An anti-counterfeiting authentication method based on supervised guided segmentation statistical difference

Through the combination of supervised and guided segmentation algorithm and random texture patterns, the problem of forged anti-counterfeiting pattern recognition is solved, and the rapid and accurate authenticity and anti-counterfeiting ability is achieved, which is suitable for the public to easily identify, avoiding the equipment dependence and material costs of traditional methods.

CN113887350BActive Publication Date: 2024-11-15WUHAN INST OF TECH
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202111101318.5
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2021-09-18
Publication Date
2024-11-15
Estimated Expiration
2041-09-18

AI Technical Summary

Technical Problem

The prior art is difficult to quickly and accurately identify forged anti-counterfeiting patterns, and traditional methods require professional equipment, which cannot meet the public's convenient identification needs, and most anti-counterfeiting methods fail after being forged.

Method used

The supervised and guided segmentation algorithm is used to randomly generate anti-counterfeiting patterns with random fine textures, bind them to the identity code, upload them to the cloud database, use mobile devices to collect images and correct them through correction algorithms, and locate the difference area to identify the authenticity.

Benefits of technology

It realizes the rapid and accurate identification of the authenticity of anti-counterfeiting images without adding safety characteristics, resists counterfeiting, has anti-counterfeiting capabilities, and is easy to collect equipment, and is suitable for public use.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN113887350B_ABST
    Figure CN113887350B_ABST
Patent Text Reader

Abstract

The present invention provides an anti-counterfeiting authentication method based on supervised guided segmentation statistical difference, by generating a digital image with a specific fine texture pattern, generating an anti-counterfeiting label based on the intrinsic difference characteristics of ink randomness, using a supervised guided segmentation algorithm to locate the difference area of ​​the reserved sample anti-counterfeiting image, comparing the difference area of ​​the reserved sample anti-counterfeiting image and the anti-counterfeiting image to be tested, correcting and extracting the detail features of the anti-counterfeiting image, and realizing the function of identifying the authenticity of the anti-counterfeiting image to be tested. The present invention can resist counterfeiting by secondary printing and printing, has the ability to identify authenticity and anti-counterfeiting, does not require any additional means, has high counterfeiting cost, good security, and is particularly useful for a single corresponding single sample. The present invention can be directly detected by a mobile phone, is convenient to collect, has mobility, is simple for users to use, etc., and has broad application prospects.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The invention belongs to the technical field of image processing, and in particular relates to an anti-counterfeiting authentication method based on supervised guided segmentation statistical differences. Background Art

[0002] Counterfeit and substandard products have long been a serious threat to the world economy and consumers. With the advent of low-cost scanning equipment, high-quality printers, and better color copiers, it has become easier to forge anti-counterfeiting patterns. Therefore, the authenticity verification of anti-counterfeiting patterns is an increasingly important task. In order to add additional security features, the anti-counterfeiting method often used in practice is to add special materials to synthesize anti-counterfeiting patterns. Although this method has good anti-counterfeiting properties, they are costly and have poor interactivity, and consumers need to learn how to verify and distinguish. Therefore, a meaningful research area is to use the printing characteristics of anti-counterfeiting patterns for anti-counterfeiting without adding additional security feature materials. Since the counterfeit anti-counterfeiting patterns are printed with different printers, the printing effects will be different, especially at the edges of geometric patterns. Even if the same printer is used to print twice, there will be differences, just like "there are no two leaves in the world that are exactly the same." These inherent differences in random inks do not require any external means, especially for single (one-to-one) correspondence verification methods, which are very useful.

[0003] However, robust extraction of useful difference features for verification under various interference conditions is a difficult point in current research. At the same time, in order to ensure high accuracy, most of the time, professional acquisition equipment is used to perform authenticity detection, which cannot meet the public's demand for convenient anti-counterfeiting identification. With the rapid development of smart phones, it is becoming easier and easier to obtain digital images. Using digital image technology to identify the authenticity of patterns has become a quick and convenient method. However, for the images of anti-counterfeiting patterns, robust extraction of useful features, removal of a large amount of redundant information in the image, and focusing on the geometric morphological feature information of important key areas of the anti-counterfeiting pattern make the application of image technology in the field of printing anti-counterfeiting more challenging.

[0004] Currently, most anti-counterfeiting methods either use special materials or require a large number of sample training, and they become completely ineffective once they are forged. Summary of the invention

[0005] The technical problem to be solved by the present invention is to provide an anti-counterfeiting authentication method based on supervised guided segmentation statistical difference, which is used to accurately and quickly identify the authenticity of an anti-counterfeiting image to be tested.

[0006] The technical solution adopted by the present invention to solve the above technical problems is: an anti-counterfeiting authentication method based on supervised guided segmentation statistical differences, comprising the following steps:

[0007] S1: Randomly generate a unique anti-counterfeiting pattern with random fine textures, bind the anti-counterfeiting pattern and the attached identity code to the unique product, and upload it to the cloud database for storage as a sample image;

[0008] S2: Collect the anti-counterfeiting pattern on the product through the mobile device and upload it to the cloud identification center as the image to be tested, and use the correction algorithm to correct the image to be tested;

[0009] S3: Use the supervised guided segmentation algorithm to locate the difference area of ​​the sample image; use the binary image of the sample image as the standard supervision model, and perform supervised guided segmentation on the binary image of the image to be tested to correct the local threshold;

[0010] S4: comparing the difference area between the sample image and the image to be tested to identify the authenticity of the anti-counterfeiting image to be tested.

[0011] According to the above scheme, in step S1, the specific steps are:

[0012] S11: Generate a binary digital image with a specific subtle texture pattern as an anti-counterfeiting pattern by using a random generation function and a fractal interpolation method;

[0013] S12: Based on the randomness of ink flow, a printer is used to print an anti-counterfeiting pattern to generate a unique anti-counterfeiting label, which is bound to a unique product;

[0014] S13: Upload the digital image of the anti-counterfeiting label to the cloud and store it as a sample image.

[0015] According to the above scheme, in step S1, the specific steps are: the anti-counterfeiting pattern is a strip-like pattern composed of arbitrarily curved thin lines or dot blocks.

[0016] According to the above scheme, in step S2, the specific steps are:

[0017] S21: Collect the anti-counterfeiting pattern on the product through the mobile device and upload it to the cloud identification center as the image to be tested; S22: Search the cloud database for a sample image that is consistent with the identity code of the image to be tested;

[0018] S23: Correcting the image to be tested by using a template matching method according to the sample image to ensure that the size, dimension and position of the image to be tested and the sample image are completely matched.

[0019] According to the above scheme, in step S3, the specific steps are:

[0020] S31: Map the binary image SB (x, y) of the sample image to the binary image TB (x, y) of the image to be tested to find the difference area D:

[0021] SB(x,y)-TB(x,y),

[0022] If |SB(x,y)-TB(x,y)|≠0, then D(x,y) is 1; if |SB(x,y)-TB(x,y)|=0, then D(x,y) is 0;

[0023] S32: According to each pixel point with a value of 1 in the difference area D, a window area S with a size of d×d is taken on the corresponding sample image and the image to be tested. w and region T w , let the sample image be S(i,j), the corrected image to be tested be T(i,j), the window size be d, and the calculation area S w The average gray value of:

[0024]

[0025] Area S w The standard deviation of

[0026]

[0027] Area T w The average gray value of:

[0028]

[0029] Area T w The standard deviation of

[0030]

[0031] Let k be the correction parameter, and construct the region S according to the local mean and local standard deviation. w The weighted local mean deviation U Sw (x,y) and region T w The weighted local mean deviation U Tw (x,y):

[0032]

[0033]

[0034] S33: The area S of the sample image is w The weighted local mean deviation U Sw (x, y) is used as the reference model. If the region T of the image to be tested is w The weighted local mean deviation U Tw (x,y) and U Sw If (x,y) have the same sign, the grayscale distribution is consistent and the binary effect is consistent:

[0035] TB(x,y)=SB(x,y);

[0036] If the area T of the image to be tested is w The weighted local mean deviation U Tw (x,y) and U Sw If (x,y) have different signs, the grayscale distribution is inconsistent and the binary effect is opposite:

[0037] TB(x,y)=TB(x,y);

[0038] S34: Update the entire binary image TB(x, y) of the image to be tested and the binary image SB(x, y) of the sample image.

[0039] Furthermore, in step S4, the specific steps are:

[0040] The similarity between the binary image of the sample image and the binary image of the image to be tested is calculated by the normalized cross-correlation similarity formula; if the similarity is greater than or equal to 90%, the image to be tested is judged to be authentic; if the similarity is less than 90%, the image to be tested is judged to be counterfeit.

[0041] A computer storage medium stores a computer program executable by a computer processor, wherein the computer program executes an anti-counterfeiting authentication method based on supervised guided segmentation statistical differences.

[0042] The beneficial effects of the present invention are:

[0043] 1. The present invention discloses an anti-counterfeiting authentication method based on supervised guided segmentation statistical differences, which generates a digital image with a specific subtle texture pattern, generates an anti-counterfeiting label based on the intrinsic difference characteristics of ink randomness, uses a supervised guided segmentation algorithm to locate the difference area of ​​the reserved sample anti-counterfeiting image, compares the difference area of ​​the reserved sample anti-counterfeiting image and the anti-counterfeiting image to be tested, corrects and extracts the detail features of the anti-counterfeiting image, resists the loss of anti-counterfeiting information caused by scanning and photographing, restores the unique feature information of anti-counterfeiting to the greatest extent, and realizes the function of identifying the authenticity of the anti-counterfeiting image to be tested.

[0044] 2. The present invention utilizes the randomness of the printing ink flow to achieve the unique random characteristics of the anti-counterfeiting pattern: it can resist secondary printing and printing counterfeiting, and has not only the ability to identify authenticity, but also the ability to resist counterfeiting; it does not require any additional means, has a high counterfeiting cost, and has good security, which is particularly useful for single corresponding single samples.

[0045] 3. The present invention can be directly detected through a mobile phone, and has many advantages that traditional anti-counterfeiting identification technologies do not have, such as convenient equipment collection, portability, and simple user use, etc., and has broad application prospects. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] Figure 1 is a flow chart of an embodiment of the present invention.

[0047] Figure 2 It is an anti-counterfeiting pattern with random fine textures according to an embodiment of the present invention.

[0048] Figure 3 4 is a diagram showing the extraction effect of the difference area image according to an embodiment of the present invention.

[0049] Figure 4 It is a window image extraction effect diagram of the difference area between the sample image and the image to be tested in an embodiment of the present invention.

[0050] Figure 5 This is a binarization effect diagram after the supervised guided segmentation processing according to an embodiment of the present invention. DETAILED DESCRIPTION

[0051] The present invention will be further described in detail below in conjunction with the accompanying drawings and specific embodiments.

[0052] The technical solution of the present invention is to first randomly generate a digital image with a specific subtle texture pattern, use a printer to generate an anti-counterfeiting label based on the randomness of ink flow, and collect the digital image of the anti-counterfeiting label to the cloud database as a sample. The customer uses a mobile phone to take a photo of the anti-counterfeiting label posted on the product and upload it to the cloud identification center. The anti-counterfeiting pattern in the image to be tested is extracted using a correction algorithm, and the difference area of ​​the reserved sample anti-counterfeiting image is located using a supervised guided segmentation algorithm. The difference area of ​​the reserved sample anti-counterfeiting image and the anti-counterfeiting image to be tested is compared to identify the authenticity of the anti-counterfeiting image to be tested.

[0053] See also Figure 1 The present invention provides an anti-counterfeiting authentication method based on supervised guided segmentation statistical difference, comprising the following steps:

[0054] (1) Generation of anti-counterfeiting images:

[0055] A binary anti-counterfeiting pattern with random fine texture is generated by randomly generating a function and combining it with a fractal interpolation method and adding an identity code, such as Figure 2 As shown. The pattern is a strip-like pattern composed of randomly curved thin lines or blocks. At the same time, each anti-counterfeiting pattern generated by the random generation function is different and unique, which maximizes the counterfeiting cost. Then it is printed out by an ink printer, and a printing pattern with anti-counterfeiting uniqueness is generated based on the randomness of the ink flow. Finally, the image of the anti-counterfeiting pattern is collected by taking a photo and uploaded to the cloud database as a sample.

[0056] (2) Anti-counterfeiting identification process

[0057] For the printed anti-counterfeiting patterns circulating in the market, use a mobile phone with a camera function to take a photo of it, and upload the collected image to the cloud identification center, which uses the anti-counterfeiting recognition module to identify it:

[0058] First, search for a sample image that is consistent with the identity code of the suspected image to be tested. Use the template matching method to correct the suspected image to be tested after taking a photo based on the sample image obtained in step (1) to ensure that the size, dimension and position of the anti-counterfeiting image to be tested and the sample image are completely aligned. Use the binary image of the sample image as the standard supervision model to perform supervised guided segmentation on the binary image of the suspected anti-counterfeiting pattern to be tested to correct the local threshold. First, directly map the binary image of the sample image to the binary image of the suspected image to be tested, i.e., SB(x, y)-TB(x, y), and find the area D with differences, such as Figure 3 As shown. If |SB(x, y)-TB(x, y)|≠0, then D(x, y) is 1. If |SB(x, y)-TB(x, y)|=0, then D(x, y) is 0. Then, for each pixel with a value of 1 in the difference area D, a window area S of size d×d is taken on the corresponding sample image and the image to be tested. w and T w ,like Figure 4 As shown, the area S is then calculated w The average gray value of:

[0059]

[0060] Area S w The standard deviation of:

[0061]

[0062] Area T w The average gray value of:

[0063]

[0064] Area T w The standard deviation of:

[0065]

[0066] Where S(i,j) is the sample image, T(i,j) is the filtered image to be tested, and d is the window size. Due to differences in lighting and photography, there are obvious differences in brightness and contrast. However, if it is the same anti-counterfeiting pattern, the grayscale distribution of its image has a certain similarity. Therefore, the local mean deviation (i.e. pixel value minus mean) can be used to solve the problem of inconsistent brightness. The difference in contrast will affect the volatility of the local mean deviation, so the weighted local mean deviation can be used to replace the local mean deviation. The weighted local mean deviation is constructed based on the local mean and local standard deviation:

[0067]

[0068]

[0069] Where k is the correction parameter (usually 1 or 2). w The weighted local mean deviation U Sw (x, y) is used as the reference model. When the image sub-region T w The weighted local mean deviation U Tw (x,y) and U Sw If (x, y) have the same sign, the grayscale distribution is consistent, and the binary effect should be consistent, that is, TB(x, y) = SB(x, y); otherwise, the binary effect should be the opposite, TB(x, y) = TB(x, y). The operation flow of the supervised guided segmentation algorithm is as follows Figure 5 Finally, the updated images SB (x, y) and TB (x, y) are used to calculate the similarity between the sample binary image and the suspected binary image to be tested through the normalized cross-correlation similarity formula. Only when the similarity reaches 90%, the anti-counterfeiting pattern to be tested is considered to be authentic, otherwise it is counterfeit.

[0070] The above embodiments are only used to illustrate the design ideas and features of the present invention, and their purpose is to enable those skilled in the art to understand the content of the present invention and implement it accordingly. The protection scope of the present invention is not limited to the above embodiments. Therefore, any equivalent changes or modifications made based on the principles and design ideas disclosed by the present invention are within the protection scope of the present invention.

Claims

1. An anti-counterfeiting authentication method based on supervised segmentation statistical differences, characterized by: The following steps are involved: S1: Randomly generate a unique anti-counterfeiting pattern with random fine textures, bind the anti-counterfeiting pattern and the attached identity code to the unique product, and upload it to the cloud database for storage as a sample image; S2: Capture the anti-counterfeiting pattern on the product through a mobile device and upload it to the cloud authentication center as the image to be tested, and use the correction algorithm to correct the image to be tested; S3: Use the supervised guided segmentation algorithm to locate the difference area of ​​the sample image; use the binary image of the sample image as the standard supervision model, and perform supervised guided segmentation on the binary image of the test image to correct the local threshold; the specific steps are: S31: Map the binary image SB(x,y) of the sample image to the binary image TB(x,y) of the image to be tested and find the difference area D: SB(x,y)-TB(x,y), If |SB(x,y)-TB(x,y)|≠0, then D(x,y) is 1; if |SB(x,y)-TB(x,y)|=0, then D(x,y) is 0; S32: Take a d×d window area S on the corresponding sample image and the image to be tested according to each pixel point with a value of 1 in the difference area D. w and region T w , let the sample image be S(i,j), the corrected image to be tested be T(i,j), the window size be d, and the calculation area S w The average gray value of: Area S w The standard deviation of: Area T w The average gray value of: Area T w The standard deviation of: Let k be the correction parameter, and construct the region S according to the local mean and local standard deviation. w The weighted local mean deviation U Sw (x,y) = S(region x,y) - T w μ S of w (add x,y weight)*local σ Sw (average x,ky value) + deviation 125 difference 6 + Uk Tw (x,y):= S33: The area S of the sample image is w The weighted local mean deviation U Sw (x,y) is used as the reference model. If the region T of the image to be measured is w The weighted local mean deviation U Tw (x,y) and U Sw If (x,y) have the same sign, the grayscale distribution is consistent and the binary effect is consistent: TB(x,y)=SB(x,y); If the area T of the image to be measured w The weighted local mean deviation U Tw (x,y) and U Sw If (x,y) have different signs, the grayscale distribution is inconsistent and the binary effect is opposite: TB(x,y)=TB(x,y); S34: Update the entire binary image TB(x, y) of the image to be tested and the binary image SB(x, y) of the sample image; S4: comparing the difference areas between the sample image and the image to be tested to identify the authenticity of the anti-counterfeiting image to be tested.

2. The anti-counterfeiting authentication method based on supervised segmentation statistical differences according to claim 1, characterized in that: In the step S1, the specific steps are: S11: Generate a binary digital image with a specific subtle texture pattern as an anti-counterfeiting pattern through a random generation function and fractal interpolation method; S12: Based on the randomness of ink flow, a printer is used to print an anti-counterfeiting pattern to generate a unique anti-counterfeiting label, which is then bound to a unique product; S13: Upload the digital image of the anti-counterfeiting label to the cloud and store it as a sample image.

3. The anti-counterfeiting authentication method based on supervised segmentation statistical differences according to claim 1, characterized in that: In the step S1, the specific steps are as follows: the anti-counterfeiting pattern is a strip-like pattern composed of arbitrarily curved thin lines or dot blocks.

4. The anti-counterfeiting authentication method based on supervised segmentation statistical differences according to claim 1, characterized in that: In the step S2, the specific steps are: S21: Capture the anti-counterfeiting pattern on the product through a mobile device and upload it to the cloud authentication center as the image to be tested; S22: Searching the cloud database for a sample image that is consistent with the identity code of the image to be tested; S23: Correcting the image to be tested by using a template matching method according to the sample image to ensure that the size, dimension and position of the image to be tested and the sample image are completely matched.

5. The anti-counterfeiting authentication method based on supervised segmentation statistical differences according to claim 1, characterized in that: In the step S4, the specific steps are: The similarity between the binary image of the sample image and the binary image of the image to be tested is calculated using the normalized cross-correlation similarity formula; if the similarity is greater than or equal to 90%, the image to be tested is judged to be authentic; if the similarity is less than 90%, the image to be tested is judged to be counterfeit.

6. A computer storage medium, characterized in that: A computer program executable by a computer processor is stored therein, and the computer program executes the anti-counterfeiting authentication method based on supervisory guided segmentation statistical difference as claimed in any one of claims 1 to 5.