Anti-counterfeiting method based on pull buckle with random elements

Abstract

The invention relates to the technical field of visual anti-counterfeiting, in particular to an anti-counterfeiting method based on a pull buckle with random elements. Comprising the following steps that S1, random patterns are generated, the pattern texture of each random pattern is unique, the pattern texture of each random pattern serves as a first pattern texture to be stored in a pattern database, and each pull buckle is endowed with the unique random pattern when the pull buckle is produced; s2, acquiring a verification picture containing a random pattern uploaded by the code scanning terminal, identifying the verification picture, extracting the random pattern in the verification picture, marking the random pattern in the verification picture as a second pattern, and identifying a second pattern texture of the random pattern in the verification picture; and S3, querying whether the first pattern texture consistent with the second pattern texture exists in the pattern database or not, generating a verification result, if so, judging that the second pattern is a real pattern, and if not, judging that the second pattern is a forged pattern.

Description

Technical Field

[0001] The present invention relates to the technical field of visual anti-counterfeiting, and in particular to an anti-counterfeiting method based on a buckle with random elements. Background Art

[0002] In the field of visual anti-counterfeiting technology, pattern-based anti-counterfeiting is an important method for verifying product authenticity. Its core goal is to achieve anti-counterfeiting through the design of unique, difficult-to-replicate patterns or texture features. Pattern-based anti-counterfeiting technology leverages the recognizability and non-replicability of visual features, combined with digital methods or physical properties, to give products a unique identity, effectively distinguishing authentic products from counterfeits.

[0003] Common pattern anti-counterfeiting applications include currency anti-counterfeiting, luxury goods labels, and pharmaceutical packaging. While the technical implementation methods are diverse, the existing mainstream methods still have significant limitations. For example, static QR codes or barcodes are widely used as basic anti-counterfeiting methods. Although they are low-cost and easy to deploy, their patterns are highly standardized and can be easily copied or tampered with in batches, significantly reducing their anti-counterfeiting effectiveness. Holographic pattern technology uses laser engraving to create a three-dimensional dynamic visual effect. Although it has certain anti-counterfeiting capabilities, the production process is complex and costly. Furthermore, with the popularization of holographic imaging technology, the threshold for imitation has gradually decreased, and some high-end holographic logos have even been cracked and released into the market. Furthermore, micro-text or invisible ink technology relies on special printing processes and requires specialized tools such as magnifying glasses or ultraviolet lamps for identification. This not only increases the production costs of enterprises, but also reduces the convenience of consumers' self-verification, making it difficult to adapt to the needs of modern digital scenarios. Random texture anti-counterfeiting technologies, which have emerged in recent years, such as those that leverage the randomness of paper fiber distribution or the surface texture of specialized materials to generate unique identifiers, have improved anti-counterfeiting capabilities to a certain extent. However, these technologies still rely on the uncontrollable nature of physical media, making precise digital management difficult. Furthermore, they are susceptible to feature loss due to material wear or environmental changes, limiting their application. Dynamic QR code technology attempts to address the shortcomings of static QR codes by binding them to timestamps or limiting the number of uses. However, its reliance on real-time network verification presents security risks, such as the potential for verification failures caused by man-in-the-middle attacks or database leaks.

[0004] A comprehensive analysis reveals that existing pattern-based anti-counterfeiting technologies generally face the following core issues: First, anti-counterfeiting features are easily copied or imitated, such as the mass reproduction of standardized patterns or the reverse engineering of holographic logos. Second, it is difficult to balance the cost of technical implementation with the anti-counterfeiting effect, making high-cost solutions unfavorable for large-scale adoption. Third, there is a conflict between user experience and verification efficiency. For example, verification methods that rely on specialized tools hinder consumer participation. Summary of the Invention

[0005] The technical problem solved by the present invention is to provide an anti-counterfeiting method based on a buckle with random elements, which can improve the convenience of visual pattern anti-counterfeiting and reduce its difficulty of use, and can be applied in more scenarios.

[0006] The basic solution provided by the present invention is an anti-counterfeiting method based on a buckle with random elements, comprising the following steps: S1. Generate a random pattern, each random pattern having a unique pattern texture, store the pattern texture of the random pattern as a first pattern texture in a pattern database, and assign a unique random pattern to each buckle when producing the buckle; S2. Obtain a verification image containing a random pattern uploaded by a code scanning terminal, identify the verification image to extract the random pattern therein, mark the random pattern in the verification image as a second pattern, and identify the second pattern texture of the random pattern in the verification image; S3. Check whether the first pattern texture is consistent with the second pattern texture in the pattern database, generate a verification result, and if so, determine that the second pattern is a genuine pattern; if not, determine that the second pattern is a forged pattern.

[0007] The principle and advantage of the present invention are as follows: after the random pattern is generated, the random pattern is recorded and stored in a pattern database. One buckle corresponds to one random pattern, and each random pattern is unique. When a user purchases a product, the random pattern can be photographed by a scanning device, such as a smartphone equipped with an APP, and the photographed random pattern can be uploaded to a server for identification. After the server obtains the uploaded random pattern from the scanning terminal, it compares the random pattern uploaded by the scanning terminal with the pattern in the pattern database to identify whether there is a pattern that is consistent with it. If so, it means that the random pattern is produced by the manufacturer and is authentic; otherwise, it is a counterfeit pattern. The verification result is fed back to the user terminal.

[0008] Compared to existing technologies, this solution uses visual pattern anti-counterfeiting to achieve anti-counterfeiting through random pattern comparison. Since the generated random pattern is unique, the random pattern on each product zipper is different. During anti-counterfeiting identification, the obtained random pattern is simply compared with the pattern in the database. This method can complete anti-counterfeiting verification simply and quickly. At the same time, because the color and texture of the random pattern are random, it is more difficult to copy. Some defects such as graphic distortion and pixel blur will occur during the copying process, resulting in failure to pass identification verification, thus reducing the reproducibility of the anti-counterfeiting mark and improving the anti-counterfeiting level.

[0009] Furthermore, the S1 comprises the following steps: S11. Obtain the newly generated random pattern image, perform feature extraction on the image data using a convolutional neural network, and obtain a high-dimensional feature vector:

[0010] Where f represents the output feature vector, CNN represents convolutional neural network, I represents the matrix of input image, and θ represents the network parameters; S12, the feature vector of the newly generated random pattern and the feature vectors of random patterns already in the pattern database Perform similarity comparison:

[0011] Where Sim represents the cosine similarity, according to the preset similarity threshold ,when , judging that the random pattern has no similar pattern in the pattern database, and adopting it, otherwise not adopting it; S13, generate a unique hash value h of the feature vector through the SimHash algorithm, and combine the hash value h and the feature vector The production time of the random pattern is associated with the production batch and stored in the pattern database, and an index is generated.

[0012] The newly generated random pattern is feature extracted and compared with random patterns in the pattern database to check for identical or similar patterns. Since each product corresponds to a random pattern in this solution, the uniqueness of the random pattern must be guaranteed. Identical or similar random patterns will not be used. Used random patterns are stored in the database and indexed.

[0013] Furthermore, the step S2 includes the following steps: S21. Perform preprocessing operations on the verification image, wherein the preprocessing operations include denoising, geometric correction, and illumination normalization; S22, identify the second image in the verification picture, and extract the second pattern high-dimensional feature vector through the pre-trained ResNet neural network ; The S3 includes the following steps: S31, searching the pattern database for records matching the features of the second image, and returning features of several similar results; S32, respectively calculating the cosine similarity and Euclidean distance between the features of the second image and the features of each similar result:

[0014]

[0015] S33. Compare the calculated pre-similarity and Euclidean distance with the preset similarity threshold and distance threshold respectively to obtain a comparison result. If the comparison results of the cosine similarity and Euclidean distance of similar results both meet preset conditions, then the second pattern is determined to be a genuine pattern; otherwise, the second pattern is determined to be a forged pattern.

[0016] The image obtained during verification is first preprocessed with denoising, geometric correction, and illumination normalization to ensure the pattern is clear and the pose is correct. Its feature vector is then identified, and several random patterns closest to the feature vector are found in the pattern database. The cosine similarity and Euclidean distance of these patterns are compared one by one. If the cosine similarity and Euclidean distance comparison results of similar results meet the preset conditions, the second pattern is judged to be authentic; otherwise, it is judged to be a forgery.

[0017] Furthermore, the S2 further includes the following steps: S23. Perform Fourier transform on the processed second image to generate a spectrogram. Perform spectral energy analysis on the spectrogram to calculate the energy proportion of the high-frequency region, determine whether there is suspected moiré, and use a sliding window to detect the probability of moiré. S24, performing edge detection on the processed second image, and continuing block edge statistics to determine whether there is suspected pixelation, and detecting the probability of pixelation using wavelet high-frequency energy; S25. The second image is jointly detected by the probability of the existence of moiré patterns and the probability of the existence of pixelation to obtain a credibility score of the second image, and a risk judgment is obtained according to the credibility interval in which the credibility score is located. The credibility interval includes a low-risk interval, a medium-risk interval, and a high-risk interval. When it is a medium-risk interval, the low image quality is fed back to the scanning terminal, and the verification picture is re-obtained. When it is a high-risk area, the detection of suspected forgery is fed back to the scanning terminal, and the verification picture is re-obtained.

[0018] Furthermore, the S23 includes the following steps: S231, perform Fourier transform on the second pattern to obtain a spectrum diagram:

[0019] Where I represents the second input pattern, are coordinates in the frequency domain, representing the frequency components in the horizontal and vertical directions respectively; S232. Calculate the energy proportion of the high-frequency area:

[0020] when When it is judged that there is suspected moiré, is the preset percentage threshold; S233, when there is a suspected existence, the trained neural network is used to identify the second pattern in blocks, and the probability of each block of the image is obtained. , according to the probability of each image Determine the probability of moiré existence: .

[0021] Furthermore, the step S24 includes the following steps: S241. Calculate the gradient amplitude of each pixel point of the second pattern, and determine the number of block edge pixels according to the gradient amplitude:

[0022] in and Represent the gradient convolution kernels in the horizontal and vertical directions respectively; S242, calculate the frequency of gradient direction mutation ; S243, when , it is determined that pixelation exists.

[0023] Furthermore, the S25 includes the following steps: S251. Calculate credibility score by weighting: .

[0024] The original random pattern has a random texture. When its fine structure overlaps with the sensor grid of a photographic device or the dot distribution of a printing device in spatial frequency, it produces periodic interference fringes, also known as moiré. When counterfeiters capture the pattern through photography or scanning, they may introduce initial moiré due to insufficient device resolution or angle issues. If the copied pattern is printed or photographed again, the added sensor / dot interference further exacerbates the moiré, creating "multi-layer interference." Original random patterns are produced directly using high-precision equipment to generate a random texture without any photography or printing steps, thus avoiding moiré caused by sensors or printing dots. Counterfeit random patterns, however, undergo a process of photography or scanning, editing, and printing, each of which can introduce new interference effects, leading to noticeable moiré. Using Fourier transforms to extract the image spectrum, moiré appears as periodic peaks in the high-frequency region.

[0025] Furthermore, during the counterfeiting process, if the original pattern is compressed, enlarged, or reduced, image detail can be lost, resulting in jagged or blocky edges, known as pixelation. With each copy, image quality degrades further, and pixelation becomes more pronounced. The original pattern is produced using high-resolution technology, resulting in crisp texture detail and no pixelation. However, counterfeit random patterns are limited by the precision of the counterfeiter's equipment, making it difficult to reproduce microscopic details, resulting in pixelation. Using the Sobel operator to detect image gradients, pixelated areas appear as stair-step edges.

[0026] Furthermore, the method further comprises the following steps: S4. When the second pattern is a genuine pattern, the product information is fed back to the code scanning terminal; when the second pattern is a counterfeit pattern, an alarm message is fed back to the code scanning terminal. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] Figure 1 Schematic diagram of an embodiment of the present invention. DETAILED DESCRIPTION

[0028] The following is further described in detail through specific implementation methods: The embodiment is basically as shown in the attached Figure 1 As shown: The anti-counterfeiting method based on a pull buckle with random elements comprises the following steps: S1. Generate a random pattern, each random pattern having a unique pattern texture, store the pattern texture of the random pattern as a first pattern texture in a pattern database, and assign a unique random pattern to each buckle when producing the buckle; S2. Obtain a verification image containing a random pattern uploaded by a code scanning terminal, identify the verification image to extract the random pattern therein, mark the random pattern in the verification image as a second pattern, and identify the second pattern texture of the random pattern in the verification image; S3. Check whether the first pattern texture is consistent with the second pattern texture in the pattern database, generate a verification result, and if so, determine that the second pattern is a genuine pattern; if not, determine that the second pattern is a forged pattern.

[0029] After the random pattern is generated, it is recorded and stored in the pattern database. One buckle corresponds to one random pattern, and each random pattern is unique. When a user purchases a product, they can use a scanning device, such as a smartphone equipped with an app, to take a photo of the random pattern and upload the captured random pattern to the server for identification. The server will obtain the uploaded random pattern from the scanning terminal and compare the random pattern uploaded by the scanning terminal with the patterns in the pattern database to identify whether there is a consistent pattern. If so, it means that the random pattern is produced by the manufacturer and is authentic. Otherwise, it is a counterfeit pattern. The verification result is fed back to the user terminal.

[0030] This solution combines visual anti-counterfeiting with random pattern comparison. Since the generated random pattern is unique, each product's zipper is different. During anti-counterfeiting identification, the obtained random pattern is simply compared with the patterns in the database. This method allows for quick and easy anti-counterfeiting verification. However, due to the random nature of the color and texture of the random pattern, it is difficult to copy. This can produce defects such as graphic distortion and pixel blur, which can lead to failure in identification verification. This reduces the reproducibility of the anti-counterfeiting mark and improves the anti-counterfeiting level.

[0031] Said S1 comprises the following steps: S11. Obtain the newly generated random pattern image, perform feature extraction on the image data using a convolutional neural network, and obtain a high-dimensional feature vector:

[0032] Where f represents the output feature vector, CNN represents convolutional neural network, I represents the matrix of input image, and θ represents the network parameters; S12, the feature vector of the newly generated random pattern and the feature vectors of random patterns already in the pattern database Perform similarity comparison:

[0033] Where Sim represents the cosine similarity, according to the preset similarity threshold ,when , judging that the random pattern has no similar pattern in the pattern database, and adopting it, otherwise not adopting it; S13, generate a unique hash value h of the feature vector through the SimHash algorithm, and combine the hash value h and the feature vector The production time of the random pattern is associated with the production batch and stored in the pattern database, and an index is generated.

[0034] The newly generated random pattern is feature extracted and compared with random patterns in the pattern database to check for identical or similar patterns. Since each product corresponds to a random pattern in this solution, the uniqueness of the random pattern must be guaranteed. Identical or similar random patterns will not be used. Used random patterns are stored in the database and indexed.

[0035] The S2 comprises the following steps: S21. Perform preprocessing operations on the verification image, wherein the preprocessing operations include denoising, geometric correction, and illumination normalization; S22, identify the second image in the verification picture, and extract the second pattern high-dimensional feature vector through the pre-trained ResNet neural network ; The S3 includes the following steps: S31, searching the pattern database for records matching the features of the second image, and returning features of several similar results; S32, respectively calculating the cosine similarity and Euclidean distance between the features of the second image and the features of each similar result:

[0036]

[0037] S33. Compare the calculated pre-similarity and Euclidean distance with the preset similarity threshold and distance threshold respectively to obtain a comparison result. If the comparison results of the cosine similarity and Euclidean distance of similar results both meet preset conditions, then the second pattern is determined to be a genuine pattern; otherwise, the second pattern is determined to be a forged pattern.

[0038] The image obtained during verification is first preprocessed with denoising, geometric correction, and illumination normalization to ensure the pattern is clear and the pose is correct. Its feature vector is then identified, and several random patterns closest to the feature vector are found in the pattern database. The cosine similarity and Euclidean distance of these patterns are compared one by one. If the cosine similarity and Euclidean distance comparison results of similar results meet the preset conditions, the second pattern is judged to be authentic; otherwise, it is judged to be a forgery.

[0039] The S2 further comprises the following steps: S23. Perform Fourier transform on the processed second image to generate a spectrogram. Perform spectral energy analysis on the spectrogram to calculate the energy proportion of the high-frequency region, determine whether there is suspected moiré, and use a sliding window to detect the probability of moiré. S24, performing edge detection on the processed second image, and continuing block edge statistics to determine whether there is suspected pixelation, and detecting the probability of pixelation using wavelet high-frequency energy; S25. The second image is jointly detected by the probability of the existence of moiré patterns and the probability of the existence of pixelation to obtain a credibility score of the second image, and a risk judgment is obtained according to the credibility interval in which the credibility score is located. The credibility interval includes a low-risk interval, a medium-risk interval, and a high-risk interval. When it is a medium-risk interval, the low image quality is fed back to the scanning terminal, and the verification picture is re-obtained. When it is a high-risk area, the detection of suspected forgery is fed back to the scanning terminal, and the verification picture is re-obtained.

[0040] The S23 includes the following steps: S231, perform Fourier transform on the second pattern to obtain a spectrum diagram:

[0041] Where I represents the second input pattern, are coordinates in the frequency domain, representing the frequency components in the horizontal and vertical directions respectively; S232. Calculate the energy proportion of the high-frequency area:

[0042] when When it is judged that there is suspected moiré, is the preset percentage threshold; S233, when there is a suspected existence, the trained neural network is used to identify the second pattern in blocks, and the probability of each block of the image is obtained. , according to the probability of each image Determine the probability of moiré existence: .

[0043] The S24 includes the following steps: S241. Calculate the gradient amplitude of each pixel point of the second pattern, and determine the number of block edge pixels according to the gradient amplitude:

[0044] in and Represent the gradient convolution kernels in the horizontal and vertical directions respectively; S242, calculate the frequency of gradient direction mutation ; S243, when , it is determined that pixelation exists.

[0045] The S25 includes the following steps: S251. Calculate credibility score by weighting: .

[0046] The original random pattern has a random texture. When its fine structure overlaps with the sensor grid of a photographic device or the dot distribution of a printing device in spatial frequency, it produces periodic interference fringes, also known as moiré. When counterfeiters capture the pattern through photography or scanning, they may introduce initial moiré due to insufficient device resolution or angle issues. If the copied pattern is printed or photographed again, the added sensor / dot interference further exacerbates the moiré, creating "multi-layer interference." Original random patterns are produced directly using high-precision equipment to generate a random texture without any photography or printing steps, thus avoiding moiré caused by sensors or printing dots. Counterfeit random patterns, however, undergo a process of photography or scanning, editing, and printing, each of which can introduce new interference effects, leading to noticeable moiré. Using Fourier transforms to extract the image spectrum, moiré appears as periodic peaks in the high-frequency region.

[0047] Furthermore, during the counterfeiting process, if the original pattern is compressed, enlarged, or reduced, image detail can be lost, resulting in jagged or blocky edges, known as pixelation. With each copy, image quality degrades further, and pixelation becomes more pronounced. The original pattern is produced using high-resolution technology, resulting in crisp texture detail and no pixelation. However, counterfeit random patterns are limited by the precision of the counterfeiter's equipment, making it difficult to reproduce microscopic details, resulting in pixelation. Using the Sobel operator to detect image gradients, pixelated areas appear as stair-step edges.

[0048] The following steps are also included: S4. When the second pattern is a genuine pattern, the product information is fed back to the code scanning terminal; when the second pattern is a counterfeit pattern, an alarm message is fed back to the code scanning terminal.

[0049] The above are only embodiments of the present invention. Common knowledge such as the known specific structures and characteristics in the scheme are not described in detail here. Ordinary technicians in the field are aware of all common technical knowledge in the technical field of the invention before the application date or priority date, can obtain all existing technologies in the field, and have the ability to apply conventional experimental means before that date. Ordinary technicians in the field can improve and implement this scheme in combination with their own abilities under the inspiration given by this application. Some typical known structures or known methods should not become obstacles for ordinary technicians in the field to implement this application. It should be pointed out that for those skilled in the art, without departing from the structure of the present invention, several variations and improvements can be made, which should also be regarded as the scope of protection of the present invention. These will not affect the effect of the implementation of the present invention and the practicality of the patent. The scope of protection required by this application shall be based on the content of its claims, and the specific implementation methods and other records in the specification can be used to interpret the content of the claims.

Claims

1. An anti-counterfeiting method based on a buckle with random elements, characterized in that: The following steps are involved: S1. Generate a random pattern, each random pattern having a unique pattern texture, store the pattern texture of the random pattern as a first pattern texture in a pattern database, and assign a unique random pattern to each buckle when producing the buckle; S2. Obtain a verification image containing a random pattern uploaded by a code scanning terminal, identify the verification image to extract the random pattern therein, mark the random pattern in the verification image as a second pattern, and identify the second pattern texture of the random pattern in the verification image; S3. Check whether the first pattern texture is consistent with the second pattern texture in the pattern database, generate a verification result, and if so, determine that the second pattern is a genuine pattern; if not, determine that the second pattern is a forged pattern.

2. The anti-counterfeiting method based on a buckle with random elements according to claim 1, characterized in that: Said S1 comprises the following steps: S11. Obtain the newly generated random pattern image, perform feature extraction on the image data using a convolutional neural network, and obtain a high-dimensional feature vector: Where f represents the output feature vector, CNN represents convolutional neural network, I represents the matrix of input image, and θ represents the network parameters; S12, the feature vector of the newly generated random pattern and the feature vectors of random patterns already in the pattern database Perform similarity comparison: Where Sim represents the cosine similarity, according to the preset similarity threshold ,when , judging that the random pattern has no similar pattern in the pattern database, and adopting it, otherwise not adopting it; S13, generate a unique hash value h of the feature vector through the SimHash algorithm, and combine the hash value h and the feature vector The production time of the random pattern is associated with the production batch and stored in the pattern database, and an index is generated.

3. The anti-counterfeiting method based on a buckle with random elements according to claim 2, characterized in that: The S2 comprises the following steps: S21. Perform preprocessing operations on the verification image, wherein the preprocessing operations include denoising, geometric correction, and illumination normalization; S22, identify the second image in the verification picture, and extract the second pattern high-dimensional feature vector through the pre-trained ResNet neural network ; The S3 includes the following steps: S31, searching the pattern database for records matching the features of the second image, and returning features of several similar results; S32, respectively calculating the cosine similarity and Euclidean distance between the features of the second image and the features of each similar result: S33. Compare the calculated pre-similarity and Euclidean distance with the preset similarity threshold and distance threshold respectively to obtain a comparison result. If the comparison results of the cosine similarity and Euclidean distance of similar results both meet preset conditions, then the second pattern is determined to be a genuine pattern; otherwise, the second pattern is determined to be a forged pattern.

4. The anti-counterfeiting method based on a buckle with random elements according to claim 1, characterized in that: The S2 further comprises the following steps: S23. Perform Fourier transform on the processed second image to generate a spectrogram. Perform spectral energy analysis on the spectrogram to calculate the energy proportion of the high-frequency region, determine whether there is suspected moiré, and use a sliding window to detect the probability of moiré. S24, performing edge detection on the processed second image, and continuing block edge statistics to determine whether there is suspected pixelation, and detecting the probability of pixelation using wavelet high-frequency energy; S25. The second image is jointly detected by the probability of the existence of moiré patterns and the probability of the existence of pixelation to obtain a credibility score of the second image, and a risk judgment is obtained according to the credibility interval in which the credibility score is located. The credibility interval includes a low-risk interval, a medium-risk interval, and a high-risk interval. When it is a medium-risk interval, the low image quality is fed back to the scanning terminal, and the verification picture is re-obtained. When it is a high-risk area, the detection of suspected forgery is fed back to the scanning terminal, and the verification picture is re-obtained.

5. The anti-counterfeiting method based on a buckle with random elements according to claim 4, characterized in that: The S23 includes the following steps: S231, perform Fourier transform on the second pattern to obtain a spectrum diagram: Where I represents the second input pattern, are coordinates in the frequency domain, representing the frequency components in the horizontal and vertical directions respectively; S232. Calculate the energy proportion of the high-frequency area: when When it is judged that there is suspected moiré, is the preset percentage threshold; S233, when there is a suspected existence, the trained neural network is used to identify the second pattern in blocks, and the probability of each block of the image is obtained. , according to the probability of each image Determine the probability of moiré existence: 。 6. The anti-counterfeiting method based on a buckle with random elements according to claim 5, characterized in that: The S24 includes the following steps: S241. Calculate the gradient amplitude of each pixel point of the second pattern, and determine the number of block edge pixels according to the gradient amplitude: in and Represent the gradient convolution kernels in the horizontal and vertical directions respectively; S242, calculate the frequency of gradient direction mutation ; S243, when , it is determined that pixelation exists.

7. The anti-counterfeiting method based on a buckle with random elements according to claim 6, characterized in that: The S25 includes the following steps: S251. Calculate credibility score by weighting: 。 8. The anti-counterfeiting method based on a buckle with random elements according to claim 7, characterized in that: The following steps are also included: S4. When the second pattern is a genuine pattern, the product information is fed back to the code scanning terminal; when the second pattern is a counterfeit pattern, an alarm message is fed back to the code scanning terminal.

Images