Blockchain-based cross-border e-commerce order intelligent processing method and system, electronic device and storage medium

Abstract

This application provides a blockchain-based intelligent processing method, system, electronic device, and storage medium for cross-border e-commerce orders, relating to the field of blockchain technology. The method includes: receiving order information, material spectral data, and genuine product feature data to construct a multi-dimensional knowledge graph; triggering hyperspectral scanning based on product identification to generate material composition data, and performing spectral matching with the knowledge graph to obtain a matching degree; simultaneously extracting key region feature sets from the hyperspectral image using the U-Net image segmentation algorithm, and comparing them with an anti-counterfeiting feature library to obtain spatial similarity; when both indicators meet customs thresholds, activating the target smart contract based on an emergency identifier and generating an execution record; finally, encapsulating the verification data into a verification package, generating a hash value, and writing it to the blockchain, establishing a bidirectional index with the order through the smart contract to achieve state synchronization. This application improves the automation level and data reliability of cross-border e-commerce product customs clearance verification.

Description

Technical Field

[0001] This application relates to the field of blockchain technology, and in particular to a blockchain-based intelligent processing method, system, electronic device, and storage medium for cross-border e-commerce orders. Background Technology

[0002] With the rapid development of cross-border e-commerce, customs faces the dual challenges of efficiency and accuracy in commodity verification. Since the verification of commodity authenticity requires the combination of material composition testing and anti-counterfeiting feature comparison, and also needs to meet the timeliness requirements in different customs clearance processes, higher standards are set for the multi-dimensional data fusion capabilities and intelligent decision-making speed of automated verification technology, thereby driving the continuous evolution of related technologies towards greater efficiency and reliability.

[0003] Currently, automated verification systems employ a combination of single-spectral analysis and image recognition. These systems first compare the product material against a pre-set spectral database, and then use traditional image processing algorithms to extract anti-counterfeiting features. The material and anti-counterfeiting feature detection results are then calculated using a simple weighted average to output the verification conclusion. In addition, for data management, the system uses a centralized database to store each verification record and connects with the existing customs clearance system through a standard API interface.

[0004] However, the reliability of the final verification conclusion is significantly limited because the results of spectral analysis and image recognition are independent of each other. In addition, the centralized data storage method itself has the potential risk of being tampered with, and the system's response speed is easily affected by fluctuations in customs clearance business volume. Therefore, the system is difficult to adapt to the real-time verification requirements in high-concurrency scenarios. Summary of the Invention

[0005] This application provides a blockchain-based intelligent processing method, system, electronic device, and storage medium for cross-border e-commerce orders, in order to solve the problems of low automation level and poor data reliability in the customs clearance verification of cross-border e-commerce goods in the prior art.

[0006] To address the aforementioned technical problems, firstly, this application provides a blockchain-based intelligent processing method for cross-border e-commerce orders, comprising:

[0007] The system receives cross-border e-commerce order information, product standard data, and hyperspectral images. The cross-border e-commerce order information includes a unique product identifier and an emergency identifier.

[0008] Based on the product standard data, a multi-dimensional knowledge graph is constructed. The product surface is scanned based on the product's unique identifier to obtain material composition data. The material composition data is then spectrally matched with the multi-dimensional knowledge graph to obtain the spectral matching degree.

[0009] Using the U-Net image segmentation algorithm, a key region feature set is extracted from the hyperspectral image. The key region feature set is then compared with a preset anti-counterfeiting feature library to obtain the spatial similarity.

[0010] When the spectral matching degree and spatial similarity degree simultaneously meet the corresponding preset verification thresholds, the corresponding target smart contract is automatically activated according to the emergency identifier, and a contract execution record is generated.

[0011] The material composition data, the key area feature set, and the contract execution record are encapsulated into a complete data packet, and a unique hash value of the complete data packet is generated and written into the blockchain.

[0012] A two-way index relationship is established between the unique hash value and the cross-border e-commerce order information through a blockchain smart contract, so as to realize the real-time synchronous update of order status and verification results.

[0013] Optionally, the step of using the U-Net image segmentation algorithm to extract a key region feature set from the hyperspectral image, and comparing the key region feature set with a preset anti-counterfeiting feature library to obtain spatial similarity, includes:

[0014] Using the U-Net image segmentation algorithm, texture feature vectors and laser point coordinate sets are identified from the hyperspectral image. The texture feature vectors and laser point coordinate sets are used to construct key region feature sets.

[0015] Retrieve the baseline texture feature vector and baseline laser point coordinate set of the same model of product from the anti-counterfeiting feature library;

[0016] The matching degree between the texture feature vector and the reference texture feature vector is calculated to generate the morphological matching degree.

[0017] The offset between the laser point coordinate set and the reference laser point coordinate set is statistically analyzed to generate a three-dimensional spatial offset.

[0018] Spatial similarity is generated by combining the morphological matching degree and the three-dimensional spatial offset.

[0019] Optionally, the step of identifying texture feature vectors and laser point coordinate sets from the hyperspectral image using the U-Net image segmentation algorithm includes:

[0020] Based on the hyperspectral image, a mask image containing pixel-level classification results is determined using the U-Net image segmentation algorithm. The mask image includes an anti-counterfeiting label area and a laser marking area.

[0021] Extract the boundary coordinates of all consecutive pixels from the anti-counterfeiting label area, and use the boundary coordinates as the geometric boundary range of the anti-counterfeiting label area.

[0022] Extract the coordinates of the smallest bounding rectangle vertex of all discrete pixel clusters from the laser marking area, and use the coordinates of the smallest bounding rectangle vertex as the positioning range of the laser marking area;

[0023] Within the geometric boundary, a texture feature vector is generated. Within the positioning range, the three-dimensional spatial coordinates of each laser marker point are determined by a multi-view triangulation method. The three-dimensional spatial coordinates of all laser marker points are collected to generate a laser point coordinate set.

[0024] Optionally, generating a texture feature vector within the geometric boundary includes:

[0025] Within the geometric boundary, multiple local image blocks are generated according to a preset size division rule;

[0026] Using an edge detection algorithm, the ridges of all grooves in each local image block are identified, and a straight line is fitted to each ridge. The angle between the fitted ridge and the horizontal baseline is calculated.

[0027] The frequency distribution of each angle value in each local image patch is statistically analyzed to form angle distribution data;

[0028] Measure the distance between adjacent ridges in the same local image patch, calculate the arithmetic mean of all distance measurements in the same local image patch, and form the average value data;

[0029] By summarizing the angular distribution data and average value data of all local image patches, a texture feature vector is generated.

[0030] Optionally, based on the material spectral data and the genuine product feature data, a multi-dimensional knowledge graph is constructed, and the product surface scan is triggered based on the product's unique identifier to obtain material composition data, including:

[0031] Based on the material type, the spectral data of the material are grouped to form a subset of spectral data corresponding to each material type;

[0032] Separate the texture feature dataset and the marker dataset from the genuine product feature data;

[0033] Establish the association between the spectral data subset, the texture feature dataset, and the labeled dataset, and construct a multi-dimensional knowledge graph based on the association;

[0034] Based on the unique identifier of the product, a command is sent to the hyperspectral scanning device to obtain the reflectance spectrum sequence of the product surface;

[0035] Material composition data is generated based on the wavelength intensity distribution in the reflection spectrum sequence.

[0036] Optionally, the step of performing spectral matching between the material composition data and the multi-dimensional knowledge graph to obtain the spectral matching degree includes:

[0037] Obtain the peak wavelength and intensity ratio of the material composition data, and the standard peak wavelength and standard intensity ratio of the multi-dimensional knowledge graph;

[0038] The peak wavelength is compared point by point with the standard peak wavelength, and the percentage of wavelength overlap points is calculated based on the comparison results.

[0039] For all wavelength coincidence points, calculate the mean absolute deviation between the intensity ratio and the standard intensity ratio;

[0040] The spectral matching degree is generated by weighted calculation based on the proportion of wavelength overlap points and the mean absolute deviation.

[0041] Optionally, when the spectral matching degree and spatial similarity simultaneously meet the corresponding preset verification thresholds, the corresponding target smart contract is automatically activated based on the emergency identifier, and a contract execution record is generated, including:

[0042] When the spectral matching degree is greater than the spectral threshold in the verification threshold and the spatial similarity is greater than the spatial threshold in the verification threshold, the verification is marked as passed, and the timestamp of the verification is recorded.

[0043] Based on the emergency identifier, the corresponding target smart contract template is matched from the preset contract triggering rule base;

[0044] Write the unique identifier of the product and the timestamp into the target smart contract template to generate a target smart contract containing executable tokens;

[0045] Based on the executable marker, the execution operation of the clearance item in the target smart contract is triggered, and a contract execution record is generated.

[0046] Secondly, this application provides a blockchain-based intelligent order processing system for cross-border e-commerce, comprising:

[0047] The receiving module is used to receive cross-border e-commerce order information, product standard data, and hyperspectral images. The cross-border e-commerce order information includes a unique product identifier and an emergency identifier.

[0048] The construction module is used to construct a multi-dimensional knowledge graph based on the product standard data, trigger a product surface scan based on the product unique identifier to obtain material composition data, and perform spectral matching between the material composition data and the multi-dimensional knowledge graph to obtain the spectral matching degree.

[0049] The extraction module is used to extract key region feature sets from the hyperspectral image using the U-Net image segmentation algorithm, and compare the key region feature sets with the preset anti-counterfeiting feature library to obtain the spatial similarity.

[0050] The activation module is used to automatically activate the corresponding target smart contract and generate a contract execution record based on the emergency identifier when the spectral matching degree and spatial similarity simultaneously meet the corresponding preset verification thresholds.

[0051] The encapsulation module is used to encapsulate the material composition data, the key area feature set, and the contract execution record into a complete data packet, generate a unique hash value for the complete data packet, and write the unique hash value into the blockchain;

[0052] A module is established to create a bidirectional index relationship between the unique hash value and the cross-border e-commerce order information through a blockchain smart contract, so as to realize real-time synchronous updates of order status and verification results.

[0053] Thirdly, this application provides an electronic device, comprising:

[0054] Memory, used to store computer programs;

[0055] A processor is configured to execute the computer program to implement the steps of the blockchain-based intelligent processing method for cross-border e-commerce orders as described in the first aspect above.

[0056] Fourthly, this application provides a computer-readable storage medium storing a computer program that, when executed by a processor, can implement the steps of the blockchain-based intelligent processing method for cross-border e-commerce orders as described in the first aspect above.

[0057] This application provides a blockchain-based intelligent processing method for cross-border e-commerce orders, which has the following advantages: First, by receiving cross-border e-commerce order information containing unique product identifiers and emergency identification, product standard data, and hyperspectral images, the method achieves digital collection and fusion of order and inspection information; then, based on the product standard data, a multi-dimensional knowledge graph is constructed, and spectral matching is performed in conjunction with material composition data obtained from hyperspectral scanning to achieve standardized characterization and objective quantitative comparison of material composition, thereby improving the accuracy of product composition verification; simultaneously, the U-Net image segmentation algorithm is used to extract key region features from the hyperspectral image, and spatial similarity is compared with the anti-counterfeiting feature library to achieve precise positioning and multi-dimensional quantitative evaluation of the micro-features of anti-counterfeiting marks, further enhancing the reliability of anti-counterfeiting verification;

[0058] Furthermore, when both spectral matching and spatial similarity meet preset thresholds, the corresponding target smart contract can be automatically activated based on the order's urgency identifier, and a contract execution record can be generated. This enables intelligent linkage between verification conclusions and customs clearance processes, significantly improving order processing efficiency. In addition, this method encapsulates material composition data, key area feature sets, and contract execution records into a complete data package, generates a unique hash value, and writes it to the blockchain. A two-way index relationship between the hash value and order information is established through smart contracts, ensuring traceability and immutability throughout the verification process. Ultimately, this achieves real-time synchronization and updating of order status and verification results.

[0059] Furthermore, the texture feature vector and laser point coordinate set are accurately extracted from the hyperspectral image using the U-Net image segmentation algorithm, realizing the quantitative representation of the microstructure of the anti-counterfeiting mark and providing calculable feature basis for anti-counterfeiting verification. Then, by comparing with the benchmark texture feature vector and benchmark laser point coordinate set of the same model of product in the anti-counterfeiting feature library, the feature data is standardized and compared to ensure that the verification process has a unified objective reference.

[0060] Furthermore, the morphological matching degree and three-dimensional spatial offset are calculated separately to achieve independent quantitative evaluation of anti-counterfeiting marks in two dimensions: morphological features and spatial layout, making the verification judgment more comprehensive and refined. Finally, a unified spatial similarity index is generated by combining the morphological matching degree and three-dimensional spatial offset, forming a multi-dimensional integrated anti-counterfeiting verification conclusion, thereby improving the accuracy of anti-counterfeiting identification and the system's anti-counterfeiting capability.

[0061] These or other aspects of this application will become more apparent in the following description of the embodiments. Attached Figure Description

[0062] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0063] Figure 1 A flowchart illustrating a blockchain-based intelligent order processing method for cross-border e-commerce, provided as an embodiment of this application.

[0064] Figure 2 A schematic diagram illustrating a specific implementation of a blockchain-based intelligent processing method for cross-border e-commerce orders, provided in this application embodiment;

[0065] Figure 3 This is a schematic diagram illustrating another specific implementation of a blockchain-based intelligent processing method for cross-border e-commerce orders, provided in this application.

[0066] Figure 4 This is a schematic diagram of the structure of a blockchain-based intelligent processing system for cross-border e-commerce orders, provided as an embodiment of this application. Detailed Implementation

[0067] Current product verification technologies in the cross-border e-commerce sector still have several shortcomings: Although existing automated verification systems combine spectral analysis and image recognition, the lack of deep correlation between the two types of detection data leads to insufficient reliability in the final verification conclusions; at the system architecture level, the current centralized storage model not only poses potential risks to data security but also struggles to meet the higher requirements for real-time processing capabilities in high-concurrency business scenarios; more importantly, traditional solutions cannot intelligently link verification results with subsequent customs clearance processes, thus failing to prioritize urgent orders for verification and limiting further improvements in overall customs clearance efficiency.

[0068] To address the aforementioned issues, this application proposes a blockchain-based intelligent processing method for cross-border e-commerce orders. Firstly, it constructs a multi-dimensional knowledge graph integrating the spectral density of product materials and anti-counterfeiting features, thereby achieving deep correlation and collaborative analysis of product feature data across different dimensions. At the technical implementation level, this method innovatively combines hyperspectral material detection technology with U-Net (Convolutional Networks for Biomedical Image Segmentation) image segmentation technology to simultaneously acquire the product's component characteristics and microscopic anti-counterfeiting features, and ensures the accuracy of the final verification result through a dual verification mechanism. Furthermore, the blockchain-based smart contract system not only provides an immutable storage method for verification data but also automatically triggers a differentiated customs clearance process based on the verification results. This method fundamentally solves the three core pain points of existing technologies: data silos, security risks, and process disconnect. While improving verification accuracy, it ultimately achieves automation and intelligent upgrading of the entire order processing process.

[0069] To enable those skilled in the art to better understand the present application, the present application will be further described in detail below with reference to the accompanying drawings and specific embodiments. Obviously, the described embodiments are merely some embodiments of the present application, and not all embodiments. Based on the embodiments in this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0070] The core of this application is to provide a blockchain-based intelligent processing method for cross-border e-commerce orders, and a flowchart of one specific implementation is shown below. Figure 1 As shown, the method includes:

[0071] Step 101: Receive cross-border e-commerce order information, product standard data, and hyperspectral images. The cross-border e-commerce order information includes a unique product identifier and an emergency identifier.

[0072] In step 101, the cross-border e-commerce order information represents a data set containing a unique product identifier and an emergency identifier, used to identify products and determine the processing order; the unique product identifier is the product's identification number, and the emergency identifier is used to indicate whether the order needs to be expedited.

[0073] The product standard data represents a standard database containing material spectral characteristics and genuine product anti-counterfeiting features. Material spectral characteristics are used to record the reflection intensity data of different materials at different wavelengths, while genuine product anti-counterfeiting features can be the surface texture and laser marking features of genuine products.

[0074] Hyperspectral images are image data containing a large amount of continuous narrow-band information. They can simultaneously record the spatial and spectral information of an object's surface, and are used to analyze material composition and identify microscopic features. The "high" in hyperspectral image refers to the fact that the image contains a large amount of continuous and narrow spectral band data. Compared to the three bands of a regular RGB image, a hyperspectral image can contain hundreds of bands. For example, a hyperspectral image may contain data from 200 consecutive bands in the wavelength range of 400-1000nm, with each band spaced only 2-3nm apart, which can accurately reflect the spectral characteristics of a substance.

[0075] For example, a cross-border e-commerce platform declares a batch of imported goods to customs. Customs receives the order data, which includes product number A123456 and an urgent mark marked "expedited". At the same time, customs retrieves pre-stored standard data of the brand's products, including the standardized spectral characteristics of its leather material and the corresponding genuine anti-counterfeiting features. Subsequently, customs scans the actual goods using hyperspectral imaging equipment, obtaining hyperspectral image data containing 200 continuous bands in the wavelength range of 400 to 1000 nanometers. In this process, product number A123456 will serve as the unique identifier for all subsequent processing steps, while the "expedited" mark will determine the specific initiation method of the final customs clearance process.

[0076] Step 102: Based on the material spectral data and the genuine product feature data, construct a multi-dimensional knowledge graph, trigger a product surface scan based on the product's unique identifier to obtain material composition data, and perform spectral matching between the material composition data and the multi-dimensional knowledge graph to obtain the spectral matching degree.

[0077] In step 102, the multi-dimensional knowledge graph represents a structured knowledge base formed by modeling the association between material spectral features and anti-counterfeiting features. Nodes represent different material types, and edges represent the association between materials and corresponding anti-counterfeiting features.

[0078] Material composition data represents the material characteristic spectrum generated by analyzing the surface reflection spectrum of the product, recording the reflection intensity distribution at a specific wavelength; spectral matching degree represents an index that quantifies the similarity between the actual product material and the standard material, with a higher value indicating a higher matching degree.

[0079] In the embodiments of this application, the specific implementation process of step 102 can be referred to the specific descriptions of steps 501 to 505 and steps 601 to 604 below, which will not be repeated here.

[0080] Step 103: Using the U-Net image segmentation algorithm, extract key region feature sets from the hyperspectral image, and compare the key region feature sets with the preset anti-counterfeiting feature library to obtain the spatial similarity.

[0081] In step 103, the key area feature set refers to the data set containing the texture features of the anti-counterfeiting area on the surface of the product and the coordinates of the laser marker points. The anti-counterfeiting feature library is a pre-set standardized database, which is derived from the structured integration of the genuine product feature data. It is formed by extracting and classifying the anti-counterfeiting elements in the genuine product feature data to form an anti-counterfeiting feature library specifically for comparison. Spatial similarity represents an index that quantifies the degree of similarity between the spatial distribution of the actual product anti-counterfeiting features and the standard features.

[0082] In the embodiments of this application, the specific description of step 103 can be referred to steps 201 to 205 below, and will not be repeated here. Furthermore, the embodiments of this application do not impose specific limitations on the specific structure and implementation process of the U-Net image segmentation algorithm, and can be set accordingly according to the actual situation.

[0083] Step 104: When the spectral matching degree and spatial similarity degree simultaneously meet the corresponding preset verification thresholds, the corresponding target smart contract is automatically activated according to the emergency identifier, and a contract execution record is generated.

[0084] In step 104, the verification threshold is the minimum qualified standard value of spectral matching degree and spatial similarity preset by the customs. When the commodity test result reaches both of these standard values ​​at the same time, it is determined that the verification is passed. The spectral threshold is used to evaluate the conformity of the material composition, and the spatial threshold is used to evaluate the conformity of the anti-counterfeiting features.

[0085] The target smart contract represents a digital customs clearance agreement that is automatically executed based on the verification results. The target smart contract may include terms such as tariff calculation and logistics arrangements. The contract execution record is a digital operation log generated after the target smart contract is triggered to execute. The contract execution record contains key information such as commodity number, verification pass time, type of customs clearance operation executed, and estimated completion time. It is used to trace the execution status of the customs clearance process and can record spectral matching degree and spatial similarity.

[0086] In the embodiments of this application, the specific description of step 104 can be referred to steps 701 to 704 below, and will not be repeated here.

[0087] Step 105: Encapsulate the material composition data, the key area feature set, and the contract execution record into a complete data packet, generate a unique hash value for the complete data packet, and write the unique hash value into the blockchain.

[0088] In step 105, the complete data package refers to a complete verification dataset containing material composition data, anti-counterfeiting feature data, and contract execution records, and the verification hash value refers to the unique fingerprint of the verification data generated by the hash algorithm.

[0089] In this embodiment, firstly, the material composition data, anti-counterfeiting feature set, and contract execution record are packaged in a specific format to form a complete verification data package; then, the verification data package is hashed to generate its unique digital fingerprint; finally, the digital fingerprint is written into the blockchain network to complete the entire data storage process.

[0090] For example, first, the material data, anti-counterfeiting feature data, and contract execution record of product number A123456 are packaged together. Then, the SHA-256 algorithm is used to calculate the hash value "a1b2c3...". Finally, the hash value is written into block number 123456 of the blockchain network.

[0091] Step 106: Establish a two-way index relationship between the unique hash value and the cross-border e-commerce order information through a blockchain smart contract to achieve real-time synchronization and updating of order status and verification results.

[0092] In step 106, the blockchain smart contract is a basic functional component of the blockchain system, responsible for performing underlying technical operations such as data uploading and index creation; the bidirectional index relationship represents the bidirectional pointer association between order data and verification data, supporting bidirectional queries.

[0093] In this embodiment of the application, a two-way link is established between order data and verification hash value through a blockchain smart contract, so that the corresponding verification result can be directly obtained when querying an order, and the corresponding order information can also be traced through the verification hash value, ultimately realizing the association and real-time synchronization of status between order and verification data.

[0094] For example, a two-way link can be established between order number A123456 and the hash value "a1b2c3...", allowing complete information to be queried from either end later.

[0095] This method automates and automates the cross-border e-commerce commodity verification process through multi-dimensional feature fusion analysis and the application of blockchain technology. Based on this, a full-process digital solution from order receipt to customs clearance execution is established. This solution effectively ensures the accuracy of verification, improves the overall processing efficiency, and ensures the security and traceability of data throughout the process, thereby meeting customs clearance needs of different urgency levels.

[0096] To improve the accuracy and automation of anti-counterfeiting verification for cross-border e-commerce goods, in some embodiments, step 103 involves: using the U-Net image segmentation algorithm to extract key region feature sets from the hyperspectral image, and comparing the key region feature sets with a preset anti-counterfeiting feature library to obtain spatial similarity, such as... Figure 2 As shown, it includes:

[0097] Step 201: Using the U-Net image segmentation algorithm, identify texture feature vectors and laser point coordinate sets from the hyperspectral image. The texture feature vectors and laser point coordinate sets are used to construct key region feature sets.

[0098] In step 201, the texture feature vector is a data set generated by analyzing the orientation angle and spacing distribution of the micro-grooves in the anti-counterfeiting label area of ​​the product, and is used to characterize the surface texture features; the laser point coordinate set is a data set generated by measuring the three-dimensional position coordinates of the laser marking points on the product surface, and is used to characterize the spatial distribution features of the laser markings.

[0099] In this embodiment, the hyperspectral image of the product is first processed using the U-Net image segmentation algorithm to identify the anti-counterfeiting label area and the laser marking area. Then, the direction and spacing of the micro-grooves are detected in the identified anti-counterfeiting label area, and the corresponding texture feature vector is generated. At the same time, the three-dimensional coordinates of each laser point are located in the laser marking area to form a laser point coordinate set. Finally, the texture feature vector and the laser point coordinate set will be used together as the basic data for subsequent comparison with the features of genuine products.

[0100] Step 202: Retrieve the baseline texture feature vector and baseline laser point coordinate set of the same model of product from the anti-counterfeiting feature library.

[0101] In step 202, the reference texture feature vector is the standard texture feature data of the same model of product stored in the anti-counterfeiting feature library, and the reference laser point coordinate set is the standard laser mark coordinate data of the same model of product stored in the anti-counterfeiting feature library.

[0102] In this embodiment of the application, standard texture feature data and standard laser coordinate data corresponding to the model information of the current product are retrieved from the pre-stored genuine product feature library. These pre-prepared standard data will serve as the reference benchmark for subsequent feature comparison.

[0103] Step 203: Calculate the matching degree between the texture feature vector and the reference texture feature vector to generate the morphological matching degree.

[0104] In step 203, the morphological matching degree is an indicator that quantifies the similarity between actual texture features and standard texture features. It is obtained by comparing the distribution of groove orientation angles and the average spacing value.

[0105] In this embodiment of the application, the actual detected texture feature vector is compared with the reference texture feature vector. Specifically, this is achieved by analyzing the similarity between the two in the distribution of the groove direction angle, and at the same time, the similarity of their average spacing values ​​is compared. Finally, a quantitative morphological matching score is calculated.

[0106] Step 204: Perform offset statistics between the laser point coordinate set and the reference laser point coordinate set to generate a three-dimensional spatial offset.

[0107] In step 204, the three-dimensional spatial offset is an indicator that quantifies the difference between the actual laser point coordinates and the standard coordinate spatial position, and is obtained by calculating the spatial distance between each corresponding point.

[0108] In this embodiment of the application, the actual detected set of laser point coordinates is first compared with the reference set of coordinates. Then, the position deviation of each corresponding laser point in the three-dimensional space is calculated, and its average offset is statistically analyzed. Finally, the corresponding three-dimensional space offset is generated.

[0109] Step 205: Combine the morphological matching degree and the three-dimensional spatial offset to generate spatial similarity.

[0110] In this embodiment, the morphological matching degree and the three-dimensional spatial offset are weighted according to preset weights to obtain a comprehensive spatial similarity score. This score can effectively reflect the overall degree of conformity between the anti-counterfeiting features of the product and the genuine product standard.

[0111] For example, the formula for calculating spatial similarity can be: ,in For spatial similarity, Where D represents the morphological matching degree, and D represents the 3D spatial offset. To allow the maximum coordinate offset, and As for the weight coefficients of the corresponding items, the embodiments of this application do not specifically limit the value of each weight coefficient, and the embodiments of this application do not specify the specific expression of the formula for calculating spatial similarity.

[0112] Here is a specific example:

[0113] After verifying the leather handbag with product number A123456, the acquired hyperspectral image was further processed. First, the precise outline of the anti-counterfeiting label area and the specific distribution of 5 laser markers were identified using the U-Net image segmentation algorithm. Subsequently, 8 main grooves were detected within the identified anti-counterfeiting label area. Analysis revealed that their directional angles appeared 3 times at 15 degrees, 3 times at 18 degrees, and 2 times at 22 degrees, with corresponding angular distribution frequencies of 37.5%, 37.5%, and 25%, respectively. Simultaneously, the distances between adjacent grooves were measured to be 0.12mm, 0.15mm, 0.13mm, 0.14mm, 0.16mm, 0.15mm, and 0.17mm, with an average distance of 0.146mm. These data collectively constitute the texture feature vector.

[0114] For the laser-marked points, the three-dimensional coordinates of five points were obtained through multi-view measurement, namely [1.2, 3.4, 0.8], [1.5, 3.6, 0.7], [1.3, 3.3, 0.9], [1.4, 3.5, 0.8], and [1.6, 3.7, 0.7], thus forming a laser point coordinate set. Next, the reference data of the same model of handbag was retrieved from the anti-counterfeiting feature library. Its texture features showed that the standard angle distribution was 40% at 15 degrees, 30% at 18 degrees, and 30% at other angles, with an average spacing of 0.15 mm. The standard coordinates of the five laser points were [1.0, 3.0, 1.0], [1.5, 3.5, 1.0], [1.0, 3.5, 1.0], [1.5, 3.0, 1.0], and [1.25, 3.25, 1.0].

[0115] When comparing the actual texture features with the benchmark data, the texture morphology matching degree is 1 - (|37.5% - 40%| + |37.5% - 30%| + |25% - 30%|) / 3 = 0.89, and the spacing matching degree is 1 - |0.146 - 0.15| / 0.15 = 0.973; the spatial offset of each point is calculated by comparing the laser point coordinates, where the offset of point 1 is sqrt[(1.2 - 1.0)² + (3.4 - 3 ... [0.8-1.0]²+(0.8-1.0)²=0.49mm. Similarly, the offsets of other points are calculated as 0.36mm, 0.61mm, 0.64mm, and 0.44mm, respectively, with an average offset of 0.51mm. The three-dimensional spatial offset = 1-0.51 / 1.0=0.49. Finally, the spatial similarity is calculated with a weight of 0.6 for morphological matching and a weight of 0.4 for three-dimensional spatial offset, resulting in 0.89×0.6+0.49×0.4=0.73.

[0116] In this embodiment, the method achieves precision and automation of anti-counterfeiting verification through multi-dimensional feature extraction and comparison, thereby providing effective technical support for customs clearance of cross-border e-commerce goods.

[0117] To improve the accuracy of anti-counterfeiting feature extraction, in some embodiments, step 201: identifying texture feature vectors and laser point coordinate sets from the hyperspectral image using the U-Net image segmentation algorithm, including:

[0118] Step 301: Based on the hyperspectral image, use the U-Net image segmentation algorithm to determine a mask image containing pixel-level classification results. The mask image includes an anti-counterfeiting label area and a laser marking area.

[0119] In step 301, the mask image is a marked image generated by an image segmentation algorithm, in which each pixel is classified as an anti-counterfeiting label area or a laser marking area to distinguish different feature areas on the surface of the product.

[0120] Anti-counterfeiting label areas refer to areas on the surface of goods that contain specific textures, patterns, or markings. These areas have unique microscopic features and are used to verify authenticity through optical detection. In this application, it specifically refers to the range of label locations with anti-counterfeiting functions identified by hyperspectral images.

[0121] A laser-marked area refers to a concentrated distribution area of ​​a set of discrete markers made on the surface of a commodity using laser technology. These markers have specific three-dimensional spatial arrangement characteristics. In this application, it specifically refers to a cluster area composed of multiple laser markers identified through image segmentation.

[0122] In this embodiment, the acquired hyperspectral image is input into the U-Net network model. Since this model is specially trained, it can accurately identify the pixel positions of anti-counterfeiting labels and laser markings, thereby outputting a classification result image for each pixel. In this image, the anti-counterfeiting label area and the laser marking area are marked with different values. It should be noted that this embodiment does not specifically limit the specific structure of the U-Net network model; it can be configured according to actual conditions.

[0123] Step 302: Extract the boundary coordinates of all consecutive pixels from the anti-counterfeiting label area, and use the boundary coordinates as the geometric boundary range of the anti-counterfeiting label area.

[0124] In step 302, the boundary coordinates are the set of position data of the edge pixels of the anti-counterfeiting label area. Connecting these points in sequence can form a complete area outline. The geometric boundary range is the minimum enclosing range of the anti-counterfeiting label area determined by these coordinate points.

[0125] In this embodiment, all consecutive pixels marked as anti-counterfeiting labels are extracted from the mask image, and the position coordinates of these edge pixels are traced sequentially in a clockwise direction to finally form a closed polygonal outline, which serves as the precise boundary range of the anti-counterfeiting label.

[0126] Step 303: Extract the coordinates of the smallest bounding rectangle vertex of all discrete pixel clusters from the laser marking area, and use the coordinates of the smallest bounding rectangle vertex as the positioning range of the laser marking area.

[0127] In step 303, the vertex coordinates of the smallest bounding rectangle are the coordinates of the four corner points of the smallest rectangle that can completely enclose the cluster of laser marker points, and the positioning range refers to the spatial range of the laser marker area determined by these rectangular coordinates.

[0128] In this embodiment, all pixel clusters marked as laser markers are located in the mask image, and the smallest bounding rectangle that can completely enclose each cluster is calculated. At the same time, the coordinates of the four vertices of the rectangle are recorded. These rectangular areas will serve as the operating range for subsequent precise three-dimensional positioning of the laser points.

[0129] Step 303: Within the geometric boundary range, generate a texture feature vector. Within the positioning range, determine the three-dimensional spatial coordinates of each laser marker point using a multi-view triangulation method. Collect the three-dimensional spatial coordinates of all laser marker points to generate a laser point coordinate set.

[0130] In this embodiment, within the defined boundary of the anti-counterfeiting label, the orientation angle of the surface grooves and the distance between adjacent grooves are detected, and these statistically obtained feature data are integrated to generate a texture feature vector; at the same time, within the positioning range of the laser mark, the precise three-dimensional coordinates of each laser point are measured by multi-angle imaging technology, and finally the coordinates of all laser points are collected together to form a laser point coordinate set.

[0131] In this embodiment of the application, the method achieves automated identification and measurement of anti-counterfeiting labels and laser markings through precise region segmentation and feature extraction, thereby providing a reliable feature data foundation for subsequent anti-counterfeiting verification.

[0132] To extract the texture features of anti-counterfeiting labels more accurately, in some embodiments, step 303: generating a texture feature vector within the geometric boundary range includes:

[0133] Step 401: Within the geometric boundary range, generate multiple local image blocks according to a preset size division rule.

[0134] In step 401, the local image block is a number of small image blocks that divide the anti-counterfeiting label area into fixed sizes, which are used for fine-grained feature detection and analysis.

[0135] In this embodiment, the entire area of ​​the anti-counterfeiting label is uniformly divided into multiple rectangular blocks of identical size according to a preset division rule. Each rectangular block serves as an independent analysis unit to facilitate subsequent extraction of local features. It should be noted that this embodiment does not specifically limit the definition and content of the preset size division rule; it can be set according to actual circumstances.

[0136] Step 402: Using an edge detection algorithm, identify the ridges of all grooves in each local image block, perform straight line fitting on each ridge, and calculate the angle between the fitted ridge and the horizontal baseline.

[0137] In step 402, the ridge line refers to the center line of the grooves on the surface of the anti-counterfeiting label, which can be identified by edge detection; the horizontal baseline refers to the reference baseline established in the anti-counterfeiting label area for measuring the direction angle of the grooves, which can be determined in the following ways: the longest bottom edge in the geometric boundary of the anti-counterfeiting label area is taken as the horizontal baseline, or when the bottom edge is not clear, the tangent line of the lowest point in the boundary coordinate set is taken as the horizontal baseline, and all groove direction angles are measured relative to this baseline; the angle value is the angle between the ridge line and the horizontal baseline, reflecting the tilt direction of the grooves.

[0138] In this embodiment of the application, in each local image block, the edges of all the gullies are first detected, then the center line of each gully is found, and these center lines are fitted with straight lines. Then the angle between the fitted straight line and the horizontal baseline is calculated, and finally the specific direction angle of each gully is obtained.

[0139] Step 403: Calculate the frequency distribution of each angle value in each local image block to form angle distribution data.

[0140] In step 403, the frequency distribution refers to the proportion of different angle values ​​appearing in each local image patch. It is used to quantitatively describe the concentration and distribution characteristics of the gully direction in the region. Specifically, it is the percentage of each specific angle value in the total number of detections. For example, 15 degrees appearing 3 times accounts for 37.5% of the total number of detections (8). The angle distribution data is the result of counting the number of different orientation angles within the local image patch, reflecting the distribution of gully direction in the region.

[0141] In this embodiment of the application, the specific occurrence frequency of all gully orientation angles in the current image block is counted, and then the proportion of each angle in all gullies is calculated, thereby forming feature data that can describe the distribution of gully orientations in the region.

[0142] Step 404: Measure the distance between adjacent ridges in the same local image patch, calculate the arithmetic mean of all distance measurements in the same local image patch, and form the average value data.

[0143] In step 404, the spacing measurement value refers to the vertical distance between the center lines of two adjacent gullies, and the average spacing value is the arithmetic mean of the spacing between all adjacent gullies in the local area.

[0144] In this embodiment of the application, step 404 is used to measure the distance between each pair of adjacent gullies in the image block, and then calculate the average value of these distances to obtain feature data that can reflect the density of gullies in the area.

[0145] Step 405: Summarize the angular distribution data and average value data of all local image patches to generate texture feature vectors.

[0146] In this embodiment of the application, the angular frequency data and average value data of all local image blocks are combined in sequence to generate a complete set of digital features, which is used to comprehensively describe the texture features of the anti-counterfeiting label.

[0147] In this embodiment, the method achieves accurate quantitative description of anti-counterfeiting label texture through local subdivision feature extraction, thereby providing reliable feature basis for subsequent authenticity verification and effectively improving the accuracy and stability of verification.

[0148] To establish a more comprehensive product feature knowledge system, in some embodiments, in step 102, a multi-dimensional knowledge graph is constructed based on the material spectral data and the genuine product feature data. The product surface is then scanned based on the product's unique identifier to obtain material composition data, including:

[0149] Step 501: Group the spectral data of the materials according to their material types to form a subset of spectral data corresponding to each material type.

[0150] In step 501, the spectral data subset is a set of spectral features stored according to material type. This set of spectral features contains the reflection intensity data of the same type of material at different wavelengths.

[0151] In this embodiment, the spectral database is first organized and categorized according to material type. Spectral data belonging to the same material are stored together to form a sub-database organized according to material type, thereby providing a structured data foundation for the subsequent construction of a knowledge graph.

[0152] Step 502: Separate the texture feature dataset and the marker dataset from the genuine product feature data.

[0153] In step 502, the texture feature dataset is a set of standard features pre-stored in the anti-counterfeiting feature library, which contains the baseline texture feature data of genuine products of the same model; the mark dataset is the set of coordinates of standard laser mark points of products of the same model stored in the anti-counterfeiting feature library, which serves as a comparison benchmark.

[0154] In this embodiment of the application, two types of key data are extracted from the genuine product feature database: the first type is the groove direction and spacing data describing the surface texture features, and the second type is the coordinate data recording the spatial position of the laser marker points. These two types of data will be used as the comparison benchmarks for subsequent texture verification and laser marker verification, respectively.

[0155] Step 503: Establish the association between the spectral data subset, the texture feature dataset, and the labeled dataset, and construct a multi-dimensional knowledge graph based on the association.

[0156] In this embodiment, step 503 is used to establish the association between material type and corresponding texture features and laser marking features, and then connect these feature data with each other through nodes and edges to finally construct a multi-dimensional knowledge network that can comprehensively describe the features of the product.

[0157] Step 504: Based on the unique identifier of the product, send a command to the hyperspectral scanning device to obtain the reflectance spectrum sequence of the product surface.

[0158] In step 504, the hyperspectral scanning device refers to a dedicated imaging device capable of acquiring continuous narrow-band spectral data of an object's surface. In this application, it is deployed in the intelligent inspection system of customs supervision sites for on-site scanning of cross-border e-commerce goods. The reflectance spectral sequence is continuous measurement data of the intensity of reflected light on the surface of the goods at different wavelengths.

[0159] In this embodiment of the application, an instruction is sent to a hyperspectral scanning device based on the unique identifier of the product. The device then scans the surface of the product and records the reflection intensity of multiple bands from short wave to long wave, thereby forming a complete spectral reflectance curve.

[0160] Step 505: Generate material composition data based on the wavelength intensity distribution in the reflection spectrum sequence.

[0161] In this embodiment of the application, intensity values ​​of specific wavelengths are extracted from the reflection spectrum sequence, and then the intensity distribution patterns of these characteristic wavelengths are analyzed to generate a feature map that can reflect the material composition. This map will be used for subsequent comparison with standard data pre-stored in the knowledge graph.

[0162] In this embodiment of the application, the method achieves a comprehensive digital representation of product characteristics by constructing a multidimensional knowledge graph and conducting precise material composition analysis, thereby improving the scientific nature and accuracy of the verification process.

[0163] To improve the accuracy of material verification, in some embodiments, in step 102, the material composition data is spectrally matched with the multi-dimensional knowledge graph to obtain the spectral matching degree, including:

[0164] Step 601: Obtain the peak wavelength and intensity ratio of the material composition data, and the standard peak wavelength and standard intensity ratio of the multi-dimensional knowledge graph.

[0165] In step 601, the peak wavelength is the characteristic band position in the material composition data where the reflection intensity is significantly higher than that of the adjacent wavelengths, the intensity ratio is the ratio of the reflection intensity at that wavelength to the maximum intensity, and the standard peak wavelength and standard intensity ratio are pre-stored characteristic data of the same type of material in the multi-dimensional knowledge graph.

[0166] Step 602: Compare the peak wavelength with the standard peak wavelength point by point, and based on the comparison results, count the percentage of wavelength overlap points.

[0167] In step 602, the percentage of wavelength overlap points is the proportion of the number of actual detected feature wavelengths that match the standard feature wavelengths to the total number of detected wavelengths.

[0168] In this embodiment of the application, the measured characteristic wavelengths are compared one by one with the standard characteristic wavelengths, and the specific number of wavelengths that match is counted. Then, the percentage of this number to the total number of detected wavelengths is calculated to obtain the wavelength matching degree index.

[0169] Step 603: For all wavelength coincidence points, calculate the average absolute deviation between the intensity ratio and the standard intensity ratio.

[0170] In step 603, the mean absolute deviation is the average of the absolute differences between the measured intensity and the standard intensity at all matched wavelengths.

[0171] In this embodiment, for each wavelength matching point, the absolute difference between the measured intensity value and the standard intensity value is calculated, and then the average of the difference values ​​of all matching points is obtained to obtain the overall level for evaluating the intensity deviation.

[0172] Step 604: Based on the proportion of wavelength overlap points and the mean absolute deviation, a weighted calculation is performed to generate the spectral matching degree.

[0173] In the embodiments of this application, weighted calculation can be understood as weighted summation. The embodiments of this application do not specifically limit the weight values ​​of each item; and the spectral matching degree can be a value between 0 and 1.

[0174] For example, firstly, the spectral data of the leather material is categorized and stored, while the texture features and laser marking data corresponding to the genuine products of the brand are extracted; then, a correlation is established between the leather material and the corresponding anti-counterfeiting features, thus forming a structured product feature knowledge base; for the specific product with the number A123456, a hyperspectral scan is initiated to obtain its reflection intensity data at characteristic wavelengths such as 580nm, 620nm, and 650nm, and the corresponding material composition data is generated; finally, the material composition data is compared in detail with the standard leather data pre-stored in the knowledge base, and by calculating the intensity similarity at each characteristic wavelength, the spectral matching score of the product is finally obtained as 0.95.

[0175] In this embodiment of the application, the method achieves accurate verification of material composition through multi-dimensional feature comparison and weighted calculation, thereby providing a reliable quantitative basis for determining the authenticity of goods.

[0176] To improve the automation level of the customs clearance process, in some embodiments, step 104: when the spectral matching degree and spatial similarity simultaneously meet the corresponding preset verification thresholds, the corresponding target smart contract is automatically activated according to the emergency identifier, and a contract execution record is generated, such as... Figure 3 As shown, it includes:

[0177] Step 701: When the spectral matching degree is greater than the spectral threshold in the verification threshold and the spatial similarity is greater than the spatial threshold in the verification threshold, mark it as verified and record the timestamp when the verification is marked as passed.

[0178] In step 701, the timestamp is time data that marks the moment the verification passed.

[0179] In this embodiment of the application, the spectral matching degree and spatial similarity of the product are compared with preset thresholds respectively. When both reach the standard, the current time is recorded as the specific time of verification, and this timestamp will be used as the time reference for subsequent processes.

[0180] Step 702: Based on the emergency identifier, match the corresponding target smart contract template from the preset contract triggering rule base.

[0181] In step 702, the contract triggering rule base is a database that stores smart contract templates for different customs clearance scenarios and stores them according to their urgency.

[0182] Step 703: Write the unique identifier of the product and the timestamp into the target smart contract template to generate a target smart contract containing executable tokens.

[0183] In step 703, the executable tag is a digital label that indicates that the contract has met the conditions for execution.

[0184] In this embodiment of the application, step 703 is used to write the product number and verification pass time into the selected contract template and add an executable tag to make it a smart contract that can be executed immediately.

[0185] Step 704: Based on the executable tag, trigger the execution operation of the clearance item in the target smart contract to generate a contract execution record.

[0186] In step 704, the customs clearance procedures include necessary steps such as tariff calculation and logistics arrangements.

[0187] In this embodiment of the application, a smart contract is initiated according to the executable marker, automatically executing the preset customs clearance operation process therein, and recording each operation step and its specific result during the execution process, ultimately forming a complete execution record.

[0188] For example, the spectral matching degree of product A123456 is 0.95 and the spatial similarity degree is 0.89, both of which are higher than the threshold of 0.85. Therefore, based on the "urgent" label in the order, the fast customs clearance contract template is automatically selected. Subsequently, a smart contract containing the product number and the current timestamp is generated. This contract automatically calculates the customs duties payable and arranges a priority logistics channel for the product. After the entire process is completed, the corresponding contract execution record is generated.

[0189] In this embodiment of the application, the method achieves seamless integration of verification results and customs clearance process through the automatic triggering and execution of smart contracts.

[0190] Figure 4 This application provides a schematic diagram illustrating a specific implementation of a blockchain-based intelligent order processing system for cross-border e-commerce, with reference to... Figure 4 The system may include:

[0191] The receiving module 41 is used to receive cross-border e-commerce order information, product standard data and hyperspectral images. The cross-border e-commerce order information includes a unique product identifier and an emergency identifier.

[0192] The construction module 42 is used to construct a multi-dimensional knowledge graph based on the product standard data, trigger a product surface scan based on the product unique identifier to obtain material composition data, and perform spectral matching between the material composition data and the multi-dimensional knowledge graph to obtain the spectral matching degree.

[0193] Extraction module 43 is used to extract key region feature sets from the hyperspectral image using the U-Net image segmentation algorithm, and compare the key region feature sets with the preset anti-counterfeiting feature library to obtain spatial similarity.

[0194] The activation module 44 is used to automatically activate the corresponding target smart contract and generate a contract execution record based on the emergency identifier when the spectral matching degree and spatial similarity simultaneously meet the corresponding preset verification thresholds.

[0195] The encapsulation module 45 is used to encapsulate the material composition data, the key area feature set, and the contract execution record into a complete data packet, generate a unique hash value for the complete data packet, and write the unique hash value into the blockchain.

[0196] Module 46 is established to create a bidirectional index relationship between the unique hash value and the cross-border e-commerce order information through a blockchain smart contract, so as to realize real-time synchronous updates of order status and verification results.

[0197] The blockchain-based intelligent processing system for cross-border e-commerce orders in this application is used to implement the aforementioned blockchain-based intelligent processing method for cross-border e-commerce orders. Therefore, the specific implementation of the blockchain-based intelligent processing system for cross-border e-commerce orders can be found in the embodiment section of the blockchain-based intelligent processing method for cross-border e-commerce orders above. The specific implementation can be referred to the description of the corresponding embodiments, which will not be repeated here.

[0198] This application also provides an electronic device, including: a memory for storing a computer program; and a processor for executing the computer program to implement the steps of any of the above-described blockchain-based intelligent processing methods for cross-border e-commerce orders.

[0199] This application also provides a computer-readable storage medium storing a computer program, which, when executed by a processor, implements the steps of any of the above-described blockchain-based intelligent processing methods for cross-border e-commerce orders.

[0200] In one exemplary embodiment, the aforementioned computer-readable storage medium may include, but is not limited to, various media capable of storing computer programs, such as USB flash drives, read-only memory, random access memory, portable hard drives, magnetic disks, or optical disks.

[0201] Embodiments of the present invention also provide a computer program product, which includes a computer program that, when executed by a processor, implements the steps in any of the above embodiments of the blockchain-based intelligent processing method for cross-border e-commerce orders.

[0202] Those skilled in the art will further recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of both. To clearly illustrate the interchangeability of hardware and software, the components and steps of the various examples have been generally described in terms of functionality in the foregoing description. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of this invention.

[0203] The foregoing has provided a detailed description of the blockchain-based intelligent processing method, system, electronic device, and storage medium for cross-border e-commerce orders. Specific examples have been used to illustrate the principles and implementation methods of this application. The descriptions of the embodiments above are merely for the purpose of helping to understand the method and its core ideas. It should be noted that those skilled in the art can make various improvements and modifications to this application without departing from its principles, and these improvements and modifications also fall within the protection scope of this application.

Claims

1. A blockchain-based intelligent order processing method for cross-border e-commerce, characterized in that, include: The system receives cross-border e-commerce order information, product standard data, and hyperspectral images. The cross-border e-commerce order information includes a unique product identifier and an emergency identifier. Based on the product standard data, a multi-dimensional knowledge graph is constructed. The product surface is scanned based on the product's unique identifier to obtain material composition data. The material composition data is then spectrally matched with the multi-dimensional knowledge graph to obtain the spectral matching degree. Using the U-Net image segmentation algorithm, a key region feature set is extracted from the hyperspectral image. The key region feature set is then compared with a preset anti-counterfeiting feature library to obtain the spatial similarity. When the spectral matching degree and spatial similarity degree simultaneously meet the corresponding preset verification thresholds, the corresponding target smart contract is automatically activated according to the emergency identifier, and a contract execution record is generated. The material composition data, the key area feature set, and the contract execution record are encapsulated into a complete data packet, and a unique hash value of the complete data packet is generated and written into the blockchain. A two-way index relationship is established between the unique hash value and the cross-border e-commerce order information through a blockchain smart contract, so as to realize the real-time synchronous update of order status and verification results.

2. The method according to claim 1, characterized in that, The process involves using the U-Net image segmentation algorithm to extract key region feature sets from the hyperspectral image, comparing these key region feature sets with a preset anti-counterfeiting feature library to obtain spatial similarity, including: Using the U-Net image segmentation algorithm, texture feature vectors and laser point coordinate sets are identified from the hyperspectral image. The texture feature vectors and laser point coordinate sets are used to construct key region feature sets. Retrieve the baseline texture feature vector and baseline laser point coordinate set of the same model of product from the anti-counterfeiting feature library; The matching degree between the texture feature vector and the reference texture feature vector is calculated to generate the morphological matching degree. The offset between the laser point coordinate set and the reference laser point coordinate set is statistically analyzed to generate a three-dimensional spatial offset. Spatial similarity is generated by combining the morphological matching degree and the three-dimensional spatial offset.

3. The method according to claim 2, characterized in that, The step of identifying texture feature vectors and laser point coordinate sets from the hyperspectral image using the U-Net image segmentation algorithm includes: Based on the hyperspectral image, a mask image containing pixel-level classification results is determined using the U-Net image segmentation algorithm. The mask image includes an anti-counterfeiting label area and a laser marking area. Extract the boundary coordinates of all consecutive pixels from the anti-counterfeiting label area, and use the boundary coordinates as the geometric boundary range of the anti-counterfeiting label area. Extract the coordinates of the smallest bounding rectangle vertex of all discrete pixel clusters from the laser marking area, and use the coordinates of the smallest bounding rectangle vertex as the positioning range of the laser marking area; Within the geometric boundary, a texture feature vector is generated. Within the positioning range, the three-dimensional spatial coordinates of each laser marker point are determined by a multi-view triangulation method. The three-dimensional spatial coordinates of all laser marker points are collected to generate a laser point coordinate set.

4. The method according to claim 3, characterized in that, The step of generating a texture feature vector within the geometric boundary includes: Within the geometric boundary, multiple local image blocks are generated according to a preset size division rule; Using an edge detection algorithm, the ridges of all grooves in each local image block are identified, and a straight line is fitted to each ridge. The angle between the fitted ridge and the horizontal baseline is calculated. The frequency distribution of each angle value in each local image patch is statistically analyzed to form angle distribution data; Measure the distance between adjacent ridges in the same local image patch, calculate the arithmetic mean of all distance measurements in the same local image patch, and form the average value data; By summarizing the angular distribution data and average value data of all local image patches, a texture feature vector is generated.

5. The method according to claim 1, characterized in that, The product standard data includes material spectral data and genuine product characteristic data; Based on the product standard data, a multi-dimensional knowledge graph is constructed. The product surface is scanned based on the product's unique identifier to obtain material composition data, including: Based on the material type, the spectral data of the material are grouped to form a subset of spectral data corresponding to each material type; Separate the texture feature dataset and the marker dataset from the genuine product feature data; Establish the association between the spectral data subset, the texture feature dataset, and the labeled dataset, and construct a multi-dimensional knowledge graph based on the association; Based on the unique identifier of the product, a command is sent to the hyperspectral scanning device to obtain the reflectance spectrum sequence of the product surface; Material composition data is generated based on the wavelength intensity distribution in the reflection spectrum sequence.

6. The method according to claim 1, characterized in that, The step of performing spectral matching between the material composition data and the multi-dimensional knowledge graph to obtain the spectral matching degree includes: Obtain the peak wavelength and intensity ratio of the material composition data, and the standard peak wavelength and standard intensity ratio of the multi-dimensional knowledge graph; The peak wavelength is compared point by point with the standard peak wavelength, and the percentage of wavelength overlap points is calculated based on the comparison results. For all wavelength coincidence points, calculate the mean absolute deviation between the intensity ratio and the standard intensity ratio; The spectral matching degree is generated by weighted calculation based on the proportion of wavelength overlap points and the mean absolute deviation.

7. The method according to claim 1, characterized in that, When both spectral matching degree and spatial similarity degree simultaneously meet the corresponding preset verification thresholds, the corresponding target smart contract is automatically activated based on the emergency identifier, and a contract execution record is generated, including: When the spectral matching degree is greater than the spectral threshold in the verification threshold and the spatial similarity is greater than the spatial threshold in the verification threshold, the verification is marked as passed, and the timestamp of the verification is recorded. Based on the emergency identifier, the corresponding target smart contract template is matched from the preset contract triggering rule base; Write the unique identifier of the product and the timestamp into the target smart contract template to generate a target smart contract containing executable tokens; Based on the executable marker, the execution operation of the clearance item in the target smart contract is triggered, and a contract execution record is generated.

8. A blockchain-based intelligent order processing system for cross-border e-commerce, characterized in that, include: The receiving module is used to receive cross-border e-commerce order information, product standard data, and hyperspectral images. The cross-border e-commerce order information includes a unique product identifier and an emergency identifier. The construction module is used to construct a multi-dimensional knowledge graph based on the product standard data, trigger a product surface scan based on the product unique identifier to obtain material composition data, and perform spectral matching between the material composition data and the multi-dimensional knowledge graph to obtain the spectral matching degree. The extraction module is used to extract key region feature sets from the hyperspectral image using the U-Net image segmentation algorithm, and compare the key region feature sets with the preset anti-counterfeiting feature library to obtain the spatial similarity. The activation module is used to automatically activate the corresponding target smart contract and generate a contract execution record based on the emergency identifier when the spectral matching degree and spatial similarity simultaneously meet the corresponding preset verification thresholds. The encapsulation module is used to encapsulate the material composition data, the key area feature set, and the contract execution record into a complete data packet, generate a unique hash value for the complete data packet, and write the unique hash value into the blockchain; A module is established to create a bidirectional index relationship between the unique hash value and the cross-border e-commerce order information through a blockchain smart contract, so as to realize real-time synchronous updates of order status and verification results.

9. An electronic device, characterized in that, include: Memory, used to store computer programs; A processor, configured to execute the computer program to implement the steps of the blockchain-based intelligent processing method for cross-border e-commerce orders as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed by a processor, enables the implementation of the blockchain-based intelligent processing method for cross-border e-commerce orders as described in any one of claims 1 to 7.

Images