Online real-time image recognition anti-counterfeiting verification method and system based on gold transaction

Abstract

The invention relates to the technical field of image recognition, in particular to an online real-time image recognition anti-counterfeiting verification method and system based on gold transaction. The method comprises the following steps: acquiring a micro texture image of the gold product; generating a unique digital fingerprint based on the physical unclonable feature of the microscopic texture image, and performing encryption in combination with a UTC timestamp to generate a dynamic encryption tag; extracting a feature descriptor of the micro texture image and performing texture comparison with a pre-stored database to obtain a texture matching degree; decrypting the dynamic encryption label to generate a dynamic code aging state; and according to the texture matching degree and the dynamic code aging state, performing anti-counterfeiting discrimination on the micro texture image of the gold product to obtain a genuine product discrimination result and a counterfeit discrimination result. According to the method, the micro texture image, the near infrared spectrum data, the dynamic encryption label and the block chain technology are combined, so that the accuracy, the safety and the efficiency of gold anti-counterfeiting are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of image recognition, and in particular to an online real-time image recognition anti-counterfeiting verification method and system based on gold trading. Background Art

[0002] The anti-counterfeiting of gold trading mainly relies on physical markings, such as engraving, laser coding, and two-dimensional codes. These physical markings can help verify the authenticity of gold, but in large-scale online transactions, the implementation costs of these methods are relatively high, and it is difficult to achieve real-time online verification. With the development of artificial intelligence and deep learning technologies, anti-counterfeiting technologies based on image recognition have emerged. By performing image analysis on multi-dimensional information such as the surface characteristics, luster, and texture of gold, accurate identification and anti-counterfeiting verification of gold can be achieved. With the introduction of technologies such as deep convolutional neural networks (CNNs), the accuracy and speed of image recognition technologies have been significantly improved. Especially today, with the continuous enhancement of big data processing and real-time processing capabilities, online real-time image recognition anti-counterfeiting verification methods have gradually become a standard configuration for gold trading platforms. However, currently, existing technologies usually rely on a single physical marking (such as laser engraving, two-dimensional code, etc.), which are easily imitated or tampered with, unable to ensure a high level of security. At the same time, they usually rely on a single verification step, resulting in a relatively high risk of counterfeits entering the market, thereby leading to relatively low security and credibility in gold trading. Summary of the Invention

[0003] Based on this, it is necessary to provide an online real-time image recognition anti-counterfeiting verification method and system based on gold trading to solve at least one of the above technical problems.

[0004] To achieve the above object, an online real-time image recognition anti-counterfeiting verification method based on gold trading, the method includes the following steps: Step S1: Collect microscopic texture images of gold products; generate a unique digital fingerprint based on the physical unclonable features of the microscopic texture images, and encrypt it in combination with the UTC timestamp to generate a dynamic encrypted tag; Step S2: Extract the feature descriptors of the microscopic texture images and perform texture comparison with a pre-stored database to obtain a texture matching degree; decrypt the dynamic encrypted tag to generate a dynamic code time-limited status; Step S3: Perform anti-counterfeiting discrimination on the microscopic texture images of the gold products according to the texture matching degree and the dynamic code time-limited status to obtain a genuine product discrimination result and a counterfeit product discrimination result; store the counterfeit product discrimination result in the blockchain for evidence recording, and generate a primary genuine gold verification storage record; Step S4: Based on the genuine product discrimination result, initiate a multi-level review mechanism, and use the multi-level review mechanism to conduct anti-counterfeiting discrimination on the material intrinsic properties of the microscopic texture image of the gold product again, obtain the secondary anti-counterfeiting discrimination result and synchronously store it in the blockchain, and generate a secondary gold anti-counterfeiting discrimination storage record.

[0005] Through the dynamic encryption label formed by encrypting the digital fingerprint generated based on the microscopic texture image and the UTC timestamp, the present invention can ensure that each gold product has a unique and non-forgeable identifier. This encrypted label changes dynamically over time, making each anti-counterfeiting label have the characteristic of being non-replicable, enhancing the security of the gold product. By extracting the feature descriptors of the microscopic texture image and comparing them with the database, the texture features of the gold product can be accurately identified. Combining with the abnormal absorption peak detection of the near-infrared spectrum data can effectively identify the spectrum deviation, further improving the accuracy of anti-counterfeiting discrimination. Storing the genuine gold discrimination result in the blockchain for evidence preservation not only ensures the immutability of the data but also provides consumers with a traceable verification record, increasing the transparency and market trust of the gold product. For the discrimination of genuine products, a multi-level review mechanism is initiated. By re-discriminating the material intrinsic properties, the possibility of counterfeiting gold products is reduced. This mechanism ensures multiple guarantees in the discrimination process and improves the reliability of the anti-counterfeiting technology. By combining texture features, spectrum data, and dynamic encryption labels, this multi-layer anti-counterfeiting method ensures that the gold product has a high level of anti-counterfeiting ability in all aspects and can adapt to different forgery means. Therefore, by combining microscopic texture images, near-infrared spectrum data, dynamic encryption labels, and blockchain technology, the present invention improves the accuracy, security, and efficiency of gold anti-counterfeiting.

[0006] Preferably, step S1 includes the following steps: Step S11: Collect the near-infrared spectrum data of the gold product through a near-infrared spectrometer; use an industrial microscope to collect the original microscopic image on the surface of the gold product, and perform denoising and contrast enhancement on the image to generate a microscopic enhanced image; Step S12: Perform lattice structure segmentation on the microscopic enhanced image to generate a lattice distribution binary mask; Step S13: Extract the pore regions within the mask, and count the pore number, spacing, and shape factor of the pore regions to generate a pore feature vector; Step S14: Fuse the lattice distribution binary mask and the pore feature vector and label them as physically unclonable feature data; generate a unique digital fingerprint for the gold product according to the physically unclonable feature data; Step S15: Perform AES-256 encryption on the unique digital fingerprint in combination with the UTC timestamp to generate a dynamic encryption label.

[0007] Through the combination of a near-infrared spectrometer and an industrial microscope, the present invention can obtain the physical and surface characteristics of gold products from different perspectives. The near-infrared spectral data provides information on the chemical composition of gold, while the microscopic images reveal details of the surface texture. The combination of multi-source data provides a more comprehensive basis for subsequent anti-counterfeiting discrimination. By denoising and enhancing the contrast, the quality of the microscopic images is improved, ensuring the accuracy of subsequent lattice structure segmentation and pore feature extraction. The enhancement of the images makes tiny details more obvious, facilitating the precise extraction of surface texture and pore information. The segmentation of the lattice structure and the extraction of pore region features quantify the physical properties of the gold surface. The number of pores, spacing, and shape factor are unique features of the gold material that cannot be replicated. These features, as unique digital fingerprints, add a layer of security to the anti-counterfeiting of gold products. Through the fusion of the lattice distribution binary mask and the pore feature vector, it is ensured that the generated physically unclonable feature data has a high degree of uniqueness and non-replicable attributes. These features become the unique identifier of the gold product, guaranteeing the authenticity and uniqueness of the gold product. According to the generated physically unclonable feature data, a unique digital fingerprint is generated and encrypted with AES-256 in combination with the UTC timestamp to form a dynamic encrypted label. The AES-256 encryption ensures the security and immutability of the label, while the real-time change of the dynamic encrypted label increases the difficulty of anti-counterfeiting, ensuring that the anti-counterfeiting label cannot be replicated or tampered with. The encryption and feature fusion technology of the whole process ensures that the anti-counterfeiting mechanism of the gold product cannot be easily cracked. Combining the digital fingerprint and the encrypted label not only enhances the security of the anti-counterfeiting technology but also provides a traceable digital record for the gold product, enabling consumers to verify the authenticity of the gold at any time.

[0008] Preferably, step S14 includes the following steps: Step S141: Fuse the lattice distribution binary mask and the pore feature vector and label them as physically unclonable feature data; Step S142: Extract the feature vector of the physically unclonable feature data and perform a high-dimensional mapping transformation on the feature vector to generate high-dimensional feature matrix data; Step S143: Perform a hash coding calculation on the high-dimensional feature matrix data to generate unique hash fingerprint data; Step S144: Perform an error correction process on the unique hash fingerprint data to generate a unique digital fingerprint.

[0009] By fusing the lattice distribution binary mask with the pore feature vector, the generated physically unclonable feature data has a high degree of uniqueness. The physical characteristics of each gold product are quantified based on its surface lattice structure and pore characteristics, making it impossible to replicate or forge, providing a solid foundation for anti-counterfeiting. Through high-dimensional mapping transformation of the feature vector of the physically unclonable feature data, high-dimensional feature matrix data is generated. High-dimensional mapping enables the original features to be represented more complexly and deeply, enabling the feature vector to express the subtle differences of gold products more comprehensively, further enhancing the accuracy of anti-counterfeiting discrimination. By calculating the high-dimensional feature matrix data through hash coding, unique hash fingerprint data is generated. The hash algorithm compresses the high-dimensional feature data into a unique hash value of a fixed length. This hash value not only has uniqueness but also has the characteristics of fast retrieval and comparison, facilitating subsequent verification and comparison. The generation of the hash fingerprint effectively avoids the risks of data leakage and forgery. By performing error correction on the unique hash fingerprint data, the accuracy and reliability of the fingerprint data are further improved. Error correction processing can reduce the impact caused by minor errors in the acquisition, processing, or storage process, ensuring the stability and consistency of the unique digital fingerprint. The generated unique digital fingerprint has high security and accuracy, ensuring that each gold product has an identity identifier that cannot be replicated or modified. This enables effective traceability of gold products in different links, while avoiding the risks of forgery and replacement, providing a more reliable verification method for consumers. This process combines physically unclonable features, high-dimensional feature mapping, hash algorithms, and error correction, providing multiple guarantees for the anti-counterfeiting of gold products. Whether in practical applications or any level of verification, it can ensure the uniqueness and anti-counterfeiting of data, reducing the possibility of forgery.

[0010] Preferably, the extracting the feature descriptor of the microscopic texture image and performing texture comparison with the pre-stored database in step S2 includes: Extracting the surface microstructure orientation feature of the microscopic texture image and analyzing the rotational invariance of the microscopic texture image to obtain local rotation features; Locating the texture key points of the microscopic texture image based on the surface microstructure orientation feature and the local rotation features; Calculating the local feature descriptor for each texture key point; Importing the feature descriptor into the pre-stored database for feature descriptor comparison to obtain descriptor matching pairs; Calculating the similarity of the descriptor matching pairs, and screening out the descriptor matching pair with the highest similarity and the descriptor matching pair with the lowest similarity for weighted mean processing to generate a texture matching degree.

[0011] By extracting the surface microstructure orientation features of the microscopic texture image and analyzing its rotational invariance, the present invention can ensure that even if the texture image of the gold product rotates, the consistency of the features can still be maintained. This process enables texture comparison to be unaffected by direction changes, increasing the stability and accuracy of the anti-counterfeiting technology at different angles. Based on the surface microstructure orientation features and local rotation features, the texture key points of the texture image can be accurately located. The accurate positioning of the key points provides a basis for the subsequent calculation of feature descriptors, ensuring the accuracy and reliability of the subsequent comparison process. By calculating local feature descriptors for each texture key point, the texture features of each gold product can be characterized in detail. The local feature descriptors can capture the local details in the image and provide accurate information during comparison, thus effectively distinguishing different gold products. By importing the extracted feature descriptors into a pre-stored database for comparison, the texture features similar to those of the gold products in the database can be quickly and efficiently matched. The descriptor comparison provides a basis for the subsequent calculation of the matching degree, enhancing the accuracy and speed of the verification process. Calculate the similarity of the descriptor matching pairs, and screen out the highest and lowest similarity matching pairs through weighted mean processing. This process ensures that the final texture matching degree can comprehensively consider the most relevant matching pairs, while reducing the possibility of false matching, making the matching result more stable and reliable. Through the above feature extraction, comparison, and weighted mean processing, the generated texture matching degree can accurately reflect the similarity between the gold product and the gold in the pre-stored database. This precise texture comparison mechanism improves the accuracy of gold anti-counterfeiting discrimination, making it impossible for counterfeits to easily pass the verification through traditional comparison methods.

[0012] Preferably, step S2 further includes: Perform spectral peak positioning on the near-infrared spectral data to generate spectral peak position data; Calculate the peak height, full width at half maximum, and peak area of the spectral peak position data respectively to obtain a spectral peak position feature data set; Perform peak shape analysis on the spectral peak position data based on the spectral peak position feature data set to generate absorption peak shape data; Use the spectral peak position feature data set and the absorption peak shape data to detect and screen abnormal absorption peaks in the near-infrared spectral data to generate abnormal absorption peak detection data and normal absorption peak screening data; Perform spectral deviation analysis on the abnormal absorption peak detection data and the normal absorption peak screening data to generate a spectral deviation value.

[0013] By performing peak positioning on the near-infrared spectral data, the present invention can accurately identify the peak positions in the spectrum and generate spectral peak position data. This provides a basis for subsequent spectral analysis, ensuring that each absorption peak can be accurately captured and analyzed in detail. By calculating the peak height, full width at half maximum (FWHM), and peak area of the spectral peak position data, a spectral peak position feature dataset is generated. This step can meticulously describe the morphology and characteristics of each absorption peak, increasing the depth of analysis of the spectral data, helping to identify abnormal absorption peaks and effectively distinguish normal absorption peaks. Based on the spectral peak position feature dataset, peak shape analysis is performed to generate absorption peak morphology data. This process can analyze the nature of the absorption peak from a morphological perspective, providing more dimensions for the detection of abnormal absorption peaks and enhancing the accuracy of abnormal peak identification. Combining the spectral peak position feature dataset and the absorption peak morphology data, abnormal absorption peak detection and screening are carried out on the near-infrared spectral data. Through this process, abnormal absorption peaks that are significantly different from normal absorption peaks can be effectively identified, avoiding errors caused by abnormal absorption peaks, thereby improving the accuracy of spectral data analysis. Spectral deviation analysis is performed on the abnormal absorption peak detection data and the normal absorption peak screening data to generate spectral deviation values. This analysis step can identify and quantify the deviations in the spectral data, helping to further correct the errors in the data, ensuring the accuracy of the spectral data, and providing accurate spectral information for subsequent anti-counterfeiting discrimination. Through abnormal absorption peak detection and screening, abnormal signals in the spectral data can be effectively removed, ensuring the cleanliness and reliability of the data. This reduces interference caused by noise or incorrect signals, improving the overall stability of the system and the reliability of anti-counterfeiting discrimination.

[0014] Preferably, the decryption of the dynamic encryption tag in step S2 includes: Analyzing the dynamic parameters of the dynamic encryption tag to perform hierarchical decoding of the tag structure of the dynamic encryption tag, thereby obtaining hierarchical encrypted data units; Deriving a time-varying key based on the hierarchical encrypted data units, thereby obtaining a dynamic decryption key sequence; Performing reverse mapping on the dynamic encryption tag according to the dynamic decryption key sequence, thereby obtaining dynamic decryption intermediate state data; Performing timeliness state synchronization and noise elimination on the dynamic decryption intermediate state data, thereby obtaining dynamic code timeliness state data; Finally, outputting the dynamic code timeliness state through the spatio-temporal dimension fusion of the dynamic code timeliness state data.

[0015] By analyzing the dynamic parameters of the dynamically encrypted tag and performing hierarchical decoding on them, the present invention can effectively disassemble the complex tag structure into multiple encrypted data units. This hierarchical process provides a clear structure for subsequent decryption operations, ensuring that each data unit can be processed individually, reducing the error rate and complexity in the decryption process. Based on the hierarchical encrypted data units, time-varying key derivation is carried out to generate a dynamic decryption key sequence. The time-varying key derivation enhances the security of the tag. The key is dynamically generated according to time during the decryption process, effectively avoiding the risk of static key cracking and greatly improving the anti-attack ability of the tag. By performing inverse mapping on the dynamic decryption key sequence, the intermediate state data in the dynamically encrypted tag can be effectively restored. The polymorphic algorithm can provide flexible decryption strategies in complex encryption environments, ensuring efficient data restoration in different situations and improving the adaptability and security of the decryption process. Performing timeliness state synchronization and noise elimination on the dynamic decryption intermediate state data helps to eliminate noise interference in the decryption process, ensuring that the obtained temporal data is more stable and accurate. The processing of noise elimination and timeliness state synchronization effectively improves the reliability and real-time performance of the final dynamic code, ensuring that the decryption result has high accuracy in various environments. By fusing the spatio-temporal dimension of the dynamic code timeliness state data, the information of the data in the spatial and temporal dimensions can be effectively combined. This fusion technology can enhance the timeliness and spatial accuracy in the tag decryption process, improve the adaptability to environmental changes in the dynamic decryption process, and enhance the decryption stability of the tag in complex scenarios.

[0016] Preferably, the anti-counterfeiting discrimination of the microscopic texture image of the gold product according to the texture matching degree and the dynamic code timeliness state in step S3 includes: The anti-counterfeiting discrimination of the microscopic texture image of the gold product is carried out according to the texture matching degree, the spectral deviation value and the dynamic code timeliness state. When the following conditions occur simultaneously, it is determined as the genuine product discrimination result: the texture matching degree is greater than or equal to 0.90 and less than or equal to 1.00, the spectral deviation is less than or equal to 0.05, and the dynamic code timeliness is greater than or equal to 90% and less than or equal to 100%; When the following conditions occur simultaneously, it is determined as the counterfeit product discrimination result: the texture matching degree is less than 0.90 and greater than or equal to 0.80, the spectral deviation is greater than 0.05 and less than or equal to 0.15, and the dynamic code timeliness is less than 90% and greater than or equal to 70%.

[0017] Through the combined discrimination of three core indicators, namely texture matching degree (measuring the matching degree between the microscopic texture image of the gold product and the database archive), spectral deviation (reflecting whether the near-infrared spectral characteristics of the material are abnormal), and dynamic code aging status (judging the timeliness of the digital fingerprint), the accuracy of anti-counterfeiting detection is effectively improved, and misjudgment caused by a single indicator is avoided. The introduction of the dynamic code aging status makes the anti-counterfeiting system time-sensitive. Even if an attacker copies the microscopic texture or spectral characteristics of the gold surface, they cannot forge the timeliness of the dynamic code, thus improving the anti-forgery ability of the system. The setting of the spectral deviation threshold ensures that even if the surface texture is forged, but the optical properties of the material are different, it can still be identified, preventing the situation of alloy plating forgery. A fake product discrimination interval is set (0.80 ≤ texture matching degree < 0.90, 0.05 < spectral deviation ≤ 0.15, 70% ≤ dynamic code timeliness < 90%), avoiding false rejection caused by overly strict genuine product judgment criteria and improving the adaptability of the system in practical applications. This flexible anti-counterfeiting discrimination strategy can not only accurately identify high-quality fakes but also reduce misjudgment of normal products, making the system more stable and reliable in actual industrial detection. This discrimination method is based on numerical calculation and does not rely on manual subjective judgment. It is applicable to an automated anti-counterfeiting detection system, greatly improving the detection speed and efficiency.

[0018] Preferably, step S4 includes the following steps: Step S41: Based on the genuine product discrimination result, start a multi-level review mechanism, where the multi-level review mechanism includes manual re-inspection and weight verification; Step S42: Conduct a microstructure stress response analysis on the gold product through the multi-level review mechanism to generate microstructure stress response data; Step S43: Detect the lattice resonance frequency of the gold product according to the microstructure stress response data to obtain resonance frequency characteristic data; Step S44: Use the resonance frequency characteristic data and the microstructure stress response data to conduct anti-counterfeiting discrimination on the microscopic texture image of the gold product to obtain a secondary anti-counterfeiting discrimination result and synchronously store it in the blockchain, generating a secondary gold anti-counterfeiting discrimination storage record.

[0019] Through a multi-level verification mechanism (manual recheck + weight verification), the present invention ensures a more in-depth verification of suspicious samples, preventing misjudgment caused by single detection errors. Combining with microstructural stress response analysis, it avoids the problem of ignoring the structural characteristics of the material itself by relying solely on surface texture, spectral data, or encryption tags, enhancing the reliability of the gold anti-counterfeiting system. Traditional anti-counterfeiting methods mainly rely on surface feature detection, while microstructural stress response analysis can detect changes in the stress distribution of gold at the microscopic level, preventing the production of highly imitated counterfeits through surface imitation means. Lattice resonance frequency detection is based on the physical properties of materials for discrimination. Even if counterfeits highly imitate parameters such as appearance and density, it is difficult to replicate their lattice resonance characteristics, thus greatly improving the anti-counterfeiting effect. The results of secondary anti-counterfeiting discrimination are synchronously stored in the blockchain, which can ensure the integrity and traceability of data, preventing artificial tampering or forgery of authentication results. The decentralized storage of the blockchain enables the verification of each detection record of gold products, improving the credibility of the anti-counterfeiting system and preventing repeated counterfeiting or data forgery.

[0020] In this specification, an online real-time image recognition anti-counterfeiting verification system based on gold transactions is provided for performing the above-mentioned online real-time image recognition anti-counterfeiting verification method based on gold transactions. The online real-time image recognition anti-counterfeiting verification system based on gold transactions includes: A label module for collecting microscopic texture images of gold products; generating a unique digital fingerprint based on the physical unclonable features of the microscopic texture images and encrypting it in combination with the UTC timestamp to generate a dynamic encrypted label; A feature analysis module for extracting feature descriptors of the microscopic texture images and comparing the textures with a pre-stored database to obtain a texture matching degree; decrypting the dynamic encrypted label to generate a dynamic code time-limited state; A primary anti-counterfeiting discrimination module for performing anti-counterfeiting discrimination on the microscopic texture images of gold products according to the texture matching degree and the dynamic code time-limited state to obtain genuine product discrimination results and counterfeit product discrimination results; storing the counterfeit product discrimination results in the blockchain for evidentiary recording and generating a primary genuine gold verification storage record; A secondary anti-counterfeiting discrimination module for starting a multi-level verification mechanism based on the genuine product discrimination results and performing anti-counterfeiting discrimination on the material intrinsic properties of the microscopic texture images of gold products again through the multi-level verification mechanism to obtain secondary anti-counterfeiting discrimination results and synchronously storing them in the blockchain to generate a secondary gold anti-counterfeiting discrimination storage record.

[0021] The beneficial effects of the present invention are as follows: By using the physical unclonable feature of the microscopic texture image to generate a unique digital fingerprint, it ensures that each gold product has a unique identification mark, preventing forgery and tampering. Combining with the UTC timestamp for dynamic encryption, it ensures that the digital fingerprint is independent at each time point. Even if an attacker obtains the current data, it cannot be used for future forgery. Using the extraction of microscopic texture features and comparison with the pre-stored database to calculate the texture matching degree can effectively prevent surface forgery methods such as surface gilding and laser engraving. Detecting the specific absorption peak of gold and analyzing the spectral deviation can prevent doping or composition forgery. By decrypting the label to obtain the aging status of the dynamic code, it ensures the integrity and traceability of the identity information of gold. Combining the texture matching degree, spectral deviation, and aging status of the dynamic code for the first round of anti-counterfeiting discrimination can quickly screen out genuine gold. Through blockchain evidence storage, the information of genuine gold is uploaded to the blockchain to ensure the immutability of its data and form a storage record of the authenticity of gold. If a genuine discrimination result appears, a multi-level review mechanism is activated, including manual re-inspection and weight verification, to further improve the accuracy. Through microstructural stress response analysis + resonance frequency feature detection, it ensures that the intrinsic properties of the gold material conform to the inherent physical properties of gold, preventing composition doping or counterfeits. Therefore, the present invention improves the accuracy, security, and efficiency of gold anti-counterfeiting by combining microscopic texture images, near-infrared spectral data, dynamic encryption labels, and blockchain technology. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] Figure 1 It is a schematic diagram of the step flow of an online real-time image recognition anti-counterfeiting verification method based on gold transactions; Figure 2 is Figure 1 a detailed implementation step flow schematic diagram of step S1 in Figure 3 is Figure 1 a detailed implementation step flow schematic diagram of step S4 in The realization, functional features, and advantages of the object of the present invention will be further described in conjunction with the embodiments with reference to the drawings. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0023] The technical method of the present invention will be clearly and completely described below with reference to the drawings. Obviously, the described embodiments are part of the embodiments of the present invention, rather than all of them. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative efforts belong to the scope of protection of the present invention.

[0024] In addition, the accompanying drawings are only schematic illustrations of the present invention and are not necessarily drawn to scale. The same reference numerals in the drawings denote the same or similar parts, and thus repeated descriptions thereof will be omitted. Some of the block diagrams shown in the drawings are functional entities and do not necessarily correspond to physically or logically independent entities. The functional entities may be implemented in software form, or implemented in one or more hardware modules or integrated circuits, or implemented in different networks and / or processor methods and / or microcontroller methods.

[0025] It should be understood that although the terms "first", "second", etc. may be used herein to describe various units, these units should not be limited by these terms. These terms are only used to distinguish one unit from another. For example, without departing from the scope of the exemplary embodiments, the first unit may be referred to as the second unit, and similarly the second unit may be referred to as the first unit. The term "and / or" used herein includes any and all combinations of one or more of the listed associated items.

[0026] To achieve the above object, please refer to Figures 1 to 3 , an online real-time image recognition anti-counterfeiting verification method based on gold trading, the method comprising the following steps: Step S1: Collect the microscopic texture image of the gold product; generate a unique digital fingerprint based on the physical unclonable feature of the microscopic texture image, and encrypt it in combination with the UTC timestamp to generate a dynamic encrypted label; Step S2: Extract the feature descriptor of the microscopic texture image and perform texture comparison with the pre-stored database to obtain the texture matching degree; decrypt the dynamic encrypted label to generate the dynamic code aging state; Step S3: Perform anti-counterfeiting discrimination on the microscopic texture image of the gold product according to the texture matching degree and the dynamic code aging state to obtain the genuine product discrimination result and the counterfeit product discrimination result; store the counterfeit product discrimination result in the blockchain for evidence recording, and generate a primary genuine gold verification storage record; Step S4: Based on the genuine product discrimination result, start a multi-level review mechanism, and perform material intrinsic property anti-counterfeiting discrimination on the microscopic texture image of the gold product again through the multi-level review mechanism to obtain the secondary anti-counterfeiting discrimination result and synchronously store it in the blockchain to generate a secondary gold anti-counterfeiting discrimination storage record.

[0027] The present invention can ensure that each gold product has a unique and non-forgeable identifier through a dynamic encryption label formed by encrypting a digital fingerprint generated based on a microscopic texture image and a UTC timestamp. This encryption label changes dynamically over time, making each anti-counterfeiting label non-replicable and enhancing the security of gold products. By extracting the feature descriptors of the microscopic texture image and comparing them with a database, the texture features of gold products can be accurately identified. Combining with the detection of abnormal absorption peaks in near-infrared spectral data can effectively identify spectral deviations, further improving the accuracy of anti-counterfeiting discrimination. The storage of the discrimination result of genuine gold in the blockchain for evidence preservation not only ensures the immutability of data but also provides consumers with a traceable verification record, increasing the transparency and market trust of gold products. For the discrimination of genuine products, a multi-level review mechanism is initiated. By re-discriminating the intrinsic properties of materials, the possibility of counterfeiting gold products is reduced. This mechanism ensures multiple guarantees in the discrimination process and improves the reliability of anti-counterfeiting technology. By combining texture features, spectral data, and dynamic encryption labels, this multi-faceted anti-counterfeiting method ensures that gold products have a high level of anti-counterfeiting ability in all aspects and can adapt to different forgery methods. Therefore, the present invention improves the accuracy, security, and efficiency of gold anti-counterfeiting by combining microscopic texture images, near-infrared spectral data, dynamic encryption labels, and blockchain technology.

[0028] In an embodiment of the present invention, with reference to Figure 1 as shown, it is a schematic diagram of the step flow of an online real-time image recognition anti-counterfeiting verification method based on gold transactions according to the present invention. In this example, the online real-time image recognition anti-counterfeiting verification method based on gold transactions includes the following steps: Step S1: Collect the microscopic texture image of the gold product; generate a unique digital fingerprint based on the physical unclonable characteristics of the microscopic texture image, and encrypt it in combination with the UTC timestamp to generate a dynamic encryption label; In the embodiments of the present invention, a high-precision microscopic imaging device with a resolution of not less than 5000 DPI (such as a confocal microscope or an electron microscope) is used to obtain the microscopic texture image of the gold surface, and a near-infrared (NIR) spectrometer with a wavelength range of 800 - 2500 nm is used to collect spectral data to ensure the integrity of the texture and material property data. To reduce the interference of environmental factors, the temperature is controlled at 22 ± 1 °C and the humidity is controlled at 50% ± 5% during the acquisition process. The equipment is calibrated using a standard gold sample, and the average value is taken through at least 10 repeated measurements to improve data consistency. Subsequently, the CNN and ResNet deep learning models are used to extract features from the microscopic texture image, and 200 - 500 key points are detected by combining the SIFT, ORB, or SURF algorithm. After removing duplicates and reducing dimensions, a unique digital fingerprint is generated. This fingerprint data is converted into a 64-byte hash value through the SHA-256 hash algorithm and stored in binary format to improve storage and transmission efficiency. Subsequently, the UTC timestamp is obtained based on the standard network time protocol (NTP) server, accurate to 1 millisecond, and after combining with the digital fingerprint, it is encrypted using the AES-256 symmetric encryption algorithm, and 256-bit ECC (elliptic curve cryptography) or 2048-bit RSA is used for key management. Finally, the generated dynamic encrypted label is stored in the form of a 300 DPI QR code, NFC tag, or RFID and attached to the gold product, and at the same time, it is uploaded to the blockchain or cloud database for evidence preservation. To verify the system stability, the microscopic texture data is repeatedly collected under different illuminations (500 - 1500 lux), temperatures (10 - 40 °C), and angles (0 - 90°, sampled every 10°), and a hash collision test is conducted to ensure that the fingerprint data does not repeat due to similar samples. At the same time, the decryptability of the encrypted label and the fingerprint matching situation are verified through database comparison, so as to ensure the stability and accuracy of the system.

[0029] Step S2: Extract the feature descriptors of the microscopic texture image and perform texture comparison with the pre-stored database to obtain the texture matching degree; decrypt the dynamic encrypted label to generate the dynamic code aging status; In the embodiments of the present invention, based on the collected microscopic texture images, 200 - 500 key points are extracted using SIFT, ORB, and SURF algorithms, and feature descriptors are calculated to generate 128 - dimensional local feature vectors. Subsequently, the FLANN (Fast Library for Approximate Nearest Neighbors) or BF (Brute - Force) algorithm is used to compare the extracted feature descriptors with the standard texture fingerprint data in the pre - stored database, and the Euclidean distance or Hamming distance is calculated. The matching degree is measured by cosine similarity, and the matching score range is set between 0 - 100%, with the threshold set at 85%. When the matching degree exceeds the threshold, it is determined to be the same gold product. Subsequently, abnormal absorption peak detection is performed on the near - infrared spectral data (800 - 2500 nm band) of the gold product. The Savitzky - Golay filtering method is used for spectral smoothing, the first - order derivative and second - order derivative are calculated, and the abnormal absorption peaks appearing in the spectral curve are screened. Based on the K - Sigma method, the abnormal absorption peak detection threshold μ±3σ (mean ± 3 times the standard deviation) is set to identify the spectral deviation points outside the normal range, and the spectral deviation value ΔA is calculated. For the case of ΔA>0.05 (absorbance unit AU), the deviation cause is further analyzed and compared with the standard spectral library to detect changes in material composition. During the decryption process of the dynamic encryption label, first, the encrypted data in the QR code, NFC, or RFID label is read, and the AES - 256 symmetric decryption algorithm is used to decrypt it in combination with the pre - stored key. The hash fingerprint data calculated by SHA - 256 and the UTC timestamp are parsed. It is verified whether the timestamp is within the allowable deviation range of ±5 minutes, and the decrypted hash value is compared with the original fingerprint stored in the database to ensure fingerprint consistency. Finally, according to the decryption result, the dynamic code aging status is generated, and the status information indicating successful or failed decryption is marked. For the case of decryption failure, the error type is recorded, including timestamp expiration, fingerprint mismatch, or data corruption, for further analysis.

[0030] Step S3: According to the texture matching degree and the dynamic code aging status, perform anti - counterfeiting discrimination on the microscopic texture image of the gold product to obtain the genuine product discrimination result and the counterfeit product discrimination result; store the counterfeit product discrimination result in the blockchain for evidence storage record, and generate a primary genuine gold verification storage record; In the embodiment of the present invention, based on the texture matching degree obtained in step S2, when the matching score exceeds 85%, it is considered that the texture images are highly similar, and further verification is carried out to determine whether it is a genuine product. If the matching score is lower than 85%, it is determined as a potential counterfeit and marked as the "suspected counterfeit" status. Based on the spectral deviation value ΔA in step S2, if ΔA is less than 0.05 and no abnormal absorption peaks are detected, it is considered that the spectral characteristics are normal and the gold product is genuine. If ΔA exceeds 0.05 or obvious spectral abnormal absorption peaks appear, it is considered that there is a compositional deviation in the gold product, which is a counterfeit, and a warning prompt is given. According to the decrypted dynamic code aging status in step S2, if the dynamic code timestamp is within the range of ±5 minutes, it is considered that the dynamic label has not expired and the verification is valid; if the timestamp of the dynamic code exceeds the allowable deviation range, it is prompted that the "dynamic code has expired", indicating that the gold product is a counterfeit. Based on the three data items of texture matching degree, spectral deviation value and dynamic code timeliness, the weighted average method is adopted for the final anti-counterfeiting discrimination. The matching degree weight is set to 40%, the spectral deviation value weight is set to 30%, and the dynamic code timeliness weight is set to 30% to calculate the comprehensive score. When the comprehensive score ≥ 80%, it is determined as a genuine product; when it is lower than 80%, it is determined as a counterfeit. If the comprehensive score ≥ 80%, the genuine product discrimination result is generated and marked as the "genuine product" status. If the comprehensive score < 80%, the counterfeit product discrimination result is generated and marked as the "counterfeit product" status. For the gold products determined to be genuine, the system automatically stores the genuine product discrimination result in the blockchain for evidence storage. After being stored in the blockchain, a genuine product gold verification storage record is generated to ensure the immutability of the data and provide a reliable basis for subsequent traceability.

[0031] Step S4: Based on the genuine product discrimination result, start a multi-level review mechanism, and through the multi-level review mechanism, conduct anti-counterfeiting discrimination on the micro-texture image of the gold product for the material intrinsic properties again, obtain the secondary anti-counterfeiting discrimination result and synchronously store it in the blockchain, and generate a secondary gold anti-counterfeiting discrimination storage record.

[0032] In the embodiments of the present invention, during the anti-counterfeiting verification process of gold products, when initially determined to be genuine, the system will automatically activate a multi-level review mechanism for further secondary anti-counterfeiting discrimination. Once initially determined to be genuine, the system automatically triggers the multi-level review mechanism. The review process is divided into three levels: Level 1 review (optional): The system automatically analyzes and compares data (including texture matching degree, spectral deviation value, dynamic code timeliness) for re-verification. Level 2 review: A human reviewer conducts a physical sample analysis of the gold product, further collects microscopic texture images through equipment, re-obtains near-infrared spectral data, and conducts in-depth analysis by comparing with previous data. Level 3 review: A third-party certification agency conducts a comprehensive appraisal, including physical and chemical property verifications such as component analysis and atomic spectrum detection, to ensure that all anti-counterfeiting clues are verified. During the multi-level review process, the system conducts anti-counterfeiting discrimination on the microscopic texture images and near-infrared spectral data of the gold product for material intrinsic properties. The microscopic texture image of the gold surface is re-obtained through a high-precision microscope, and deep feature extraction is performed using the latest image recognition algorithms (such as convolutional neural networks). By comparing with the standard gold samples in the pre-stored database, the similarity is calculated. If the similarity is higher than the set threshold (such as 90%), it is determined to be a gold product that conforms to the intrinsic properties. By re-collecting near-infrared spectral data and using statistical analysis methods such as PLS-DA (partial least squares discriminant analysis) to compare with the standard spectral data, it is judged whether it conforms to the spectral characteristics of the gold material. If the spectral data matches the standard data, it proves to be a gold product, otherwise it is a counterfeit. Based on the analysis results in the multi-level review mechanism, the system conducts secondary anti-counterfeiting discrimination on the gold product. If it is confirmed to be genuine at all review levels, a "secondary anti-counterfeiting discrimination result" is generated and marked as "genuine". If any level of review confirms it to be a counterfeit, a "secondary anti-counterfeiting discrimination result" is generated and marked as "counterfeit". For the gold products that are secondarily determined to be genuine, the system synchronously stores the secondary anti-counterfeiting discrimination results in the blockchain for evidence storage. At the same time, a secondary gold anti-counterfeiting discrimination storage record is generated and uploaded to the blockchain database in an immutable manner to provide a permanent verifiable certificate for the authenticity of the gold product.

[0033] As an example of the present invention, refer to Figure 2 shown, in this example, step S1 includes: Step S11: Collect the near-infrared spectral data of the gold product through a near-infrared spectrometer; collect the original microscopic image of the gold product surface using an industrial microscope, and perform denoising and contrast enhancement on the image to generate a microscopic enhanced image; Step S12: Segment the lattice structure of the microscopic enhanced image to generate a lattice distribution binary mask; Step S13: Extract the pore regions within the mask, and count the pore number, spacing, and shape factor of the pore regions to generate a pore feature vector; Step S14: Fuse the lattice distribution binary mask with the pore feature vector and label it as physical unclonable feature data; generate a unique digital fingerprint for the gold product based on the physical unclonable feature data. Step S15: Combine the UTC timestamp to perform AES-256 encryption on the unique digital fingerprint to generate a dynamic encrypted label.

[0034] In the embodiments of the present invention, a near-infrared spectrometer (800 - 2500 nm band) is used to scan gold products to collect their near-infrared spectral data. To ensure the accuracy of the data, during the collection process, the spectral resolution is ensured to be 1 nm, the data collection time is 5 seconds, and the relative reflectance method is used for spectral intensity correction. Through this method, the obtained spectral data reflects the internal chemical composition and molecular structure of the gold material. An industrial microscope (magnification of 1000x) is used to scan the gold surface to obtain the original microscopic image. To enhance the usability of the image, first, noise reduction processing is performed on the image. Gaussian filters are used for noise elimination, and histogram equalization technology is used to enhance the image contrast, thereby generating a clear microscopic enhanced image. Through this image, the minute differences in the gold surface structure can be clearly observed. The lattice structure of the microscopic enhanced image is segmented, and the adaptive threshold segmentation algorithm is used to separate the lattice structure and the amorphous part in the image. Through calculation, a lattice distribution binary mask is generated, where the proportion of the lattice structure within the mask is approximately 70%. This mask is used for further analysis of the microscopic features of the gold surface, with high accuracy and clarity. Based on the lattice distribution binary mask, the pore regions within the mask are extracted, and morphological operations (such as dilation and erosion) are used to further highlight the pore features. Through calculation, the number of pores, pore spacing, and shape factor in the pore region are statistically analyzed. For example, the number of pores is 100 per square millimeter, the pore spacing is 0.3 mm, and pores with a shape factor greater than 0.8 account for approximately 60% of the total number of pores. These pore feature data generate a pore feature vector, which reflects the microscopic pore characteristics of the gold surface material. The lattice distribution binary mask and the pore feature vector are fused to generate a unique dataset containing physical unclonable feature data. This dataset not only contains information on the microscopic texture of the gold surface but also combines the pore characteristics of the gold material, which is its unique physical fingerprint and cannot be replicated on other gold products. Based on this physical unclonable feature data, the SHA-256 hash algorithm is used to generate a digital fingerprint for the data, obtaining a unique digital fingerprint. This digital fingerprint will serve as the unique identifier of the gold product. Combining the current UTC timestamp (e.g., March 18, 2025, 15:00:00), the AES-256 symmetric encryption algorithm is used to encrypt the digital fingerprint. After encryption, a dynamic encryption label is generated. This label contains timestamp information, and the label generated by each decryption is unique and tamper-proof. This label will play an important role in the anti-counterfeiting verification process of the gold product, ensuring that a brand-new encrypted identifier can be generated during each verification to prevent forgery and tampering.

[0035] Particularly importantly, step S12 further includes the following steps: Step S121: Enhance the lattice edge features of the microscopic enhanced image to generate edge enhancement data; perform lattice boundary detection on the microscopic enhanced image according to the edge enhancement data to obtain boundary feature data; Step S122: Extract the lattice skeleton from the microscopic enhanced image through the boundary feature data to obtain lattice skeleton data; perform lattice morphology analysis on the microscopic enhanced image using the lattice skeleton data to generate morphological feature data; Step S123: Extract the lattice texture features to obtain lattice texture feature data; perform regional lattice growth segmentation on the lattice skeleton data according to the lattice texture feature data to generate initial segmentation data; Step S124: Correct the lattice contour of the initial segmentation data to obtain contour optimization data, where the lattice contour correction includes curve fitting, contour smoothing, defect repair, and boundary refinement; construct a mask based on the contour optimization data to obtain the lattice distribution binary mask.

[0036] In the embodiments of the present invention, the lattice edge features of the micro-enhanced image are enhanced by using an edge detection algorithm (such as the Sobel operator). The Sobel operator calculates the gradient of the image to identify the changing regions of the edges in the image, thereby highlighting the edge information of the lattice structure. The edge-enhanced image can more clearly display the boundaries and details of the lattice. Then, the lattice boundary detection is performed on the micro-enhanced image using the edge-enhanced data. By methods such as threshold segmentation and contour tracing, the boundary feature data of the lattice are extracted, providing clear boundary information for subsequent image analysis. After obtaining the lattice boundary features, the image is further subjected to skeleton extraction. The skeleton extraction uses a thinning algorithm (such as the Zhang-Suen thinning algorithm) to simplify the lattice region in the image into a single-pixel-width skeleton structure, extracting the skeleton data of the lattice. These skeleton data can more intuitively represent the main structural features of the lattice. Subsequently, the morphological analysis of the micro-enhanced image is performed using the lattice skeleton data to analyze the morphological features of the lattice, such as shape, symmetry, pore structure, etc. This process uses morphological operations, such as erosion, dilation, opening and closing operations, to generate morphological feature data to further describe the geometric features of the lattice. Based on the skeleton data and morphological analysis, a texture feature extraction algorithm (such as the gray-level co-occurrence matrix method (GLCM) or local binary pattern (LBP)) is used to extract the texture feature data of the lattice. These features include information such as the roughness, contrast, and directionality of the texture, which can reflect the microscopic structural characteristics of the lattice region. Using these texture feature data, the region growing algorithm is used to perform region segmentation on the lattice skeleton data. The region growing algorithm expands from the seed points and merges similar regions according to the texture features to generate the initial lattice segmentation data. After obtaining the initial segmentation data, the contour is further corrected by curve fitting technology to smooth the irregular boundaries, making the contour more accurate. Methods such as B-spline curves are used to smooth the contour to eliminate the noise and defects in the image. In addition, the missing or damaged regions in the image are filled by defect repair technology to ensure the integrity of the lattice contour. The boundary refinement operation optimizes the details of the contour through an iterative algorithm to make it more precise. After the lattice contour is optimized, binarization processing is used to convert the optimized contour data into a lattice distribution binary mask. This process clearly distinguishes the lattice region and the amorphous region, setting the pixel value of the lattice region to 1 and the pixel value of the amorphous region to 0. The final lattice distribution binary mask can accurately reflect the microscopic lattice distribution characteristics of the gold product, providing an accurate data basis for subsequent anti-counterfeiting analysis.

[0037] Preferably, step S14 includes the following steps: Step S141: Fuse the lattice distribution binary mask with the pore feature vector and label it as physical unclonable feature data; Step S142: Extract the feature vectors of the physical unclonable feature data, and perform a high-dimensional mapping transformation on the feature vectors to generate high-dimensional feature matrix data; Step S143: Perform a hash coding calculation on the high-dimensional feature matrix data to generate unique hash fingerprint data; Step S144: Perform error correction processing on the unique hash fingerprint data to generate a unique digital fingerprint.

[0038] In the embodiment of the present invention, by using the previously generated lattice distribution mask and pore feature vectors, the two are fused by the weighted average method. Each pixel point of the mask is combined with the corresponding pore feature values (such as the number of pores, spacing, shape factor) to generate a set of physical unclonable feature data. These features include microscopic texture and pore characteristics, and have uniqueness and high discrimination. The fused data set is professionally marked as "physical unclonable feature data", which will be used as the core basis for the anti-counterfeiting of gold products. Use a high-precision algorithm to extract the key feature vectors in the physical unclonable feature data. Each feature vector includes texture information, pore information, and their spatial relationships. Then, these feature vectors are transformed by high-dimensional mapping techniques such as principal component analysis (PCA) or t-SNE (t-distributed stochastic neighbor embedding) to generate a high-dimensional feature matrix data. This matrix reflects the high-dimensional data space of the microscopic structure of the gold product, and each data point represents a unique feature of the gold. Assuming that the dimension of the high-dimensional matrix is 1024×1024 and each dimension contains 1000 feature points, the original material characteristics of the gold product can be accurately reflected through these feature matrices. Input the high-dimensional feature matrix data into a hash algorithm (such as SHA-512) to perform a hash coding calculation on it. The hash algorithm compresses the high-dimensional feature matrix into a hash value of a fixed length. This hash value is the unique fingerprint of the gold product, and no matter how the input data changes, the hash value is always unique and not easily reverse-engineered. Assuming that the length of the hash value is 512 bits, each generated hash fingerprint has extremely high uniqueness and cannot be copied or forged by any means. After the hash coding calculation, in order to ensure the accuracy and stability of the fingerprint data, error correction processing is performed. Use an error correction algorithm (such as Hamming coding or Reed-Solomon algorithm) to encode the generated hash fingerprint to eliminate calculation errors or external interference. After error correction, the generated unique digital fingerprint data has extremely high fault tolerance and anti-interference ability, ensuring that each decrypted digital fingerprint is unique and accurate. Finally, the digital fingerprint will be used as the identity identifier of the gold product and stored in the blockchain together with all its historical transaction records and storage information to achieve permanent preservation and anti-tampering.

[0039] Preferably, the extraction of the feature descriptors of the microscopic texture image and the texture comparison with the pre-stored database in step S2 include: Extract the surface microstructure orientation features of the microscopic texture image, analyze the rotational invariance of the microscopic texture image, and obtain the local rotation features; Locate the texture key points of the microscopic texture image based on the surface microstructure orientation features and the local rotation features; Calculate the local feature descriptor for each texture key point; Import the feature descriptor into the pre-stored database for feature descriptor comparison to obtain the descriptor matching pairs; Calculate the similarity of the descriptor matching pairs, and screen out the descriptor matching pairs with the highest similarity and the lowest similarity for weighted mean processing to generate the texture matching degree.

[0040] In the embodiments of the present invention, the surface microstructure orientation features of the microscopic texture image are extracted by using image processing algorithms (such as the gray-level co-occurrence matrix method (GLCM) or wavelet transform). The microstructure orientation features can describe the main directionality of the surface texture and reflect the orientation of the lattice structure. Then, rotation invariance techniques (such as the scale-invariant feature transform (SIFT) or speeded up robust features (SURF) algorithm) are used to perform rotation invariance analysis on the image to obtain local rotation features. This step ensures that the texture features remain stable under rotation changes, making the features consistent at different angles and suitable for subsequent feature matching. After obtaining the surface microstructure orientation features and local rotation features, feature point localization algorithms (such as Harris corner detection or FAST corner detection) are used to determine the texture key points in the microscopic texture image. By combining the surface orientation and rotation features, these key points can accurately describe the local features of the texture, especially remaining consistent at different rotation angles, thereby enhancing the reliability of texture recognition. Once the texture key points are located, descriptor generation algorithms (such as SIFT, SURF or ORB) are used to calculate local feature descriptors for each key point. These descriptors are high-dimensional vectors generated from the texture information in the surrounding area and can accurately characterize the texture features of the local area. Each descriptor contains the image information related to its corresponding texture key point and can effectively distinguish different texture regions. The extracted feature descriptors are compared with the texture descriptors pre-stored in the database. The comparison process uses feature matching algorithms, such as brute-force matching or matching based on the FLANN (Fast Approximate Nearest Neighbor Search) algorithm. By calculating the distance between the feature descriptors (usually using the Euclidean distance or Hamming distance), one-to-one descriptor matching pairs can be obtained. Each pair of matching descriptors represents similar texture regions in the image. For each matching pair, its similarity is calculated, usually using methods such as cosine similarity or normalized mutual information measure. The higher the similarity, the more similar the two texture regions are. Then, the descriptor matching pairs with the highest similarity and the lowest similarity are selected. The features of these two extreme matching pairs are of great significance for evaluating texture similarity. Next, through weighted mean processing, the similarity values of these matching pairs are combined to generate the final texture matching degree. The weights of the weighted mean can be adjusted according to the range of similarity and the reliability of the texture region to ensure that the calculation results accurately reflect the texture similarity.

[0041] Preferably, step S2 further includes: Performing spectral peak localization on the near-infrared spectral data to generate spectral peak position data; Calculating the peak height, full width at half maximum, and peak area of the spectral peak position data respectively to obtain a spectral peak position feature data set; Performing peak shape analysis on the spectral peak position data based on the spectral peak position feature data set to generate absorption peak shape data; Using the spectral peak position feature dataset and the absorption peak morphology data, detect and screen the abnormal absorption peaks in the near-infrared spectral data to generate abnormal absorption peak detection data and normal absorption peak screening data; Perform spectral deviation analysis on the abnormal absorption peak detection data and the normal absorption peak screening data to generate a spectral deviation value.

[0042] In the embodiments of the present invention, peak positioning is performed on the near-infrared spectral data by using a spectral peak detection algorithm (such as an algorithm based on the second derivative method, the maximum derivative method, or an adaptive threshold). These algorithms can accurately identify the absorption peaks existing in the spectral data and determine their positions. The spectral peak position data generated in this step contains the wavelength or frequency information of each absorption peak, serving as the basis for subsequent analysis. Based on the spectral peak position data, standard spectral analysis methods are used to calculate the peak height, full width at half maximum (FWHM), and peak area of each spectral peak respectively. These characteristic data can further describe the morphological characteristics of each absorption peak. The peak height represents the maximum value of the absorption peak, reflecting the magnitude of the absorption intensity in the spectrum; the full width at half maximum (FWHM) represents the width of the absorption peak, reflecting the distribution range of the absorption peak; the peak area represents the area under the absorption peak, reflecting the light absorption intensity within a specific wavelength band. By calculating these characteristics, detailed data support can be provided for subsequent peak shape analysis. Through peak shape analysis (such as using Gaussian fitting, Lorentz fitting, etc.) on the spectral peak position feature dataset, the morphological characteristics of each absorption peak are further extracted. These morphological characteristics include the shape, symmetry, and goodness of fit of the absorption peak. During the analysis process, denoising processing of the peak shape is also required to reduce the influence of noise on the peak shape. The generated absorption peak morphology data will help identify whether there are abnormal absorption peaks. Using the spectral peak position feature dataset and the absorption peak morphology data, detect the abnormal absorption peaks in the spectral data through an anomaly detection algorithm (such as anomaly detection based on a statistical distribution model, an anomaly detection model based on machine learning, etc.). This step can identify abnormal absorption peaks by setting thresholds (such as thresholds for parameters such as peak height, peak width, and goodness of fit). Abnormal absorption peaks usually appear as peaks with waveforms that do not conform to the normal absorption peak morphology. Finally, through screening, the normal absorption peaks and abnormal absorption peaks are separated to generate abnormal absorption peak detection data and normal absorption peak screening data. Perform spectral deviation analysis on the abnormal absorption peak detection data and the normal absorption peak screening data. By calculating the deviation between the normal absorption peak and the abnormal absorption peak (such as the shift of the peak value, the difference in spectral intensity, the drift of the frequency, etc.), a spectral deviation value is generated.

[0043] Preferably, the decryption of the dynamic encryption tag in step S2 includes: Analyze the dynamic parameters of the dynamic encryption tag to perform hierarchical decoding of the tag structure of the dynamic encryption tag, so as to obtain hierarchical encrypted data units; Derive time-varying keys based on hierarchical encrypted data units to obtain a dynamic decryption key sequence; Inverse-map the dynamic encryption tags according to the dynamic decryption key sequence to obtain dynamic decryption intermediate-state data; Perform timeliness state synchronization and noise elimination on the dynamic decryption intermediate-state data to obtain dynamic code timeliness state data; Finally output the dynamic code timeliness state through the spatio-temporal dimension fusion of the dynamic code timeliness state data.

[0044] In the embodiments of the present invention, dynamic parameters of a dynamically encrypted tag are parsed by using a tag structure parsing algorithm (such as a hierarchical decoding method based on a data structure). These parameters generally include a timestamp, an encrypted identifier, an encryption algorithm type, and related encryption key information. The tag structure is organized by using multi-level data units, and information of each encrypted data unit can be obtained through layer-by-layer decoding. Through the hierarchical decoding of the tag structure, hierarchical encrypted data units related to time, space, and encryption type can be extracted. The dynamic parameters extracted from the hierarchical encrypted data units are used to deduce a key sequence by using a time-varying key deduction algorithm (such as an encryption algorithm based on time-series data and a key update mechanism). These time-varying keys generally contain key information at each moment in the time series, and based on the rules of the encryption algorithm, the keys change at different time points. Specifically, assuming that the timestamp data is T1 = 1632680400 and T2 = 1632680500, the key deduction will generate a new key sequence based on the change in each time period, such as K1 = 0x4F7B9D2C and K2 = 0x3A8D0B5F, to ensure the timeliness of the decryption process. The dynamically encrypted tag is inversely mapped by using the deduced dynamic decryption key sequence (such as by using an exclusive OR algorithm, a DES inverse mapping, or an AES reverse decryption technique, etc.). Through this process, the ciphertext data in the encrypted tag is combined with the key sequence, thereby gradually restoring the intermediate decrypted state data of the tag. Assuming that the encrypted data C is decrypted by using the key sequences K1 and K2, an intermediate data D_intermediate = 0x9F4C8A3E can be obtained, which represents an intermediate state of tag decryption. After obtaining the dynamically decrypted intermediate state data, timeliness state synchronization is performed. This step corrects the time deviation in the decryption process by comparing the timeliness factor (such as the system time T_sync = 1632680700) with the decrypted data. At the same time, a noise rejection algorithm (such as a Kalman filter, a smoothing algorithm, etc.) is applied to remove the noise components in the data and improve the accuracy of the data. Finally, the processed dynamically decrypted intermediate state data will become dynamically coded timeliness state data, such as T_state = 0x01A2B3C4. Finally, through the space-time dimension fusion of the dynamically coded timeliness state data (by using an algorithm based on a space-time correlation model), the time and space information is comprehensively processed to generate the final dynamically coded timeliness state. For example, the final dynamically coded timeliness state C_final = 0xB7C9D3F8 can be obtained by using the time-series information T_sync = 1632680700 in combination with data such as the device position X = 23.5 and Y = 45.6 through a space-time data fusion algorithm (such as a Kalman filter, a convolutional neural network, etc.).

[0045] Preferably, the anti-counterfeiting discrimination of the microscopic texture image of the gold product according to the texture matching degree and the dynamically coded timeliness state in step S3 includes: Perform anti-counterfeiting discrimination on the microscopic texture image of the gold product according to the texture matching degree, spectral deviation value, and dynamic code timeliness status. When the following conditions occur simultaneously, it is determined as the genuine product discrimination result: the texture matching degree is greater than or equal to 0.90 and less than or equal to 1.00, the spectral deviation is less than or equal to 0.05, and the dynamic code timeliness is greater than or equal to 90% and less than or equal to 100%. When the following conditions occur simultaneously, it is determined as the counterfeit product discrimination result: the texture matching degree is less than 0.90 and greater than or equal to 0.80, the spectral deviation is greater than 0.05 and less than or equal to 0.15, and the dynamic code timeliness is less than 90% and greater than or equal to 70%.

[0046] As an example of the present invention, refer to Figure 3 As shown, in this example, step S4 includes: Step S41: Activate a multi-level review mechanism based on the genuine product discrimination result, where the multi-level review mechanism includes manual re-inspection and weight verification; Step S42: Perform microstructural stress response analysis on the gold product through the multi-level review mechanism to generate microstructural stress response data; Step S43: Detect the lattice resonance frequency of the gold product according to the microstructural stress response data to obtain resonance frequency characteristic data; Step S44: Use the resonance frequency characteristic data and the microstructural stress response data to perform anti-counterfeiting discrimination on the microscopic texture image of the gold product, obtain the secondary anti-counterfeiting discrimination result, and synchronously store it in the blockchain to generate a secondary gold anti-counterfeiting discrimination storage record.

[0047] In the embodiments of the present invention, after the preliminary authenticity discrimination of the gold product, the system automatically activates a multi-level review mechanism. First, in the manual re-inspection step, the authenticity discrimination result generated by the manual expert verification system is verified. This step includes manual inspection of the appearance, weight, and other visible features of the gold product to ensure the accuracy of the preliminary discrimination. Secondly, the weight verification is to measure the weight of the gold product with a high-precision electronic balance (the precision can reach 0.0001 grams) and compare it with the standard weight of the gold product to ensure that the weight of the product meets the standard. If the weight of the product is within the allowable error range, the next step is continued; if there is a significant error, it is determined to be a fake and the subsequent verification is stopped. Through the multi-level review mechanism, the gold product enters the microstructural stress response analysis stage. A piezoelectric material stress sensor (precision of ±1%) is used to apply a controllable pressure or stress to the gold product to simulate the mechanical response in the actual environment. The sensor measures the reaction signal of the gold product in real time after the stress is applied, and captures the stress distribution at the microscopic level. Through a stress analysis algorithm (such as the finite element analysis method), the microstructural stress response data is obtained, which characterizes the microstructural reaction characteristics of the gold product under different stress actions. Based on the data obtained from the microstructural stress response analysis, a laser resonance frequency test device (precision of 0.01 Hz) is used to detect the lattice resonance frequency of the gold product. The device measures the resonance response of the gold product at different frequencies and extracts the resonance frequency characteristics of the gold material. The microscopic lattice structure of the gold product will generate specific resonance frequencies, and these frequency characteristics reflect its inherent physical properties, such as the elastic modulus and hardness of the lattice. By comparing the resonance frequency range of the standard gold product, the resonance frequency characteristic data is generated. After obtaining the resonance frequency characteristic data and the microstructural stress response data, the system combines the microscopic texture image of the gold product to perform anti-counterfeiting discrimination of the material intrinsic properties. First, a special anti-counterfeiting algorithm is used to fuse the resonance frequency characteristic data, the microstructural stress response data, and the characteristics of the texture image to generate a comprehensive anti-counterfeiting model. This model can identify the uniqueness of the physical properties and microscopic structure of the gold product. By comparing with the genuine product standard data in the database, a secondary anti-counterfeiting discrimination result is obtained. If the gold product meets the genuine product standard, it is marked as genuine; if it does not meet the standard, it is determined to be a fake. Finally, the system stores the secondary anti-counterfeiting discrimination result and related data (such as resonance frequency, microstructural stress response data, etc.) in the blockchain to ensure the immutability of the discrimination record, generates a valid gold anti-counterfeiting discrimination storage record, and provides a data basis for subsequent traceability.

[0048] Particularly importantly, step S43 further includes the following steps: Step S43: Detect the lattice resonance frequency of the gold product according to the microstructural stress response data to obtain the resonance frequency characteristic data; Step S431: Perform time-frequency decomposition on the microstructural stress response data to obtain time-frequency characteristic data; analyze the stress wave propagation characteristics of the microstructural stress response data based on the time-frequency characteristic data to generate wave propagation data; Step S432: Extract the stress response modal characteristics of the microstructural stress response data through the wave propagation data, and construct a lattice vibration model to generate vibration model data; calculate the lattice elastic constant of the vibration model data, and analyze the thermal vibration characteristics of the vibration model data through the lattice elastic constant to obtain thermal vibration data; Step S433: Perform a swept-frequency excitation test on the gold product based on the thermal vibration data to obtain frequency response data; identify the resonance peak of the frequency response data to generate resonance frequency characteristic data.

[0049] In the embodiments of the present invention, time-frequency analysis is performed on the microstructure stress response data of gold products. Assume that the microstructure stress response data has been collected by a high-precision stress sensor (such as a piezoelectric sensor) with a sampling rate of 10 kHz and a total data length of 5 seconds. The wavelet transform is used to analyze the time-frequency characteristics. The Daubechies 4 (db4) wavelet is selected as the base wavelet, and an appropriate decomposition level (such as 4 levels) is chosen to obtain the time-frequency characteristic data after each level of decomposition. For the wavelet transform data of each level, the time-frequency spectrogram is extracted, from which the frequency distribution within each time window can be obtained, and then the distribution of the microstructure stress response wave of the gold product in time and frequency can be observed. For example, time-frequency analysis shows that the gold product has the strongest response in the frequency band from 2 kHz to 6 kHz, and the wave amplitudes are significant at 2.5 kHz and 4 kHz. According to the time-frequency characteristic data, the propagation mode of the stress wave in the gold product is analyzed. The finite difference method or the transfer matrix method is used to simulate the propagation process of the stress wave to obtain the wave propagation data of the microstructure of the gold product. For example, it can be obtained that the wave propagation speed is 1500 m / s and the wave attenuation coefficient is 0.5 dB / cm. The main stress response modal characteristics of the gold product are extracted from the wave propagation data. These modal characteristics usually manifest as significant responses at specific frequencies. For example, at 3 kHz and 5 kHz, the gold product shows strong resonance characteristics, and the wave propagation data indicates that these frequencies are the resonance frequencies of the lattice structure of the gold product. Assume that the lattice structure of the gold product is an FCC (face-centered cubic) lattice, and the finite element analysis (FEA) method is used to model it. In this model, the elastic modulus of the material is set to 120 GPa, the Poisson's ratio is 0.3, and the lattice mesh size is set to 0.01 mm. A solver (such as ANSYS) is used to perform a vibration analysis on this model to calculate its natural frequencies. Through calculation, it is found that the gold product has significant modal responses at 3.2 kHz and 5.1 kHz. During this process, the elastic constants of the gold product lattice are calculated by simulation. Assume that a typical elastic constant is 50 GPa, and the lattice stress response characteristics of gold are calculated to obtain the thermal vibration characteristics. These data provide a basis for subsequent thermal vibration analysis. The sweep frequency excitation method is used to conduct experimental tests on the gold product. In the experiment, the excitation signal is scanned in the range from 1 kHz to 10 kHz, the sampling frequency is set to 10 kHz, and the scanning step is 1 Hz. A piezoelectric exciter is used to apply the excitation signal to the gold product, and the response at each frequency is recorded. For example, during the scanning process, the scanning frequency starts from 1 kHz and increases by 1 Hz each time until it reaches 10 kHz. At each frequency, its frequency response amplitude and phase are recorded. Under the sweep frequency excitation, a set of frequency response data is obtained.Suppose the frequency response amplitude is -25 dB at 3.1 kHz, and the response amplitude at 5.0 kHz is -18 dB, indicating that the stress response of the gold product is more obvious at these frequencies. Based on the frequency response data, the resonance peaks in the frequency response are identified through a peak detection algorithm (such as a peak search algorithm). Through the identification of the resonance peak values, the resonance frequency characteristic data are obtained. For example, 3.2 kHz and 5.1 kHz are respectively identified as the two main resonance frequencies of the gold product, and these frequencies will be used as the unique physical characteristics of the gold product for subsequent anti-counterfeiting determination.

[0050] In this specification, an online real-time image recognition anti-counterfeiting verification system based on gold trading is provided for performing the above-mentioned online real-time image recognition anti-counterfeiting verification method based on gold trading. The online real-time image recognition anti-counterfeiting verification system based on gold trading includes: A label module, configured to collect the microscopic texture image of the gold product; generate a unique digital fingerprint based on the physical unclonable feature of the microscopic texture image, and encrypt it in combination with the UTC timestamp to generate a dynamic encrypted label; A feature analysis module, configured to extract the feature descriptor of the microscopic texture image and compare the texture with a pre-stored database to obtain the texture matching degree; decrypt the dynamic encrypted label to generate the dynamic code aging status; A primary anti-counterfeiting discrimination module, configured to perform anti-counterfeiting discrimination on the microscopic texture image of the gold product according to the texture matching degree and the dynamic code aging status to obtain a genuine product discrimination result and a counterfeit product discrimination result; store the counterfeit product discrimination result in the blockchain for evidence recording, and generate a primary genuine gold verification storage record; A secondary anti-counterfeiting discrimination module, configured to start a multi-level review mechanism based on the genuine product discrimination result, and perform anti-counterfeiting discrimination on the material intrinsic properties of the microscopic texture image of the gold product again through the multi-level review mechanism to obtain a secondary anti-counterfeiting discrimination result and synchronously store it in the blockchain to generate a secondary gold anti-counterfeiting discrimination storage record.

[0051] The beneficial effect of the present invention is that a unique digital fingerprint is generated through the physical non-clonable features of the micro-texture image, ensuring that each gold product has a unique identification mark to prevent counterfeiting and tampering. Dynamic encryption is performed in combination with the UTC timestamp to ensure that the digital fingerprint is independent at each time point, and even if an attacker obtains the current data, it cannot be used for future counterfeiting. The micro-texture feature extraction is compared with the pre-stored database to calculate the texture matching degree, effectively preventing surface counterfeiting methods such as surface gold plating and laser engraving. The specific absorption peak of gold is detected, and the spectral deviation is analyzed to prevent tampering or counterfeiting by adulteration or components. By decrypting the label, the dynamic code aging status is obtained to ensure that the identity information of the gold is complete and traceable. Combined with the texture matching degree, spectral deviation and dynamic code aging status, the first round of anti-counterfeiting identification is carried out to quickly screen out genuine gold. Through blockchain evidence storage, the information of genuine gold is chained to ensure that its data cannot be tampered with, forming a storage record of gold authenticity. If a genuine identification result appears, a multi-level review mechanism is activated, including manual re-inspection and weight verification, to further improve accuracy. Through microstructure stress response analysis + resonant frequency characteristic detection, it is ensured that the intrinsic properties of gold materials conform to the inherent physical properties of gold, and that component doping or imitation is eliminated. Therefore, the present invention improves the accuracy, security and efficiency of gold anti-counterfeiting by combining microscopic texture images, near-infrared spectral data, dynamic encryption tags and blockchain technology.

[0052] Therefore, the embodiments should be regarded as illustrative and non-restrictive from all points, and the scope of the present invention is limited by the appended claims rather than the above description, and it is therefore intended that all changes falling within the meaning and range of equivalent elements of the application documents are included in the present invention.

[0053] The above description is only a specific embodiment of the present invention, so that those skilled in the art can understand or implement the present invention. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to the embodiments shown herein, but should conform to the widest scope consistent with the principles and novel features invented herein.

Claims

1. An online real-time image recognition anti-counterfeiting verification method based on gold trading, characterized in that: The following steps are involved: Step S1: Collect the micro texture image of the gold product; generate a unique digital fingerprint based on the physical unclonable features of the micro texture image, and encrypt it in combination with the UTC timestamp to generate a dynamic encrypted tag; Step S2: extract the feature descriptor of the micro texture image and compare the texture with the pre-stored database to obtain the texture matching degree; decrypt the dynamic encryption tag and generate the dynamic code time status; Step S3: Anti-counterfeiting identification is performed on the micro texture image of the gold product according to the texture matching degree and the aging state of the dynamic code to obtain the authenticity identification result and the counterfeit identification result; The counterfeit identification result is stored in the blockchain for evidence recording, generating a genuine gold verification storage record; Step S4: Based on the authenticity identification result, a multi-level review mechanism is initiated, and the micro-texture image of the gold product is again subjected to material intrinsic property anti-counterfeiting identification through the multi-level review mechanism, and a secondary anti-counterfeiting identification result is obtained and synchronously stored in the blockchain, generating a secondary gold anti-counterfeiting identification storage record.

2. The online real-time image recognition anti-counterfeiting verification method based on gold trading according to claim 1 is characterized in that: Step S1 includes the following steps: Step S11: collecting near-infrared spectrum data of the gold product by a near-infrared spectrometer; collecting the original microscopic image of the surface of the gold product by an industrial microscope, and performing denoising and contrast enhancement on the image to generate a microscopic enhanced image; Step S12: performing lattice structure segmentation on the microscopic enhanced image to generate a lattice distribution binary mask; Step S13: extracting the pore area in the mask, and counting the number, spacing and shape factor of the pores in the pore area to generate a pore feature vector; Step S14: fusing the lattice distribution binary mask and the pore feature vector and marking them as physical unclonable feature data; generating a digital fingerprint of the gold product according to the physical unclonable feature data to obtain a unique digital fingerprint; Step S15: Perform AES-256 encryption on the unique digital fingerprint in combination with the UTC timestamp to generate a dynamic encryption tag.

3. The online real-time image recognition anti-counterfeiting verification method based on gold trading according to claim 2 is characterized in that: Step S14 includes the following steps: Step S141: fusing the lattice distribution binary mask with the pore feature vector and marking them as physically unclonable feature data; Step S142: extracting feature vectors of the physical unclonable feature data, and performing high-dimensional mapping conversion on the feature vectors to generate high-dimensional feature matrix data; Step S143: performing hash coding calculation on the high-dimensional feature matrix data to generate unique hash fingerprint data; Step S144: Perform error correction processing on the unique hash fingerprint data to generate a unique digital fingerprint.

4. The online real-time image recognition anti-counterfeiting verification method based on gold trading according to claim 1 is characterized in that: Extracting feature descriptors of the micro texture image and performing texture comparison with a pre-stored database in step S2 includes: Extract the surface microstructure trend features of the microtexture image, and analyze the rotation invariance of the microtexture image to obtain the local rotation features; Locate texture key points of micro texture images based on surface micro structure orientation features and local rotation features; Calculate the local feature descriptor for each texture keypoint; Importing feature descriptors into a pre-stored database to perform feature descriptor comparison and obtain descriptor matching pairs; The similarity of the descriptor matching pairs is calculated, and the descriptor matching pairs with the highest similarity and the descriptor matching pairs with the lowest similarity are screened out for weighted average processing to generate the texture matching degree.

5. The online real-time image recognition anti-counterfeiting verification method based on gold trading according to claim 1 is characterized in that: Step S2 also includes: Perform spectral peak location on near-infrared spectral data to generate spectral peak position data; The peak height, full width at half maximum and peak area of ​​the spectral peak position data are calculated respectively to obtain the spectral peak position characteristic data set; Perform peak shape analysis on the spectral peak position data based on the spectral peak position characteristic data set to generate absorption peak morphology data; Detect and screen abnormal absorption peaks of near-infrared spectral data using spectral peak position feature data sets and absorption peak morphology data, and generate abnormal absorption peak detection data and normal absorption peak screening data; Perform spectral deviation analysis on abnormal absorption peak detection data and normal absorption peak screening data to generate spectral deviation values.

6. The online real-time image recognition anti-counterfeiting verification method based on gold trading according to claim 1 is characterized in that: Decrypting the dynamic encryption tag in step S2 includes: Parsing the dynamic parameters of the dynamic encryption tag to perform hierarchical decoding of the tag structure of the dynamic encryption tag, thereby obtaining a hierarchical encryption data unit; Derivation of time-varying keys is performed based on hierarchical encrypted data units, thereby obtaining a dynamic decryption key sequence; Reverse mapping the dynamic encryption tag according to the dynamic decryption key sequence to obtain dynamic decryption intermediate state data; Perform timeliness state synchronization and noise elimination on the dynamic decryption intermediate state data to obtain dynamic code timeliness state data; By integrating the space-time dimension of the dynamic code time-limited status data, the dynamic code time-limited status is finally output.

7. The online real-time image recognition anti-counterfeiting verification method based on gold trading according to claim 1 is characterized in that: The anti-counterfeiting determination of the micro texture image of the gold product according to the texture matching degree and the aging state of the dynamic code in step S3 includes: The anti-counterfeiting judgment is carried out on the microscopic texture image of the gold product according to the texture matching degree, spectral deviation value and dynamic code aging status. When the following conditions are present at the same time, it is judged as the genuine judgment result: the texture matching degree is greater than or equal to 0.90 and less than or equal to 1.00, the spectral deviation is less than or equal to 0.05, and the dynamic code aging is greater than or equal to 90% and less than or equal to 100%; When the following conditions occur at the same time, it is judged as a counterfeit identification result: the texture matching degree is less than 0.90 and greater than or equal to 0.80, the spectral deviation is greater than 0.05 and less than or equal to 0.15, and the dynamic code timeliness is less than 90% and greater than or equal to 70%.

8. The online real-time image recognition anti-counterfeiting verification method based on gold transaction according to claim 1, characterized in that: Step S4 includes the following steps: Step S41: starting a multi-level review mechanism based on the authenticity identification result, wherein the multi-level review mechanism includes manual review and weight verification; Step S42: performing microstructural stress response analysis on the gold product through a multi-level review mechanism to generate microstructural stress response data; Step S43: performing lattice resonance frequency detection on the gold product according to the microstructure stress response data to obtain resonance frequency characteristic data; Step S44: Use the resonance frequency characteristic data and microstructure stress response data to perform material intrinsic property anti-counterfeiting identification on the microtexture image of the gold product, obtain a secondary anti-counterfeiting identification result and store it synchronously in the blockchain, and generate a secondary gold anti-counterfeiting identification storage record.

9. An online real-time image recognition anti-counterfeiting verification system based on gold trading, characterized in that: Used to execute the online real-time image recognition anti-counterfeiting verification method based on gold trading as claimed in claim 1, the online real-time image recognition anti-counterfeiting verification system based on gold trading comprises: The label module is used to collect the micro texture image of gold products; generate a unique digital fingerprint based on the physical unclonable characteristics of the micro texture image, and encrypt it in combination with the UTC timestamp to generate a dynamic encrypted label; Feature analysis module, used to extract feature descriptors of micro texture images and compare textures with pre-stored databases to obtain texture matching degree; decrypt dynamic encryption tags and generate dynamic code validity status; The one-time anti-counterfeiting identification module is used to perform anti-counterfeiting identification on the microscopic texture image of the gold product according to the texture matching degree and the time-limited status of the dynamic code, so as to obtain the results of genuine identification and counterfeit identification; the counterfeit identification results are stored in the blockchain for evidence storage and record, and a one-time genuine gold verification storage record is generated; The secondary anti-counterfeiting identification module is used to start the multi-level verification mechanism based on the authenticity identification result, and to perform anti-counterfeiting identification of the material intrinsic properties of the micro-texture image of the gold product again through the multi-level verification mechanism, obtain the secondary anti-counterfeiting identification result and store it synchronously in the blockchain, and generate a secondary gold anti-counterfeiting identification storage record.

Images