facial feature data of the user is first extracted from the reorganized data set. These feature data exist in the form of high-dimensional vectors, representing the unique information of the user's face. Then, the pre-saved user face templates are obtained from the stored face template library. These templates are collected and generated by the face recognition system when the user registers. Next, the similarity between the current user's face feature vector and the template vector of the corresponding user in the template library is calculated, preferably using algorithms such as cosine similarity and Euclidean distance. If the similarity between the two exceeds the set threshold, it indicates that the recognition is successful, the user's identity is verified, and the payment request can continue to be processed; conversely, if the similarity is lower than the threshold, it is considered that the face match fails and the payment request is rejected.

[0095] Step S620: Perform consistency verification processing on the payment request data and payment scenario data in the reorganized data set, verify whether the payment request is initiated within a legal time and area by matching the transaction content, time and geographic information, and obtain a payment consistency verification result;

[0096] In this step, payment request data and related payment scenario data are first extracted from the reorganized data set, where the payment request data usually includes the transaction amount, merchant information, payment timestamp, etc., while the payment scenario data includes information such as the time and geographic location of the transaction. This information is matched, and the payment time is first verified against the legal time range preset by the system to ensure that the transaction is carried out within the permitted time. For example, the system may limit payments to only be made within a specific time period, and payment requests outside this time range will be considered abnormal. At the same time, the geographic location data in the payment request is compared with the actual geographic information in the payment scenario to verify whether the payment request is initiated within a legal geographic area. For example, if the payment request comes from a high-risk area or does not match the user's known location, the request is marked as a suspicious transaction. Through such consistency verification of time and geographic information, potential fraud can be effectively identified to ensure the security and compliance of the payment process.

[0097] Step S630: Perform a comprehensive judgment based on the user face matching result and the payment consistency verification result to generate a final payment verification result.

[0098] This step can verify the authenticity of the payment in multiple dimensions by combining the scenario data of identity authentication and payment request, thereby effectively protecting the interests of users and improving the credibility of the payment process.

[0099] Embodiment 5:

[0100] like Figure 2 As shown, this embodiment provides a face-scanning payment data transmission system for implementing the face-scanning payment data transmission method described in any one of Embodiments 1-4, and the system includes:

[0101] Acquisition module 901, used to acquire payment data and scenario data, the payment data includes user facial image data, payment request data and device identification data, and the scenario data includes payment timestamp data, payment location data and transaction action data;

[0102] Extraction module 902, used to perform facial feature extraction processing based on the user's facial image data, extract the user's facial features, and verify the time validity of the image data in combination with the payment timestamp data to obtain the user's facial feature data that passes the time sequence verification;

[0103] The encryption module 903 is used to perform feature association encryption processing according to the user's facial feature data, the payment request data and the device identification data, generate a unique device key through the device identification data to encrypt the user's facial feature data, and bind the encrypted data with the payment request data to obtain a device-bound encrypted data packet;

[0104] The marking module 904 dynamically analyzes the regional importance of the device-bound encrypted data packet based on the payment location data, marks the transmission priority of data blocks of different importance and adds location-related metadata to obtain data packets based on location blocks;

[0105] The decompression module 905 is used to perform decompression and action verification processing according to the data packets divided into blocks based on the location, decompress the data packets and verify the integrity of the data block transmission sequence according to the transaction action data, and use the device key to decrypt the data blocks to obtain a reorganized data set that passes the action verification;

[0106] The verification module 906 is used to perform payment verification and feedback processing according to the reorganized data set to obtain the final payment verification result.

[0107] Embodiment 3:

[0108] Corresponding to the above method embodiment, this embodiment also provides a face-scanning payment data transmission device. The face-scanning payment data transmission device described below and the face-scanning payment data transmission method described above can refer to each other.

[0109] Figure 3 FIG. 8 is a block diagram of a face-scanning payment data transmission device 800 according to an exemplary embodiment. Figure 3 As shown, the face-swiping payment data transmission device 800 may include: a processor 801 and a memory 802. The face-swiping payment data transmission device 800 may also include one or more of a multimedia component 803, an I / O interface 804, and a communication component 805.

[0110] In an exemplary embodiment, a face-scanning payment data transmission device 800 can be implemented by one or more application-specific integrated circuits (ASIC), digital signal processors (DSP), digital signal processing devices (DSPD), programmable logic devices (PLD), field programmable gate arrays (FPGA), controllers, microcontrollers, microprocessors or other electronic components to execute the above-mentioned face-scanning payment data transmission method.

[0111] In another exemplary embodiment, a computer-readable storage medium including program instructions is also provided, and when the program instructions are executed by a processor, the steps of the above-mentioned face-swiping payment data transmission method are implemented. For example, the computer-readable storage medium may be the above-mentioned memory 802 including program instructions, and the above-mentioned program instructions may be executed by a processor 801 of a face-swiping payment data transmission device 800 to complete the above-mentioned face-swiping payment data transmission method.

[0112] The above description is only a specific implementation mode of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can easily think of changes or substitutions within the technical scope disclosed by the present invention, which should be covered by the protection scope of the present invention.

Claims

1. A face-scanning payment data transmission method, characterized in that: include: Acquire payment data and scenario data, wherein the payment data includes user facial image data, payment request data and device identification data, and the scenario data includes payment timestamp data, payment location data and transaction action data; Performing facial feature extraction processing according to the user facial image data, extracting the user facial features and verifying the time validity of the image data in combination with the payment timestamp data, to obtain user facial feature data that passes the timing verification; Perform feature association encryption processing according to the user facial feature data, the payment request data and the device identification data, generate a unique device key through the device identification data to encrypt the user facial feature data, and bind the encrypted data to the payment request data to obtain a device-bound encrypted data packet; Dynamically analyzing the regional importance of the device-bound encrypted data packet based on the payment location data, marking the transmission priority of data blocks of different importance and attaching location-related metadata to obtain data packets based on location blocks; Decompressing and verifying the action according to the position-based block data packet, decompressing the data packet and verifying the integrity of the data block transmission sequence according to the transaction action data, decrypting the data block using the device key, and obtaining a reorganized data set that passes the action verification; Payment verification and feedback processing are performed according to the reorganized data set to obtain a final payment verification result.

2. The face recognition payment data transmission method according to claim 1, characterized in that: Performing facial feature extraction processing according to the user facial image data, extracting the user facial features, and verifying the time validity of the image data in combination with the payment timestamp data, to obtain the user facial feature data that passes the timing verification, including: Perform feature point detection processing on the user's facial image data, extract key facial regions based on a multi-scale pyramid feature extraction method, and obtain candidate facial feature region data; Perform feature encoding processing according to the candidate facial feature region data, generate a standardized facial feature vector through deep embedding representation, and obtain preliminary facial feature vector data; Performing time matching processing according to the preliminary facial feature vector data, filtering valid image frames through the payment timestamp data, and obtaining preliminary facial feature vector data with valid time matching; Noise verification is performed on the preliminary facial feature vector data that is effective in time matching, the cosine similarity between the feature vector and the standard distribution is calculated based on the distribution of the preliminary feature vector in the feature space, abnormal features that exceed the set threshold are eliminated, and user facial feature data that passes the timing verification is obtained.

3. The face recognition payment data transmission method according to claim 1, characterized in that: Performing feature-association encryption processing according to the user facial feature data, the payment request data and the device identification data, generating a unique device key through the device identification data to encrypt the user facial feature data, and binding the encrypted data with the payment request data to obtain a device-bound encrypted data packet, including: Performing key generation processing according to the device identification data, and obtaining a device key by performing dynamic key calculation using a preset random factor; Using the device key to encode and encrypt the user facial feature data, and obtaining encrypted user facial feature data by processing different regions of the feature data one by one; Binding the encrypted user facial feature data with the payment request data, and obtaining a binding data packet by constructing a JSON data structure and serializing it; Integrity verification is performed based on the intermediate data packet and device identification data. A verification value is generated and encapsulated by using a hash algorithm. The source authenticity and transmission integrity of the data packet are verified and encapsulated in combination with the uniqueness of the device identification to obtain a device-bound encrypted data packet.

4. The face recognition payment data transmission method according to claim 1, characterized in that: Dynamically analyzing the regional importance of the device-bound encrypted data packet based on the payment location data, marking the transmission priority of data blocks of different importance and attaching location-related metadata, to obtain a data packet based on location segmentation, including: Matching the payment location data with a preset geographic priority matrix, and obtaining regional weights by calculating the importance of different locations to the payment data; Based on the regional weight, the device-bound encrypted data packet is parsed in blocks, and the importance score of each data block is obtained by quantitatively calculating the importance of each data block; Performing transmission priority marking processing according to the importance score, sorting the data blocks from high to low according to the importance score and assigning priority tags to obtain a set of data blocks with transmission priority tags; Payment location metadata and priority tags are embedded in each data block based on the data block set, and encapsulated into a final data packet.

5. The face recognition payment data transmission method according to claim 4, characterized in that: According to the matching of the payment location data with the preset geographic priority matrix, the regional weights are obtained by calculating the importance of different locations to the payment data, including: The calculation formula is: Where (x, y) represents the latitude and longitude coordinates of the current payment location; W region (x, y) represents the regional weight of the current payment location (x, y); k represents the location point number in the geographic priority matrix; m represents the total number of all known priority locations in the geographic priority matrix; R k represents the basic priority score of the k-th known position; d((x,y),(x k ,y k )) indicates the current payment position (x, y) and the kth high priority position (x k ,y k ) geographical distance; σ represents the hyperparameter that controls the distance decay rate; H k represents the historical payment weight of the kth high priority position; l represents the index variable used for normalization calculation; p represents the total number of reference points involved in normalization; α represents the weight adjustment factor of the risk score; S(x,y) represents the risk score of the current payment position (x,y).

6. The face recognition payment data transmission method according to claim 1, characterized in that: Decompressing and verifying the action according to the position-based block data packet, decompressing the data packet and verifying the integrity of the data block transmission sequence according to the transaction action data, decrypting the data block using the device key, and obtaining a reorganized data set that passes the action verification, including: Decompressing the data packet based on the position-based block, performing decompression operation by extracting the position-related metadata and priority tag in the data packet, and recovering the compressed data block by block using a block-based decoding algorithm to obtain a decompressed data packet; Performing data block transmission sequence verification processing according to the decompressed data packet and the transaction action data, extracting time series information and comparing it with the priority tag and building a transmission sequence verification model, to obtain a verified decompressed data packet; Decryption processing is performed on the decompressed data packets that have passed the verification, the decrypted data blocks are reordered according to the priority order and the sequence information in the position-related metadata, and the data integrity is verified using a checksum to obtain a recombined data set.

7. The face recognition payment data transmission method according to claim 6, characterized in that: The data block transmission sequence verification process is performed according to the decompressed data packet and the transaction action data, and the decompressed data packet that passes the verification is obtained by extracting the time series information and comparing it with the priority label and building a transmission sequence verification model, including: Performing model building processing according to the decompressed data packets, extracting data block sequences based on priority tags and embedding transmission offset values ​​to obtain a transmission sequence model sorted by priority; Model building and processing are performed based on transaction action data. A mapping model based on time series is constructed by parsing the operation sequence and timestamp information in the transaction action data. The model structure is adjusted in combination with the time constraint parameters of the payment scenario to obtain a time series model. Performing matching processing according to the transmission sequence model and the time series model, calculating the matching scores of the two models through a dynamic sequence alignment algorithm, and obtaining a sequential verification model; The order of the data blocks in the decompressed data packet is adjusted and screened according to the sequence verification model, and the sequence consistency of each data block is verified, abnormal data blocks are removed and the order is adjusted to obtain a decompressed data packet that passes the verification.

8. The face recognition payment data transmission method according to claim 7, characterized in that: Matching processing is performed according to the transmission sequence model and the time series model, and matching scores of the two models are calculated by a dynamic sequence alignment algorithm to obtain a sequential verification model, including: The calculation formula is: Among them, MatchScore represents the final matching score; i represents the sequence number of the data block in the transmission model; j represents the sequence number of the time point in the time series model; W time,i represents the time importance weight of the i-th data block in the payment scenario; T (j) represents the jth time point in the time series model; O i Indicates the offset value of the i-th data block in the transmission model.

9. The face recognition payment data transmission method according to claim 1, characterized in that: Payment verification and feedback processing are performed according to the reorganized data set to obtain a final payment verification result, including: Acquire a face template library stored in the system, and perform face feature matching processing on the user face feature data in the reorganized data set and the face template library, and obtain a user face matching result by calculating the similarity between the feature vector of the user face feature data and the user face template vector pre-stored in the template library; Perform consistency verification processing on the payment request data and payment scenario data in the reorganized data set, verify whether the payment request is initiated within a legal time and area by matching the transaction content, time and geographic information, and obtain a payment consistency verification result; A comprehensive judgment is made based on the user face matching result and the payment consistency verification result to generate a final payment verification result.

10. A face recognition payment data transmission system, characterized in that: include: An acquisition module, used to acquire payment data and scenario data, wherein the payment data includes user facial image data, payment request data and device identification data, and the scenario data includes payment timestamp data, payment location data and transaction action data; An extraction module, used to perform facial feature extraction processing based on the user facial image data, extract the user facial features, and verify the time validity of the image data in combination with the payment timestamp data to obtain the user facial feature data that passes the timing verification; An encryption module is used to perform feature association encryption processing according to the user facial feature data, the payment request data and the device identification data, generate a unique device key through the device identification data to encrypt the user facial feature data, and bind the encrypted data with the payment request data to obtain a device-bound encrypted data packet; A marking module dynamically analyzes the regional importance of the device-bound encrypted data packet based on the payment location data, marks the transmission priority of data blocks of different importance and attaches location-related metadata to obtain a data packet based on location blocks; A decompression module, used to perform decompression and action verification processing according to the data packet based on the position block, decompress the data packet and verify the integrity of the data block transmission sequence according to the transaction action data, and use the device key to decrypt the data block to obtain a reorganized data set that passes the action verification; The verification module is used to perform payment verification and feedback processing according to the reorganized data set to obtain a final payment verification result.

Images