An intelligent information hiding and visualization method and system for entity commodity anti-counterfeiting

Abstract

The application discloses an intelligent information hiding and visualization method and system for entity anti-counterfeiting, and belongs to the fields of image processing and information security. A multi-objective optimization function is used to balance hiding capacity, visual quality and robustness, and an embedding feasible region insensitive to human eyes is selected in combination with region detection. Anti-counterfeiting information is encoded into an energy-constrained invisible matrix, and is embedded into an image space to generate a carrier containing the information. An image is collected by using a smart phone, an information matrix is extracted through edge detection, and then, visualized information is reconstructed through a decoding library, and authenticity verification is completed by using cosine similarity, Euclidean distance or SSIM. The system is composed of an embedding end module and a verification end module, and can run on a mobile phone APP / mini program without the need of special equipment. The method has the advantages of high concealment, convenient verification, high robustness, reliable anti-counterfeiting, wide application range, and can be applied to the anti-counterfeiting of printed matters such as commodities, certificates and tickets, and can balance security and user experience.

Description

Technical Field

[0001] This invention relates to the fields of image processing and information security technology, and to intelligent information hiding and visualization methods and systems for anti-counterfeiting images, specifically to an intelligent information hiding and visualization method and system for anti-counterfeiting of physical goods. Background Technology

[0002] In the field of product anti-counterfeiting, technologies such as holographic labels, fluorescent inks, and QR codes have been widely used and have played a certain role in combating counterfeit and substandard products. However, with the continuous improvement of counterfeiting technology and consumers' increasing demands for convenient verification, the shortcomings of the above-mentioned existing technologies have become increasingly prominent. Holographic labels and fluorescent inks rely on special physical or chemical characteristics to achieve anti-counterfeiting, making them easy to be counterfeited with high precision; moreover, verification requires specialized equipment such as ultraviolet lamps, which cannot meet the needs of ordinary consumers for instant and convenient verification. Although QR codes can be verified by scanning with a mobile phone without the need for specialized equipment, their patterns are obvious, damaging the product's appearance design. At the same time, because the information is public and statically readable, they are easily copied and tampered with in batches, making it impossible to achieve reliable authentication.

[0003] Information hiding technology can embed anti-counterfeiting data into a carrier image, achieving covert anti-counterfeiting without significantly altering the image's appearance, making it an important research direction. However, existing information hiding methods have significant shortcomings in practical applications: First, it is difficult to balance concealment and robustness. If the embedding strength is too low, the information is easily lost during printing, scanning, compression, or daily use; if the embedding strength is too high, it easily causes image distortion, exposing the hidden location. Second, the convenience of verification is insufficient. Most methods require the original image to be involved when extracting information, or rely on complex algorithms and professional software. Ordinary users cannot verify quickly and independently, making it difficult to achieve popular and lightweight anti-counterfeiting verification.

[0004] Therefore, there is an urgent need for an anti-counterfeiting method that addresses the problems of traditional anti-counterfeiting methods, such as easy counterfeiting, the need for specialized equipment, and damage to the appearance. Summary of the Invention

[0005] To address the shortcomings of the existing technologies, the present invention aims to provide an intelligent information hiding and visualization method and system for anti-counterfeiting of physical goods, which combines high concealment, high robustness, and high convenience. It can embed anti-counterfeiting information into the visual carrier of the product in a form invisible to the human eye, without damaging the appearance design, and can complete information extraction and visual authenticity verification with one click through common terminals such as smartphones. This will build a new type of anti-counterfeiting system that is safe, concealed, and widely applicable, providing effective technical support for brand protection and consumer rights protection.

[0006] This invention is achieved through the following technical solution: A smart information hiding and visualization method for anti-counterfeiting of physical goods includes the following steps: Step 1: Obtain the original carrier image and the hidden information to be embedded, construct a multi-objective optimization function, perform region detection on the original carrier image, and determine the feasible region for information embedding; Step 2: Encode the hidden information into an invisible information matrix within the feasible region, embed it into the spatial domain coefficients of the carrier image, and generate a dense carrier image; Step 3: At the verification end, a hidden information matrix is ​​extracted by using a handheld terminal to collect images of the dense carrier and edge detection. Step 4: Input the information matrix into the pre-built decoding library for reconstruction, output visualized information, and verify the completeness and authenticity by comparing the loss function.

[0007] The present invention also has the following technical features: Preferably, the multi-objective optimization function described in step one simultaneously considers the hiding capacity, the visual quality of the carrier image, and robustness, and determines the optimal embedding parameters by weighted summation; The region detection is based on edge distribution or saliency detection, marking regions with rich textures and insensitivity to the human eye as feasible regions.

[0008] Preferably, the process of encoding the hidden information to be embedded into an invisible information matrix within the feasible domain in step two includes learning the mapping encoding in stages to match the carrier features, and constraining the amplitude to be lower than the human visual threshold through energy constraints. The process of embedding spatial coefficients of the carrier image involves using an additive embedding rule to directly modify the grayscale value or color component of the carrier image pixels in the spatial domain, embedding the information coefficients into the feasible domain that allows modification.

[0009] Preferably, the edge detection in step three uses Canny, Sobel, or an edge detection network to extract subtle edge changes caused by embedded information in order to locate and recover the information matrix.

[0010] Preferably, the decoding library in step four is a pre-generated overlay; the contrast loss function includes cosine similarity, Euclidean distance or structural similarity index, used to measure the consistency between the reconstructed information and the original information, and to determine the authenticity according to a threshold.

[0011] This invention also protects an intelligent information hiding and visualization system for anti-counterfeiting of physical goods, comprising: The embedded module is used to perform steps one and two above, generate a dense carrier image and output it to a printing or display medium; The verification module is used to perform steps three and four above, acquiring images through a handheld terminal and completing information extraction and authenticity determination.

[0012] Preferably, the handheld terminal is a smartphone or tablet computer, and the verification module runs as an application, requiring no dedicated hardware device.

[0013] Compared with the prior art, the present invention has the following beneficial effects: This invention embeds information into areas insensitive to human vision through multi-objective optimization and region detection, and makes the embedding traces completely invisible through energy confinement, without damaging the original appearance design of the product, thus providing strong concealment. This invention can complete the entire process of image capture, information extraction, visual development, and authenticity determination using only a smartphone. It does not require special equipment such as ultraviolet lamps, making verification convenient. Ordinary consumers can operate it with one click, making it highly accessible. This invention uses a spatial domain embedding method combined with edge detection technology, which can effectively resist various interferences such as printing, scanning, image compression and minor daily wear and tear. It has excellent robustness and ensures that anti-counterfeiting information is not easily lost and is stably retained. This invention uses a contrastive loss function to verify information consistency, which can accurately identify copying and tampering behavior. Its anti-counterfeiting security is far superior to traditional technologies such as QR codes and fluorescent inks, and its anti-counterfeiting is reliable. This invention has wide adaptability and can flexibly embed various types of information such as pictures, text, and logos. It has a large information capacity and can be widely used in various physical anti-counterfeiting scenarios such as commodities, certificates, and tickets. Its application scope covers a wide range of fields. Attached Figure Description

[0014] Figure 1 The extracted image is from Embodiment 1 of the invention; Figure 2 This is the display effect of Embodiment 1 of the invention. Detailed Implementation

[0015] The present invention will be further described in detail below with reference to specific embodiments. These descriptions are for explanation purposes only and are not intended to limit the scope of the invention.

[0016] This invention provides an intelligent information hiding and visualization method for anti-counterfeiting of physical goods, comprising the following steps: Step 1: Obtain the original carrier image and the hidden information to be embedded, construct a multi-objective optimization function, perform region detection on the original carrier image, and determine the feasible region for information embedding; the multi-objective optimization function considers the hiding capacity, visual quality of the carrier image, and robustness at the same time, and sets three objectives: peak signal-to-noise ratio (PSNR) ≥ 40dB (to measure visual quality), hiding capacity ≥ 4kb, and high robustness; The formula for calculating the multi-objective optimization function is as follows: Where F is the original carrier image; For embedded parameters; To hide the capacity; Visual quality is typically measured using peak signal-to-noise ratio (PSNR). For robustness; For the weighting coefficients, satisfying .

[0017] The optimal embedding parameters are determined by weighted summation; Region detection is based on edge distribution or saliency detection, marking regions with rich textures and insensitive human vision (regions with saliency values ​​below the threshold of 0.3) as feasible regions.

[0018] Step 2: Encode the hidden information into an invisible information matrix within the feasible region, embed it into the spatial coefficients of the carrier image, and generate a dense carrier image. Scale each element in this matrix so that its absolute value does not exceed 5 pixels. Then, embed it into the feasible region using an additive embedding rule: new pixel value = original pixel value + corresponding element (operations are performed on all three RGB channels). After modifying all selected pixels, the dense carrier image is obtained and then printed.

[0019] The process of encoding the hidden information to be embedded into an invisible information matrix within the feasible domain includes learning the mapping encoding in stages to match the carrier features, and constraining the amplitude to be lower than the human visual threshold through energy constraints. The process of embedding spatial coefficients of the carrier image is to use an additive embedding rule to directly modify the grayscale value or color component of the carrier image pixels in the spatial domain, and embed the information coefficients into the feasible domain that allows modification.

[0020] The information matrix representation is as follows: Where v is the information matrix; For matrix transformation, the parameters are ; m represents the original hidden information; The matrix dimension (e.g., 256); The energy constraint is expressed as: in, The energy threshold is used to control the embedding strength to ensure it is invisible to the human eye; The segmented learning encoding process is represented as follows: in, For phased mapping encoding functions, The hidden image after preprocessing. Features of the carrier image; The human visual threshold is expressed as: in, The human visual threshold; Spatial pixel embedding is represented as: in, For the original image in Pixel value at channel c; It is a carrier of information; For global embedding strength factor; Let k be the k-th component of the information matrix.

[0021] Step 3: At the verification end, a hidden carrier image is acquired through a handheld terminal, and the hidden information matrix is ​​extracted using edge detection. Edge detection uses Canny, Sobel, or an edge detection network to extract subtle edge changes caused by the embedded information in order to locate and recover the information matrix.

[0022] Performing Canny edge detection is represented as follows: Where EdgeMap is the output edge map; For edge detection algorithms, the common Canny gradient operator can be used, or other operators can be used; The minimum error reconstruction information matrix is ​​represented as: in, The estimated information matrix represents the hidden information recovered from the dense image. The extraction function has the following parameters: ; This is an information embedding function.

[0023] Step 4: Input the information matrix into the pre-built decoding library for reconstruction, output visualized information, and verify the completeness and authenticity by comparing loss functions. The decoding library is a pre-generated overlay; the comparison loss functions include cosine similarity, Euclidean distance, or structural similarity index, used to measure the consistency between the reconstructed information and the original information, and determine authenticity based on a threshold.

[0024] The visual decoding process is represented as follows: in, For the reconstruction of visual information; For decoding library, the parameters are: ; Cosine similarity loss; The loss is due to Euclidean distance. For structural similarity index loss; Example 1 The original hidden information is a Lena image (1024×1024 pixels, RGB color image). The original carrier image is a decorative background pattern on the product packaging (1024×1024 pixels, RGB color image).

[0025] Step 1: Construct the corresponding multi-objective optimization function based on the shading map of the Lena diagram, and find the feasible region of the shading for embedding preparation.

[0026] Step 2: Generate an information matrix from the Lena diagram in Step 1, and enlarge each element of this matrix by 3 pixels. Then, apply the bonus embedding rule to embed it into the feasible region. After embedding, proceed with printing, as shown below. Figure 1 As shown.

[0027] Step 3: Obtain the printed image of the dense carrier. Scan the image using a WeChat mini-program. The mini-program first automatically corrects the shooting angle and lighting, then performs Canny edge detection on the image to obtain an edge map. By analyzing the edge map, the corresponding dense information matrix is ​​obtained.

[0028] Step 4: Input the obtained cryptic information matrix into the decoding library for matrix decoding, and display the corresponding Lena diagram on the mini-program, such as... Figure 2 As shown, the displayed Lena diagram is compared with the original Lena diagram to calculate its structural similarity index (SSIM). If the SSIM is greater than 0.9, the WeChat mini program displays "Genuine Product" and is accompanied by a prompt sound; otherwise, it displays "Suspicious Counterfeit".

[0029] Example 2 The original hidden information is a school emblem image (512×512 pixels, RGB color image). The original carrier image is a decorative background pattern on product packaging (512×512 pixels, RGB color image).

[0030] Step 1: Construct a corresponding multi-objective optimization function based on the background pattern of the school emblem, and find the feasible region of the background pattern for embedding preparation.

[0031] Step 2: Generate an information matrix from the school emblem image in Step 1, enlarge each element of the matrix by 6 pixels, and then embed it into the feasible region using the bonus embedding rule. After embedding, proceed with printing.

[0032] Step 3: Use the corresponding WeChat mini-program to perform Sobel edge detection on the image to obtain an edge map. By analyzing the edge map, the corresponding dense information matrix is ​​obtained.

[0033] The image of the dense carrier is photographed and extracted to obtain the corresponding dense information matrix.

[0034] Step 4: Input the obtained encrypted information matrix into the decoding library for matrix decoding, and display the corresponding school emblem image on the mini-program. Compare the displayed school emblem image with the original school emblem image to calculate its structural similarity index (SSIM).

[0035] Example 3 The original hidden information is a trademark text image (2048×2048 pixels, RGB color image). The original carrier image is a decorative background pattern on the product packaging (2048×2048 pixels, RGB color image).

[0036] Step 1: Construct a corresponding multi-objective optimization function based on the background pattern of the trademark text image, and find the feasible domain of the background pattern for embedding preparation.

[0037] Step 2: Generate an information matrix from the trademark text image in Step 1, enlarge each element of the matrix by 1 pixel, and then embed it into the feasible region using the bonus embedding rule. After embedding, proceed with printing.

[0038] Step 3: Use the corresponding WeChat mini program and then use an edge detection network to perform edge detection on the image to obtain an edge map; by analyzing the edge map, obtain the corresponding dense information matrix.

[0039] Step 4: Input the obtained encrypted information matrix into the decoding library for matrix decoding, and display the corresponding trademark text image on the mini program. Compare the displayed trademark text image with the original trademark text image to calculate its structural similarity index (SSIM).

[0040] Example 4 An intelligent information hiding and visualization system for anti-counterfeiting of physical goods includes: an embedded module and a verification module; The embedded module is a PC workstation equipped with image processing software, and is configured with an image acquisition unit, a parameter optimization unit, an information encoding unit, and a spatial domain embedding unit; The verification module uses a smartphone or computer (with built-in camera and processor) to run an anti-counterfeiting verification mini-program / APP, which includes an image acquisition and correction unit, an edge detection unit, an information extraction unit, a decoding and reconstruction unit, and a authenticity discrimination unit.

[0041] The working process of the embedded module includes: Read the product packaging background image (512×512 RGB) and the brand logo information to be hidden; Run a multi-objective optimization function, set weights α=0.3, β=0.5, γ=0.2, and determine the optimal embedding parameter θ with PSNR≥40 dB, capacity≥4kb, and robustness≥0.95 as objectives; Saliency detection is performed on the carrier image, and regions with saliency < 0.3 are marked as embedding feasible regions; The brand identity is encoded as a 256-dimensional information matrix, and after L2 norm normalization and energy constraint processing, η=5; With an intensity factor of λ=1, the information matrix is ​​embedded point by point into the feasible region pixels to generate a dense carrier image; Output the dense image to the printing press to be printed as product packaging.

[0042] The working process of the verification module includes: Users can use their smartphones to open the anti-counterfeiting mini-program and take a picture of the secret image on the product packaging. The mini-program automatically completes distortion correction, illumination equalization, and image preprocessing. The Canny edge detection algorithm is called to extract weak edge features caused by information embedding, resulting in an edge map; Based on the minimum reconstruction error criterion, the estimated information matrix is ​​extracted from the edge map; Input the information matrix into the pre-built decoding library to reconstruct the brand logo visual pattern; Calculate the SSIM value of the original information and the reconstructed information. When SSIM > 0.9, it is determined to be genuine, and the interface displays a visual icon and prompts that the verification has passed; otherwise, it is determined to be suspicious and prompts that the verification has failed.

[0043] The embodiments described above are merely illustrative of several implementations of the present invention and should not be construed as limiting the scope of the present invention. Those skilled in the art can make various modifications and improvements without departing from the concept of the present invention, and these modifications and improvements all fall within the scope of protection of the present invention. Therefore, the scope of protection of this patent should be determined by the appended claims.

Claims

1. A method for intelligent information hiding and visualization in anti-counterfeiting of physical goods, characterized in that, Includes the following steps: Step 1: Obtain the original carrier image and the hidden information to be embedded, construct a multi-objective optimization function, perform region detection on the original carrier image, and determine the feasible region for information embedding; Step 2: Encode the hidden information to be embedded into an invisible information matrix within the feasible domain, embed the spatial domain coefficients of the carrier image, and generate a dense carrier image. Step 3: At the verification end, a hidden information matrix is ​​extracted by using a handheld terminal to collect images of the dense carrier and edge detection. Step 4: Input the information matrix into the pre-built decoding library for reconstruction, output visualized information, and verify the completeness and authenticity by comparing the loss function.

2. The intelligent information hiding and visualization method for anti-counterfeiting of physical goods according to claim 1, characterized in that, The multi-objective optimization function described in step one simultaneously considers the hiding capacity, the visual quality of the carrier image, and robustness, and determines the optimal embedding parameters through weighted summation; the region detection is based on edge distribution or saliency detection, and regions with rich texture and insensitivity to the human eye are marked as feasible regions.

3. The intelligent information hiding and visualization method for anti-counterfeiting of physical goods according to claim 1, characterized in that, The process of encoding the hidden information to be embedded into an invisible information matrix within the feasible domain in step two includes learning the mapping encoding in stages to match the carrier features, and constraining the amplitude to be lower than the human visual threshold through energy constraints. The process of embedding spatial coefficients of the carrier image involves using an additive embedding rule to directly modify the grayscale value or color component of the carrier image pixels in the spatial domain, embedding the information coefficients into the feasible domain that allows modification.

4. The intelligent information hiding and visualization method for anti-counterfeiting of physical goods according to claim 1, characterized in that, The edge detection described in step three uses Canny, Sobel, or an edge detection network to extract subtle edge changes caused by embedded information in order to locate and recover the information matrix.

5. The intelligent information hiding and visualization method for anti-counterfeiting of physical goods according to claim 1, characterized in that, The decoding library mentioned in step four is a pre-generated overlay; the contrast loss function includes cosine similarity, Euclidean distance or structural similarity index, which is used to measure the consistency between the reconstructed information and the original information, and to determine the authenticity according to a threshold.

6. An intelligent information hiding and visualization system for anti-counterfeiting of physical goods, characterized in that, include: An embedded module is used to perform steps one and two as described in any one of claims 1 to 3, generating a dense carrier image and outputting it to a printing or display medium; The verification module is used to perform steps three and four as described in any one of claims 1, 4, and 5, and to collect images and complete information extraction and authenticity determination through a handheld terminal.

7. The intelligent information hiding and visualization system for anti-counterfeiting of physical goods according to claim 6, characterized in that, The handheld terminal is a smartphone or tablet computer, and the verification module runs as an application, requiring no dedicated hardware device.

Images