A font-based steganographic encoding anti-counterfeiting tracing method and system and enterprise production center data platform

Abstract

The present application relates to a kind of font steganography encoding-based anti-counterfeiting traceability method, system and enterprise production center data platform, belong to anti-counterfeiting technical field.A kind of font steganography encoding-based anti-counterfeiting traceability method, comprising: S1. according to product information, generate the anti-counterfeiting code serial number corresponding to product, pass through product information and anti-counterfeiting code serial number, construct anti-counterfeiting serial number database;S2. generate hidden key through encoding encryption module, hidden key has hidden key information, hidden key information is associated with anti-counterfeiting code serial number, the production date image of each product day has hidden key information and different font;S3. the production date image generated in S2 step in day is sprayed on product;S4. production date image is parsed, obtains hidden key therein, obtains hidden key information after parsing hidden secret key, by hidden key information and anti-counterfeiting serial number database matching, trace product information and verify authenticity.It has the advantages of strong concealment and robustness.

Description

Technical Field

[0001] This invention relates to the field of anti-counterfeiting technology, and in particular to an anti-counterfeiting traceability method, system, and enterprise production center data platform based on font steganography. Background Technology

[0002] In the current market environment, counterfeit goods are rampant in various fields, from food, pharmaceuticals, health products, alcoholic beverages, and daily necessities to 3C digital electronics, jewelry, and luxury goods. Counterfeit products enter the market by imitating the appearance of genuine products and forging traceability information, infringing on consumer rights and causing significant economic losses to brands. Therefore, reliable anti-counterfeiting technology has become the core of building a genuine product protection defense. Among these, information hiding technology, as a core support in the field of anti-counterfeiting and traceability, aims to conceal key information, such as keys and identifiers, within images and text, further enhancing the concealment and tamper resistance of anti-counterfeiting information. Currently, traditional font steganography methods generally rely on parametric deformation technology, with thin-plate spline transformation being a typical example. Its core logic is to modify the control point positions of the font outline to cause subtle mechanical deformation of the font shape, thereby carrying key information. However, this type of method suffers from insufficient security, susceptibility to cracking, weak robustness, and easy decoding errors in practical applications, failing to meet the higher requirements of anti-counterfeiting. Summary of the Invention

[0003] Based on this, in order to address the shortcomings of traditional font steganography technology in terms of security, concealment, and robustness, this paper provides a method, system, and enterprise production center data platform based on font steganography for anti-counterfeiting and traceability.

[0004] A method for anti-counterfeiting and traceability based on font steganography, comprising the following steps: S1. Generate an anti-counterfeiting serial number corresponding to each product based on the product information, and build an anti-counterfeiting serial number database using the product information and the anti-counterfeiting serial number; S2. A hidden key is generated through the encoding and encryption module. The hidden key corresponds to hidden key information, which is associated with the anti-counterfeiting code serial number of each product. The production date image on the product contains hidden key information, and the font of the production date image on each product is different. S3. Inkjet print the production date image generated in step S2 onto the product; S4. Analyze the production date image to obtain the hidden key. After parsing the hidden key, obtain the hidden key information. Match the hidden key information with the anti-counterfeiting serial number database to trace the product information and verify its authenticity.

[0005] Unlike traditional explicit anti-counterfeiting technologies such as QR codes and laser marking, this method embeds hidden key information into the production date image through font steganography. By changing the color and density of the font pixels, the resulting production date font image visually presents different font outlines and styles. This steganographic feature is invisible to the naked eye and difficult to extract with ordinary image analysis tools, fundamentally preventing counterfeiters from cracking, copying, or mass-producing counterfeit information. Each product's production date has eight digits, each using a different font, making the production date image unique and unpredictable for each product. Even if a counterfeiter cracks the font steganography pattern of a product's production date image, they cannot reuse it for other products of the same date or products from other dates, significantly raising the barrier to mass counterfeiting. The hidden key information is bound one-to-one with the anti-counterfeiting code serial number, forming a unique mapping from key to serial number to product information. Even if a counterfeiter obtains a key, if they cannot match the corresponding serial number in the database, they cannot pass verification, logically eliminating vulnerabilities such as key misappropriation. The production date image generated on the same day is automatically printed using inkjet printing equipment, eliminating the need for complex additional processes on the production line. This adapts to the continuous operation requirements of high-speed production lines and solves the problem of cumbersome processes affecting production efficiency in traditional anti-counterfeiting technologies. Finally, the hidden key is obtained by parsing the production date image and then matched with the anti-counterfeiting serial number database, achieving a complete closed loop from image recognition to key extraction and database verification. The verification result is directly linked to the original production information, avoiding information distortion caused by human intervention or tampering in intermediate links, ensuring the authority and accuracy of the traceability results.

[0006] In one embodiment, the specific steps of step S1 are as follows: S11. Collect product production information, including production entity information, place of origin information, factory number, production line number, batch number, production date, shelf life, etc. S12. Match the collected production information with the corresponding products, generate a serial number for each product to form an anti-counterfeiting serial number database.

[0007] By collecting comprehensive production information such as production entity information, place of origin information, and factory number, and associating it with the generated anti-counterfeiting code serial number, a structured database is constructed. This enables traceability information to not only verify authenticity but also locate the product's production source, such as the production line and batch. This provides precise data support for quality recall and responsibility traceability, and solves the limitation of traditional anti-counterfeiting methods that can only verify authenticity but cannot trace the source.

[0008] In one embodiment, the specific steps of step S2 are as follows: S21. Perform input preprocessing and semantic markup; S22. Link the hidden key with the anti-counterfeiting code serial number one by one; S23. Perform multimodal semantic encoding and style vector generation; S24. Perform conditional neural radiation field rendering.

[0009] Through the layered process from steps S21 to S24, the hidden key and the anti-counterfeiting serial number are dynamically bound together, ensuring that each serial number corresponds to unique steganographic information. At the same time, the combination of multimodal semantic coding and conditional neural radiation field rendering ensures both the visual naturalness of the font image and avoids it from being identified as anomaly code, while also ensuring the implicitness of the key information, thus balancing the contradiction between concealment and normal appearance.

[0010] In one embodiment, the specific steps of step S21 are as follows: S211. Convert all fonts in the anti-counterfeiting font library into text descriptions, and then set the text description of each font as the corresponding predefined prompt word template; S212. Receive a 128-bit binary encryption key and a tag representing the target numeric characters; S213. Verify the key length and validity; S214. Convert character labels into text descriptions using predefined prompt word templates.

[0011] The 128-bit key undergoes length and validity verification to prevent encoding errors caused by invalid keys from the outset. Character labels are converted into text descriptions using predefined prompt word templates, ensuring standardized input of semantic information and providing high-quality foundational data for subsequent multimodal fusion, thereby improving the stability and fault tolerance of the encoding process.

[0012] In one embodiment, the anti-counterfeiting font library is constructed through the following steps: S2111. Extract the online font library to obtain the first font library; S2112. Use font editing tools to adjust and modify the fonts in the first font library to obtain the second font library; S2113. By using font editing tools, different fonts can be designed independently to obtain a third font library; S2114. Merge the second and third font libraries to obtain the anti-counterfeiting font library.

[0013] Through design and modification, an anti-counterfeiting font library containing 20 or more different fonts is obtained, corresponding to an 8-digit production date. Each digit has 20 or more different font representations, and at least 20 can be generated daily. 8The system features various production date images, and can further meet the needs of daily mass production by iterating and adding more fonts to the anti-counterfeiting font library. The font is also dynamically changed daily without a fixed pattern, which greatly increases the difficulty for counterfeiters to crack and mass-produce fakes.

[0014] In one embodiment, the specific steps of step S23 are as follows: S231. Input the text description into the text encoder to obtain the corresponding text feature vector; S232. Convert the 128-bit encryption key into an initial vector through a linear projection layer; S233. Input the transformed initial vector into the adapter network that processes the key, and output the key condition vector; S234. By fusing the text feature vector and the key condition vector through a cross-attention mechanism and performing calculations, a 64-dimensional stylized steganographic vector is generated.

[0015] By combining text encoder, linear projection layer and cross attention mechanism, text semantics and key information are deeply fused into a 64-dimensional stylized steganographic vector. This ensures that the production date image can be correctly identified as date characters, achieving dual protection of character semantic recognizability and key information concealment. It also makes the key information impossible to separate from a single feature, enhancing the ability to resist analysis attacks.

[0016] In one embodiment, the specific steps of step S24 are as follows: S241. Concatenate the coordinates (x, y) of each pixel in the target character image with the stylized steganography vector to form the query vector [x, y, S]; S242. Input the query vector [x,y,S] into the multilayer perceptron network; S243. Predict the density and color of the corresponding coordinate points and calculate the final color value of the pixels to generate a production date character image containing hidden key information.

[0017] The 64-dimensional stylized steganographic vector contains information such as font outline shape, character spacing, and character width-to-height ratio. By concatenating it with pixel coordinates as input, the multilayer perceptron network can dynamically predict the density and color for each pixel, thereby making the generated font image visually present different font outlines and styles. At the same time, it also makes the generated production date image closer to ordinary fonts in terms of edge smoothness and color naturalness, and has stronger visual concealment.

[0018] In one embodiment, the specific steps of step S3 are as follows: S31. Distribute the production date image with different fonts generated online each day to different industrial control software. The production date image has hidden key information and is associated with the anti-counterfeiting serial number. S32. The production line's inkjet printer sends an online coding request to the industrial control software; S33. The industrial control software accepts and executes the production date image inkjet printing instruction that hides the key information; S34. The inkjet printing equipment performs inkjet printing of the production date image; S35. The intelligent barcode reader scans the production date image printed on the product and reads the key information therein; S36. The key information returned to the system is correlated with the anti-counterfeiting code serial number.

[0019] Each day, a unique production date image in a different font is generated and distributed via industrial control software. Combined with inkjet printing equipment and intelligent barcode readers, this system automates the entire production line process, including online inkjet printing, intelligent barcode reading, data acquisition and association, information verification, and visual inspection. This avoids human intervention that could lead to the leakage of anti-counterfeiting information or incorrect association. The intelligent barcode reader performs real-time scanning and data acquisition after inkjet printing, returning key information for data association verification. Visual inspection equipment promptly removes substandard printed products due to missed printing, incorrect printing, blurred production date images, or failure to associate the production date with the anti-counterfeiting serial number, ensuring inkjet printing quality and preventing the failure of anti-counterfeiting information from affecting traceability. The differentiated design of different fonts gives the same batch of products diverse steganographic features, increasing the difficulty of mass counterfeiting.

[0020] In one embodiment, the specific steps of step S4 are as follows: S41. Scan the production date character image of the corresponding hidden key information on the product through a dedicated mini-program or App, and transmit the identified information back to the system; S42. The system adjusts and normalizes the size of the noisy character image after it is returned; S43. Extract the high-dimensional feature vector of the adjusted image; S44. Map the high-dimensional feature vector to a 128-dimensional vector, activate it with the Sigmoid function, and obtain the predicted encryption key; S45. The predicted encryption key is thresholded to obtain the final binary encryption key, which is then matched with the unique anti-counterfeiting serial number to trace and verify the authenticity of the product.

[0021] Through dedicated scanning and standardized processing on the system side, it adapts to image input under different shooting conditions. Simultaneously, high-dimensional feature extraction and Sigmoid activation decoding ensure accurate key recovery and database matching even with noisy or slightly tampered images. This solves the problems of traditional decoding relying on clear images and having weak anti-interference capabilities, improving the verification success rate for both consumers and businesses.

[0022] A font steganography-based anti-counterfeiting and traceability system, including The anti-counterfeiting information module is used to generate an anti-counterfeiting code serial number corresponding to each product based on the product information, and to build an anti-counterfeiting serial number database through the product information and the anti-counterfeiting code serial number. The encoding and encryption module is used to generate hidden keys. Each hidden key corresponds to hidden key information, which in turn corresponds to the anti-counterfeiting code serial number of each product. Each hidden key is matched with the production date image on each product, and the font of the production date image on each product is different. The data image coding module is used to print the production date image generated on the day onto the product. The image recognition and decoding traceability module is used to parse the production date image, obtain the hidden key, and obtain the hidden key information after parsing the hidden key. The hidden key information is then matched with the anti-counterfeiting serial number database to trace product information and verify authenticity.

[0023] A data platform for enterprise production center based on font steganography for anti-counterfeiting and traceability, wherein the data platform includes the anti-counterfeiting and traceability system based on font steganography described in the above embodiment. Attached Figure Description

[0024] Figure 1 This is a flowchart illustrating an anti-counterfeiting and traceability method based on font steganography. Figure 2 This is a flowchart illustrating the sub-steps of step S1; Figure 3 This is a flowchart illustrating the sub-steps of step S2; Figure 4 This is a flowchart illustrating the sub-steps of step S21; Figure 5 This is a flowchart illustrating the sub-steps of step S211; Figure 6 This is a flowchart illustrating the sub-steps of step S23; Figure 7 This is a flowchart illustrating the sub-steps of step S24; Figure 8 This is a flowchart illustrating the sub-steps of step S3; Figure 9 This is a flowchart illustrating the sub-steps of step S4; Figure 10 A schematic diagram of a module for an anti-counterfeiting and traceability system based on font steganography. Figure 11 This is a diagram illustrating the effect of generating date images for different fonts. Detailed Implementation

[0025] To better understand the above-mentioned objectives, features, and advantages of the present invention, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments. It should be noted that, unless otherwise specified, the embodiments and features described in these embodiments can be combined with each other.

[0026] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and therefore the scope of protection of the invention is not limited to the specific embodiments disclosed below.

[0027] The following describes, with reference to the accompanying drawings, an anti-counterfeiting and traceability method based on font steganography according to some embodiments of the present invention.

[0028] like Figures 1 to 11 As shown, this embodiment discloses an anti-counterfeiting and traceability method based on font steganography, including the following steps: S1. Generate an anti-counterfeiting serial number corresponding to each product based on the product information, and build an anti-counterfeiting serial number database using the product information and the anti-counterfeiting serial number; S2. A hidden key is generated through the encoding and encryption module. The hidden key corresponds to hidden key information, which is associated with the anti-counterfeiting code serial number of each product. The production date image on the product contains hidden key information, and the font of the production date image on each product is different. S3. Inkjet print the production date image generated in step S2 onto the product; S4. Analyze the production date image to obtain the hidden key. After parsing the hidden key, obtain the hidden key information. Match the hidden key information with the anti-counterfeiting serial number database to trace the product information and verify its authenticity.

[0029] Unlike traditional explicit anti-counterfeiting technologies such as QR codes and laser marking, this method embeds hidden key information into the production date image through font steganography. By changing the color and density of the font pixels, the resulting font image visually presents different font outlines and styles. This steganographic feature is invisible to the naked eye and difficult to extract with ordinary image analysis tools, fundamentally preventing counterfeiters from cracking, copying, or mass-producing counterfeit information. The production date of each product has eight digits, each using a different font, making the production date image unique and unpredictable for each product. Even if a counterfeiter cracks the font steganography pattern of a product's production date image, they cannot reuse it for other products of the same date or products from other dates, significantly raising the barrier to mass counterfeiting. The hidden key information is bound one-to-one with the anti-counterfeiting code serial number, forming a unique mapping from key to serial number to product information. Even if a counterfeiter obtains a key, if they cannot match the corresponding serial number in the database, they cannot pass verification, logically eliminating vulnerabilities such as key misappropriation. The production date image generated on the same day is automatically printed using inkjet printing equipment, eliminating the need for complex additional processes on the production line. This adapts to the continuous operation requirements of high-speed production lines and solves the problem of production efficiency being affected by the cumbersome processes of traditional anti-counterfeiting technologies. Finally, the hidden key is obtained by parsing the production date image and then matched with the anti-counterfeiting serial number database, achieving a complete closed loop from image recognition to key extraction to database verification. The verification result is directly linked to the original production information, avoiding information distortion caused by human intervention or tampering in intermediate links, and ensuring the authority and accuracy of the traceability results.

[0030] like Figure 2 As shown, in addition to the features of the above embodiments, this embodiment further defines the specific steps of step S1 as follows: S11. Collect product production information, including production entity information, place of origin information, factory number, production line number, batch number, production date, shelf life, etc. S12. Match the collected production information with the corresponding products, generate corresponding anti-counterfeiting serial numbers, and construct an anti-counterfeiting serial number database.

[0031] By collecting comprehensive production information such as production entity information, place of origin information, and factory number, and associating it with the generated anti-counterfeiting code serial number, a structured database is constructed. This enables traceability information to not only verify authenticity but also locate the product's production source, such as the production line and batch. This provides precise data support for quality recall and responsibility traceability, and solves the limitation of traditional anti-counterfeiting methods that can only verify authenticity but cannot trace the source.

[0032] like Figure 3 As shown, in addition to the features of the above embodiments, this embodiment further defines the specific steps of step S2 as follows: S21. Perform input preprocessing and semantic markup; S22. Link the hidden key with the anti-counterfeiting code serial number one by one; S23. Perform multimodal semantic encoding and style vector generation; S24. Perform conditional neural radiation field rendering.

[0033] Through the layered process from steps S21 to S24, the hidden key and the anti-counterfeiting serial number are dynamically bound together, ensuring that each serial number corresponds to unique steganographic information. At the same time, the combination of multimodal semantic coding and conditional neural radiation field rendering ensures both the visual naturalness of the font image and avoids it from being identified as anomaly code, while also ensuring the implicitness of the key information, thus balancing the contradiction between concealment and normal appearance.

[0034] like Figure 4 As shown, in addition to the features of the above embodiments, this embodiment further defines the specific steps of step S21 as follows: S211. Convert all fonts in the anti-counterfeiting font library into text descriptions, and then set the text description of each font as the corresponding predefined prompt word template; S212. Receive a 128-bit binary encryption key and a tag representing the target numeric characters; S213. Verify the key length and validity; S214. Convert character labels into text descriptions using predefined prompt word templates.

[0035] The 128-bit key undergoes length and validity verification to prevent encoding errors caused by invalid keys from the outset. Character labels are converted into text descriptions using predefined prompt word templates, ensuring standardized input of semantic information and providing high-quality foundational data for subsequent multimodal fusion, thereby improving the stability and fault tolerance of the encoding process.

[0036] Specifically, the system accepts two inputs, including a 128-bit binary encryption key. and a label representing the target numeric character. Characters are displayed using predefined prompt templates. The data is transformed into a rich text description T, where the predefined cue word template must include relevant descriptions such as high resolution, target digit character, font style, and background. For example, T = a high-resolution image of the digit '5' with a clear and natural font style and a solid color background. These steps embed discrete labels into a continuous, high-dimensional semantic space, laying the foundation for subsequent multimodal understanding.

[0037] like Figure 5 As shown, in addition to the features of the above embodiments, this embodiment further defines the anti-counterfeiting font library as being constructed through the following steps: S2111. Extract the online font library to obtain the first font library; S2112. Use font editing tools to adjust and modify the fonts in the first font library to obtain the second font library; S2113. By using font editing tools, different fonts can be designed independently to obtain a third font library; S2114. Merge the second and third font libraries to obtain the anti-counterfeiting font library.

[0038] Through design and modification, an anti-counterfeiting font library containing 20 or more different fonts is obtained, corresponding to an 8-digit production date. Each digit has 20 or more different font representations, and at least 20 can be generated daily. 8 The system features various production date images, and can further meet the needs of daily mass production by iterating and adding more fonts to the font library. The font is also dynamically changed daily without a fixed pattern, which greatly increases the difficulty for counterfeiters to crack and mass-produce fakes.

[0039] like Figure 6 As shown, in addition to the features of the above embodiments, this embodiment further defines the specific steps of step S23 as follows: S231. Input the text description into the text encoder to obtain the corresponding text feature vector; S232. Convert the 128-bit encryption key into an initial vector through a linear projection layer; S233. Input the transformed initial vector into the adapter network that processes the key, and output the key condition vector; S234. By fusing the text feature vector and the key condition vector through a cross-attention mechanism and performing calculations, a 64-dimensional stylized steganographic vector is generated.

[0040] By combining text encoder, linear projection layer and cross attention mechanism, text semantics and key information are deeply fused into a 64-dimensional stylized steganographic vector. This ensures that the production date image can be correctly identified as date characters, achieving dual protection of character semantic recognizability and key information concealment. It also makes the key information impossible to separate from a single feature, enhancing the ability to resist analysis attacks.

[0041] Specifically, a lightweight multimodal large model containing a text encoder and a key adapter network is adopted. The text description T is input into the text encoder to obtain the text feature vector Ftext∈Rd (Rd represents a real vector of dimension d). The 128-bit encryption key K is transformed into an initial vector through a linear projection layer and then input into the key adapter network to obtain the key condition vector Fkey∈Rd (Rd represents a real vector of dimension d). The key adapter network is composed of a multilayer perceptron. Then, the text feature vector Ftext and the key condition vector Fkey are fused through a cross-attention mechanism, according to the formula... Perform calculations, where For query vector, For key vectors, For value vectors, key vector The dimension is used to generate a 64-dimensional stylized steganographic vector S∈R64.

[0042] like Figure 7 As shown, in addition to the features of the above embodiments, this embodiment further defines the specific steps of step S24 as follows: S241. Concatenate the coordinates (x, y) of each pixel in the target character image with the stylized steganography vector to form the query vector [x, y, S]; S242. Input the query vector [x,y,S] into the multilayer perceptron network; S243. Predict the density and color of the corresponding coordinate points and calculate the final color value of the pixels to generate a production date character image containing hidden key information.

[0043] The 64-dimensional stylized steganographic vector contains information such as font outline shape, character spacing, and character width-to-height ratio. By concatenating it with pixel coordinates as input, the multilayer perceptron network can dynamically predict the density and color for each pixel, thereby making the generated font image visually present different font outlines and styles. At the same time, it also makes the generated production date image closer to ordinary fonts in terms of edge smoothness and color naturalness, and has stronger visual concealment.

[0044] Specifically, a conditional neural radiation field micro-renderer is used to concatenate the coordinates (x, y) of each pixel in the target character image with the stylized steganography vector S to form a query vector [x, y, S]. This query vector is then input into the multilayer perceptron network contained in the conditional neural radiation field micro-renderer, and the density of the pixel is predicted using a formula. With RGB colors ,Right now ,in It is a conditional neural radiation field micro-renderer. It calculates the final color value of each pixel using a volume rendering formula. The formula is: ,in The total number of sampling points in volume rendering. Let be the transmittance at the i-th sampling point. Let be the density value of the i-th sampling point. Let be the distance between the i-th sampling point and the previous sampling point. Let be the color value of the i-th sampling point. Specifically, ,in This represents the cumulative absorption of the first i-1 sampling points.

[0045] like Figure 8 As shown, in addition to the features of the above embodiments, this embodiment further defines the specific steps of step S3 as follows: S31. Distribute the unique, variable production date image with different fonts generated online each day to different industrial control software. The production date image has hidden key information and is associated with the anti-counterfeiting serial number. S32. The production line's inkjet printer sends an online coding request to the industrial control software; S33. The industrial control software accepts and executes the production date image inkjet printing instruction that hides the key information; S34. The inkjet printing equipment performs inkjet printing of the production date image; S35. The intelligent barcode reader scans the production date image printed on the product and reads the key information therein; S36. The key information returned to the system is correlated with the anti-counterfeiting code serial number.

[0046] Each day, a unique production date image in different fonts is generated and distributed via industrial control software. Combined with inkjet printing equipment and intelligent barcode readers, this system automates the entire production line process, including online inkjet printing, intelligent barcode reading, data acquisition and association, information verification, and visual inspection. This avoids information leakage or association errors caused by manual intervention. The intelligent barcode reader performs real-time scanning and data acquisition after inkjet printing, returning key information for data association detection. Visual inspection equipment promptly removes substandard printed products due to omissions, blurred production date images, or failure to correlate with the anti-counterfeiting serial number, ensuring inkjet printing quality and preventing anti-counterfeiting information failure that could affect traceability. The differentiated design of different fonts creates diverse steganographic features within the same batch of products, increasing the difficulty of mass counterfeiting.

[0047] Specifically, the marking equipment can be laser printer, high-resolution inkjet printer, or UV inkjet printer.

[0048] like Figure 9 As shown, in addition to the features of the above embodiments, this embodiment further defines the specific steps of step S4 as follows: S41. Scan the production date character image of the corresponding hidden key information on the product through a dedicated mini-program or App, and transmit the identified information back to the system; S42. The system adjusts and normalizes the size of the noisy character image after it is returned; S43. Extract the high-dimensional feature vector of the adjusted image; S44. Map the high-dimensional feature vector to a 128-dimensional vector, activate it with the Sigmoid function, and obtain the predicted encryption key; S45. The predicted encryption key is thresholded to obtain the final binary encryption key, which is then matched with the unique anti-counterfeiting serial number to trace and verify the authenticity of the product.

[0049] Through dedicated scanning and standardized processing on the system side, it adapts to image input under different shooting conditions. Simultaneously, high-dimensional feature extraction and Sigmoid activation decoding ensure accurate key recovery and database matching even with noisy or slightly tampered images. This solves the problems of traditional decoding relying on clear images and having weak anti-interference capabilities, improving the verification success rate for both consumers and businesses.

[0050] Specifically, after the system acquires a noisy character image that has been printed, photographed, or transmitted, adjusts and normalizes the image size, the preprocessed image is input into a pre-trained visual base model with frozen weights, such as DINOv2-ViT-Small. From the output of the penultimate layer of this model, the output corresponding to the [CLS] Token (classification label) is found. This output is a high-dimensional real vector (the dimension is determined by the model structure, denoted as d_vit). This vector is named Fcls and is extracted as a global feature vector representing the entire input image, or by performing average pooling on all image patch features to obtain a robust feature vector. Further, a lightweight key decoding head composed of a 3-layer multi-layer perceptron is used. The robust feature vector is input into this lightweight key decoding head, mapped to a 128-dimensional vector, and then activated by the Sigmoid function to obtain the prediction key. ,Right now Then, a thresholding process is applied to K′, where the threshold can be set to 0.5, to obtain the binary encryption key. Finally, a cyclic redundancy check (CRC) mechanism can be introduced to verify the correctness of the encryption key in the binary encryption key verification stage.

[0051] In one embodiment, a font steganography-based anti-counterfeiting traceability system is provided, comprising: The anti-counterfeiting information module is used to generate an anti-counterfeiting code serial number corresponding to each product based on the product information, and to build an anti-counterfeiting serial number database through the product information and the anti-counterfeiting code serial number. The encoding and encryption module is used to generate hidden keys. Each hidden key corresponds to hidden key information, which in turn corresponds to the anti-counterfeiting code serial number of each product. Each hidden key is matched with the production date image on each product, and the font of the production date image on each product is different. The data image coding module is used to print the production date image generated on the day onto the product. The image recognition and decoding traceability module is used to parse the production date image, obtain the hidden key, and obtain the hidden key information after parsing the hidden key. The hidden key information is then matched with the anti-counterfeiting serial number database to trace product information and verify authenticity.

[0052] like Figure 10 As shown, in one embodiment, a data platform for anti-counterfeiting and traceability of an enterprise production center based on font steganography is provided, the enterprise production center data platform including the anti-counterfeiting and traceability system based on font steganography described in the above embodiment.

[0053] Specifically, the method in the above embodiments is the core process. The anti-counterfeiting traceability system based on font steganography is the hardware or software carrier of the method. The data platform of the anti-counterfeiting traceability enterprise production center based on font steganography is the core supporting component of the system. The steps of the method are executed by the modules of the system, and the functions of the modules depend on the data processing capabilities of the data platform.

[0054] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0055] The embodiments described above are merely illustrative of several implementations of the present invention, and while the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the invention patent. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of the present invention, and these all fall within the protection scope of the present invention. Therefore, the protection scope of this invention patent should be determined by the appended claims.

Claims

1. A method for anti-counterfeiting and traceability based on font steganography, characterized in that, The method includes the following steps: S1. Generate an anti-counterfeiting serial number corresponding to each product based on the product information, and build an anti-counterfeiting serial number database using the product information and the anti-counterfeiting serial number; S2. A hidden key is generated through the encoding and encryption module. The hidden key corresponds to hidden key information, which is associated with the anti-counterfeiting code serial number of each product. The production date image on the product contains the hidden key information, and the production date image on each product is different. S3. Inkjet print the production date image generated in step S2 onto the product; S4. Analyze the production date image to obtain the hidden key. After parsing the hidden key, obtain the hidden key information. Match the hidden key information with the anti-counterfeiting serial number database to trace the product information and verify its authenticity. The specific steps of step S2 are as follows: S21. Perform input preprocessing and semantic markup, convert all fonts in the anti-counterfeiting font library into text descriptions, set the text description of each font as a corresponding predefined markup template, receive a 128-bit binary encryption key and a tag representing the target numeric character, verify the key length and validity, and convert the character tag into a text description through the predefined markup template; S22. Link the hidden key with the anti-counterfeiting code serial number one by one; S23. Perform multimodal semantic encoding and style vector generation. Input the text description into the text encoder to obtain the corresponding text feature vector. Convert the 128-bit encryption key into an initial vector through a linear projection layer. Input the converted initial vector into the adapter network that processes the key and output the key condition vector. Fuse the text feature vector and the key condition vector through a cross-attention mechanism and perform calculations to generate a 64-dimensional stylized steganography vector S. S24. Perform conditional neural radiation field rendering, concatenate the coordinates (x,y) of each pixel in the target character image with the stylized steganography vector S to form a query vector [x,y,S]. Input the query vector [x,y,S] into a multilayer perceptron network to predict the density and color of the corresponding coordinate points and calculate the final color value of the pixel, thereby generating a production date character image containing hidden key information.

2. The anti-counterfeiting and traceability method based on font steganography according to claim 1, characterized in that, The specific steps of step S1 are as follows: S11. Collect the product's production information, which includes production entity information, place of origin information, factory number, production line number, batch number, production date, and shelf life; S12. Match the collected production information with the corresponding products, generate a serial number for each product to form an anti-counterfeiting serial number database.

3. The anti-counterfeiting and traceability method based on font steganography according to claim 1, characterized in that, The anti-counterfeiting font library is constructed through the following steps: S211. Extract the online font library to obtain the first font library; S212. Use font editing tools to adjust and modify the fonts in the first font library to obtain the second font library; S213. By using font editing tools, different fonts can be designed independently to obtain a third font library; S214. Merge the second and third font libraries to obtain the anti-counterfeiting font library.

4. The anti-counterfeiting and traceability method based on font steganography according to claim 1, characterized in that, The specific steps of step S3 are as follows: S31. Distribute the production date image with a variable font, which is generated online daily and has a unique date for that day, to different industrial control software. The production date image has hidden key information and is associated with the anti-counterfeiting serial number. S32. The production line's inkjet printer sends an online coding request to the industrial control software; S33. The industrial control software accepts and executes the production date image inkjet printing instruction that hides the key information; S34. The inkjet printing equipment performs inkjet printing of the production date image; S35. The intelligent barcode reader scans the production date image printed on the product and reads the key information therein; S36. The key information returned to the system is correlated with the anti-counterfeiting code serial number.

5. The anti-counterfeiting and traceability method based on font steganography according to claim 1, characterized in that, The specific steps of step S4 are as follows: S41. Scan the production date character image of the corresponding hidden key information on the product through a dedicated mini-program or App, and transmit the identified information back to the system; S42. The system adjusts and normalizes the size of the noisy character image after it is returned; S43. Extract the high-dimensional feature vector of the adjusted image; S44. Map the high-dimensional feature vector to a 128-dimensional vector, activate it with the Sigmoid function, and obtain the predicted encryption key; S45. The predicted encryption key is thresholded to obtain the final binary encryption key, which is then matched with the unique anti-counterfeiting serial number to trace and verify the authenticity of the product.

6. A counterfeit-proof traceability system based on font steganography, characterized in that, include: The anti-counterfeiting information module is used to generate an anti-counterfeiting code serial number corresponding to each product based on the product information, and to build an anti-counterfeiting serial number database through the product information and the anti-counterfeiting code serial number. The encoding and encryption module is used to generate hidden keys. Each hidden key corresponds to hidden key information, which in turn corresponds to the anti-counterfeiting code serial number of each product. Each hidden key is matched with the production date image on each product, and the font of the production date image on each product is different. The encoding and encryption module generates a production date image containing hidden key information by performing the following steps: input preprocessing and semantic markup; converting all fonts in the anti-counterfeiting font library into text descriptions; setting the text description of each font as a corresponding predefined markup template; receiving a 128-bit binary encryption key and a tag representing the target numeric character; verifying the key length and validity; converting the character tag into a text description using the predefined markup template; associating the hidden key with the anti-counterfeiting code serial number; performing multimodal semantic encoding and style vector generation; inputting the text description into the text encoder to obtain the corresponding text feature vector; and then... A 128-bit encryption key is converted into an initial vector through a linear projection layer. The converted initial vector is then input into an adapter network that processes the key, outputting a key condition vector. The text feature vector and the key condition vector are fused through a cross-attention mechanism and calculated to generate a 64-dimensional stylized steganography vector S. Conditional neural radiation field rendering is then performed, concatenating the coordinates (x, y) of each pixel in the target character image with the stylized steganography vector S to form a query vector [x, y, S]. The query vector [x, y, S] is then input into a multilayer perceptron network to predict the density and color of the corresponding coordinates and calculate the final color value of the pixel, generating a character image containing the production date information hidden in the key. The data image coding module is used to print the production date image generated on the day onto the product. The image recognition and decoding traceability module is used to parse the production date image, obtain the hidden key, obtain the hidden key information after parsing the hidden key, and match the hidden key information with the anti-counterfeiting serial number database to trace product information and verify authenticity.

7. A data platform system for an enterprise production center, characterized in that, The enterprise production center data platform system includes the anti-counterfeiting traceability system based on font steganography encoding as described in claim 6.

Images