Anti-counterfeit label capable of being verified off line based on dark watermark technology and manufacturing and identifying method of anti-counterfeit label

Abstract

The invention discloses an anti-counterfeit label capable of being verified off line based on a dark watermark technology and a manufacturing and identifying method thereof, and belongs to the field of information security. The anti-counterfeit label is low in manufacturing cost, only a small piece of non-woven fabric is needed, an ordinary printer is used for printing the anti-counterfeit pattern, networking is not needed for verification, the number of counterfeit distinguishing times is not limited, counterfeit distinguishing can be achieved only through a picture with the non-woven fabric and the anti-counterfeit pattern, and offline counterfeit distinguishing can be completed through an ordinary mobile phone with a camera. The content of the anti-counterfeiting pattern can be selected at will, a product trademark can be selected as the anti-counterfeiting pattern, and extra information such as production numbers and production places can be hidden in the pattern so as to assist product traceability.

Description

Technical Field

[0001] The invention belongs to the technical field of anti-counterfeiting labels, and relates to an anti-counterfeiting label that can be verified offline based on dark watermark technology and a method for making and identifying the label. Background Art

[0002] Anti-counterfeiting labels are widely used and can be used for the anti-counterfeiting of commodities and banknotes. However, existing anti-counterfeiting technologies have many defects. They either require high production costs to prevent illegal copying, or require online verification, resulting in cumbersome verification steps, and generally only one counterfeit can be identified. The present invention relates to an anti-counterfeiting label technology that can be verified offline based on dark watermark technology. The production cost is low. Only a small piece of non-woven fabric and an anti-counterfeiting pattern printed using an ordinary printer are required. There is no need to connect to the Internet for verification. The number of times of identifying counterfeits is unlimited. Only a photo with non-woven fabric and anti-counterfeiting pattern can be used to identify counterfeits using the software module of the present invention. For example, an ordinary mobile phone with a camera can complete offline identification. The content of the anti-counterfeiting pattern can be selected arbitrarily. For example, a product trademark can be selected as an anti-counterfeiting pattern. Additional information can be hidden in the pattern, such as production number, place of origin and other information, to assist product traceability. This anti-counterfeiting label technology can be used in all fields where existing anti-counterfeiting labels are applied, and there is no need to connect to the Internet. Summary of the invention

[0003] The present invention provides an anti-counterfeiting label based on dark watermark technology that can be verified offline and a method for making and identifying the label, which can efficiently identify counterfeits without the need for networking and only requires a photo of the anti-counterfeiting label.

[0004] The technical solution adopted by the present invention is as follows:

[0005] An offline verifiable anti-counterfeiting label based on dark watermark technology mainly consists of two parts: one is a piece of non-woven fabric with obvious pores and textures, and the other is an anti-counterfeiting icon. From the production process of non-woven fabrics, it can be known that the texture of each piece of non-woven fabric is unique, so the unique recognition of the object can be obtained through the particularity of the non-woven fabric. The anti-counterfeiting icon refers to a picture or trademark that can prove the association with the anti-counterfeiting label.

[0006] As a further feature, nonwoven fabrics can also be replaced by other materials with unique texture characteristics that are difficult to replicate.

[0007] The production and verification of the anti-counterfeiting label that can be verified offline includes the following steps:

[0008] The label making process is as follows:

[0009] Step 1-1: Obtain an image of the non-woven fabric by photographing or scanning, wherein the image contains texture features of the non-woven fabric.

[0010] Step 1-2: Obtain a feature vector of a preset dimension related to the content information of the non-woven fabric image. The feature vector is generally represented by a sequence of 0 and 1. The method for obtaining the feature vector can use the image encoding technology in deep learning, that is, the technology of encoding an image into a vector of a fixed dimension using a neural network, and the distance between the vectors encoded by similar images is also small.

[0011] As a further feature, the image encoding method may also use image hashing technology.

[0012] Further features, after obtaining the feature vector, an encryption process can be added, and the encryption can use a symmetric or asymmetric encryption algorithm.

[0013] Step 1-3: Use dark watermark technology to embed the feature vector of the non-woven fabric image obtained in step 1-2 into the image of the anti-counterfeiting icon, and obtain an image of the anti-counterfeiting icon with a watermark embedded with the texture features of the non-woven fabric. Dark watermark technology can use deep learning technology, that is, input the image and watermark information into a deep network. The deep network is divided into two parts: an encoding network and a decoding network. The encoding network fuses the image and the watermark information to generate an image with watermark information. The generated image is difficult to see the difference from the input image with the naked eye, and the decoding network can correctly decode the input watermark information from the watermarked image.

[0014] As a further feature, the watermark information embedded in the anti-counterfeiting trademark image can additionally include information about the item associated with the anti-counterfeiting label, which can be used to trace the source, production date, etc. of the product.

[0015] Step 1-4: The non-woven fabric and the image with watermark technology obtained in step 1-3 are combined into an anti-counterfeiting label.

[0016] The tag verification process is as follows:

[0017] Step 2-1: Observe whether the label is divided into two parts, non-woven fabric and anti-counterfeiting trademark. Identify the non-woven fabric part of the label to determine whether it is made of non-woven fabric. If not, it is considered fake. If it is made of non-woven fabric, scan or take a photo of the non-woven fabric to obtain an image with non-woven fabric texture characteristics.

[0018] Step 2-2: Encode the image obtained in step 2-1. This process is the same as steps 1-3 in the label making process, and obtain a feature vector related to the texture features of the non-woven fabric.

[0019] Step 2-3: Use the dark watermark decoding function to decode the image of the anti-counterfeiting trademark and extract the watermark information, which is a vector of a preset dimension.

[0020] The watermark information needs to be decrypted, and the key needs to use the key of the corresponding encryption process. Adding this encryption and decryption process can authenticate or control the user of the label authentication, and can also increase the difficulty of cracking the entire system.

[0021] Step 2-4: Compare the similarity of the vectors obtained in step 2-2 and step 2-3. If the similarity is greater than a preset threshold, the label is determined to be true, otherwise the label is determined to be false. The vector similarity measurement can be measured by cosine similarity or Hamming distance.

[0022] As a further feature, the extra information needs to be removed from the watermark information used for comparison.

[0023] The anti-counterfeiting label of the present invention has low production cost, only needs a small piece of non-woven fabric and uses an ordinary printer to print the anti-counterfeiting pattern, and verification does not need to be connected to the Internet. The number of times of identification is unlimited, and only a photo with the non-woven fabric and the anti-counterfeiting pattern is needed to achieve identification, for example, an ordinary mobile phone with a camera can complete offline identification. The content of the anti-counterfeiting pattern can be arbitrarily selected, for example, the product trademark can be selected as the anti-counterfeiting pattern, and additional information can be hidden in the pattern, such as production number, place of origin and other information, to assist product traceability. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] Figure 1 Flowchart of label production.

[0025] Figure 2 Flowchart of label authentication. DETAILED DESCRIPTION

[0026] The specific embodiments of the present invention are described in detail below in combination with the technical solutions and the accompanying drawings.

[0027] Example 1

[0028] As shown in the accompanying drawings, the main steps include:

[0029] 1-1. When a user wants to make an anti-counterfeiting label, first select a piece of non-woven fabric, scan it with a scanner, and obtain an image with the texture characteristics of the non-woven fabric. Then select a trademark image.

[0030] 1-2. Use image hashing technology to encode the non-woven fabric image to obtain a 0, 1 sequence corresponding to the texture features of the non-woven fabric.

[0031] 1-3. Use the private key in the asymmetric key to encrypt the 0, 1 sequence to obtain an encrypted 0, 1 sequence.

[0032] 1-4. The encrypted 0, 1 sequence is embedded into the image of the trademark selected by the user using the dark watermark function, and the watermarked image is printed to the user-specified location. The image and the non-woven fabric form an anti-counterfeiting label.

[0033] Label authentication process:

[0034] 1-5. The user makes a preliminary judgment. The label should be divided into two parts, non-woven fabric and trademark. If they do not match, it is considered fake. Otherwise, proceed to the following steps.

[0035] 1-6. Photograph or scan the nonwoven fabric to obtain an image with the texture characteristics of the nonwoven fabric.

[0036] 1-7. Decode the dark watermark of the trademark image to obtain an encrypted 0, 1 sequence, and then use the public key in the asymmetric key to decrypt it to obtain a 0, 1 sequence.

[0037] 1-8. Use the same method as step 1-2 to obtain the coded 0, 1 sequence.

[0038] 1-9. Compare the similarities of the 0, 1 sequences obtained in steps 1-7 and 1-8. If the similarity is greater than a preset threshold, the label is determined to be true, otherwise the label is determined to be false.

[0039] Embodiment 2:

[0040] 2-1. A user wants to make an anti-counterfeiting label for a product and use the product's trademark pattern as the anti-counterfeiting trademark.

[0041] A piece of nonwoven fabric is also selected and photographed to obtain an image thereof.

[0042] 2-2. Use the image hash function to obtain the hash value of the non-woven fabric image, which is usually a 0, 1 sequence, and then add a 0, 1 sequence encoding information such as the origin and production date of the product.

[0043] 2-3. Use the symmetric key to encrypt the 0, 1 sequence to obtain the encrypted 0, 1 sequence.

[0044] 2-4. The encrypted 0, 1 sequence is embedded into the anti-counterfeiting trademark image using the dark watermark function, and the watermarked image and non-woven fabric are combined into an anti-counterfeiting label.

[0045] Label authentication process:

[0046] 2-5. The user initially identifies the label and divides it into two parts: one is the trademark pattern and the other is the non-woven fabric. Scan or take a photo of the label to obtain the non-woven fabric and trademark images.

[0047] 2-6. Decode the dark watermark of the trademark image to obtain an encrypted 0, 1 sequence, and then use the symmetric key to decrypt it to obtain a 0, 1 sequence. According to the origin and production date information, the authenticity and quality of the product can be preliminarily judged or traced. Remove the part of the 0, 1 sequence containing the origin and production date information and continue to the next step.

[0048] 2-7. Use the same method as step 2-2 to get the hash value of the non-woven fabric image, which is a sequence of 0s and 1s without adding any additional information.

[0049] 2-8. Compare the similarities of the 0 and 1 sequences obtained in steps 2-6 and 2-7. If the similarity is greater than a preset threshold, the label is determined to be true, otherwise the label is determined to be false.

Claims

1. An anti-counterfeiting label based on dark watermark technology that can be verified offline, characterized in that: It mainly consists of two parts: one part is a non-woven fabric with obvious pores and texture, and the other part is an anti-counterfeiting icon; the texture of each non-woven fabric is unique, so that the uniqueness of the item can be recognized through the particularity of the non-woven fabric; the anti-counterfeiting icon refers to a picture or trademark that can prove its association with the anti-counterfeiting label.

2. The anti-counterfeiting label according to claim 1, characterized in that: Nonwovens are replaced by other materials that have unique texture characteristics that are difficult to replicate.

3. The method for making the anti-counterfeiting label according to claim 1 or 2, characterized by the following steps: Step 1-1: Obtain an image of the non-woven fabric by taking a photo or scanning, wherein the image contains texture features of the non-woven fabric; Step 1-2: Obtain a feature vector of a preset dimension related to the content information of the non-woven fabric image. The feature vector is generally represented by a sequence of 0 and 1. The method for obtaining the feature vector can use the image coding technology in deep learning, that is, the technology of encoding an image into a vector of a fixed dimension using a neural network, and the distance between vectors encoded by similar images is also small. Step 1-3: The feature vector of the non-woven fabric image obtained in step 1-2 is embedded into the image of the anti-counterfeiting icon using dark watermark technology to obtain an image of the anti-counterfeiting icon with a watermark embedded with the texture features of the non-woven fabric; the dark watermark technology can use deep learning technology, that is, the image and the watermark information are input into a deep network, and the deep network is divided into two parts: an encoding network and a decoding network. The encoding network fuses the image and the watermark information to generate an image with the watermark information. The difference between the generated image and the input image is difficult to see with the naked eye, and the decoding network can correctly decode the input watermark information from the watermarked image; Step 1-4: The non-woven fabric and the image with watermark technology obtained in step 1-3 are combined into an anti-counterfeiting label.

4. The method for making an anti-counterfeit label according to claim 3, characterized in that: In step 1-2, it is further characterized that the encoding method of the image uses image hashing technology.

5. The method for making an anti-counterfeit label according to claim 3, characterized in that: In step 1-2, after obtaining the feature vector, an encryption process is added, and the encryption uses a symmetric or asymmetric encryption algorithm.

6. The method for making an anti-counterfeit label according to claim 3, characterized in that: In step 1-3, the watermark information embedded in the anti-counterfeiting trademark image may be supplemented with information about the item associated with the anti-counterfeiting label, and the information may be used to trace the source and production date of the product.

7. The method for verifying the anti-counterfeit label according to claim 1 or 2, characterized by the following steps: Step 2-1: Observe whether the label is divided into two parts, non-woven fabric and anti-counterfeiting trademark. Identify the non-woven fabric part of the label to determine whether it is made of non-woven fabric. If not, it is considered fake. If it is made of non-woven fabric, scan or take a photo of the non-woven fabric to obtain an image with non-woven fabric texture features. Step 2-2: Encode the image obtained in step 2-1. This process is the same as steps 1-3 in the label making process, and obtain a feature vector related to the texture features of the non-woven fabric; Step 2-3: Use the dark watermark decoding function to decode the image of the anti-counterfeiting trademark and extract the watermark information, which is a vector of a preset dimension; The watermark information needs to be decrypted, and the key needs to use the key of the corresponding encryption process; adding this encryption and decryption process can authenticate or control the user who authenticates the label, and can also increase the difficulty of cracking the entire system; Step 2-4: Compare the similarity of the vectors obtained in step 2-2 and step 2-3. If the similarity is greater than the preset threshold, the label is determined to be true, otherwise the label is determined to be false; the vector similarity measurement can be measured by cosine similarity or Hamming distance.

8. The method for verifying an anti-counterfeit label according to claim 7, characterized in that: This extra information needs to be removed from the watermark information used for comparison.

Images