Method for preventing duplication of real-name registration two-dimensional code of unmanned aerial vehicle

Abstract

The invention discloses a method for preventing duplication of a real-name registration two-dimensional code of an unmanned aerial vehicle. The method comprises the following steps of 1, performing block processing on a two-dimensional code image through discrete cosine transform (DCT), improving the first-time printing resistance by adopting an intermediate-frequency coefficient energy enhancement technology, and encrypting watermark information in combination with Logistic chaotic mapping, and 2, adjusting symbol distribution of an intermediate-frequency coefficient selected in a DCT block by adopting a dynamic embedding strategy to realize watermark embedding. And 3, extracting a watermark through statistical coefficient symbol distribution during verification, setting a threshold value, and judging whether the two-dimensional code is copied or not by using a normalized correlation coefficient. The method can accurately distinguish behaviors of primary printing and secondary copying, realizes high sensitivity to copying attacks while ensuring readability of the two-dimensional code, has the advantages of low cost, low complexity, mobile terminal compatibility and the like, effectively solves the problem of identity counterfeiting in the scene of one-machine one-code of the unmanned aerial vehicle, and improves the identity counterfeiting efficiency of the unmanned aerial vehicle. And a reliable anti-counterfeiting technical scheme is provided for low-altitude safety supervision.

Description

TECHNICAL FIELD

[0001] The application belongs to the field of information security, and proposes a method for preventing copying of real-name registration two-dimensional code of unmanned aerial vehicle. The method realizes the robustness of the watermark to the first printing scanning and the sensitivity to the secondary copying by combining the intermediate frequency coefficient energy enhancement module, the Logistic encryption and the discrete cosine transform domain watermark embedding strategy, and provides a deployable anti-fake scheme for the security authentication of the real-name registration system of unmanned aerial vehicle. BACKGROUND

[0002] Under the vigorous rise of the low-altitude economy field, the regulatory framework of civil unmanned aerial vehicle is facing severe technological innovation requirements. To meet this challenge, the relevant authoritative management regulations have been officially promulgated and implemented, which clearly stipulate that the owner must complete the real-name registration process in the official designated integrated management system, and firmly attach the unique identifier containing identity information and two-dimensional code generated by the system to the unmanned aerial vehicle body. The core intention of this identification system is to achieve fine digital management of "one machine one code" through a unique coding system, and its function positioning is highly similar to the license plate system of ground transportation tools, aiming to build the core technical support for air space safety regulation.

[0003] However, the current widely used QR code scheme based on static plaintext information has inherent and fundamental security vulnerabilities due to its physical form that can be easily copied. Attackers can easily scan a legitimate two-dimensional code and then use high-resolution printing devices to accurately reproduce it, and then paste the fake identification on an unauthorized aircraft, creating a "fake license" or identity fraud regulatory blind area. This inherent structural defect limits the function of the existing identification system to basic identity information display, and this structural defect makes the existing system only achieve identity recognition function, and cannot meet the urgent need of the aviation safety field for "active anti-fake verification" capability, Figure 1 The figure shows the process of copying the real-name registration two-dimensional code of unmanned aerial vehicle by the attacker. The "Civil Unmanned Aerial Vehicle Operation Safety Management Regulations" requires users to paste their real-name registration two-dimensional code on the civil unmanned aerial vehicle, and regulatory personnel can obtain part of the information of the unmanned aerial vehicle and the user by scanning the real-name registration two-dimensional code. The attacker can print and paste the real-name registration two-dimensional code of other users on his own unmanned aerial vehicle after scanning it, so as to achieve the purpose of forging the identity of other users.

[0004] The root of this security risk can be analyzed from multiple technical dimensions. At the information-bearing medium level, the modular structure of the standard QR code only introduces predictable geometric distortions during the printing and scanning cycle, which can be easily repaired by conventional image correction techniques, thus achieving perfect reproduction of the encoding matrix. At the authentication logic level, existing systems rely entirely on one-way information reading mode, which lacks any embedded anti-counterfeiting features that can be verified, making counterfeit identification objects able to pass through the official authentication process without any obstacles. In terms of the applicability of the technical solution, existing solutions in the market are polarized: physical anti-counterfeiting means such as radio frequency identification tags have prohibitively high implementation and deployment costs; while advanced cryptographic solutions often require dedicated decoding hardware support, both of which are difficult to effectively balance the three key requirements of low cost, high reliability, and portable verification in the specific scenario of unmanned aerial vehicle mass application.

[0005] Therefore, there is an urgent need in the industry to develop an innovative digital watermarking technology that can cleverly implant an authentication signal with specific semi-fragile properties while maintaining the original reading function of the two-dimensional code. An ideal authentication signal should be able to stably tolerate minor signal distortion caused by legal operations such as initial printing and normal scanning, while showing high sensitivity to significant signal degradation caused by attack behaviors such as secondary copying or malicious reprinting and triggering verification failure. Although research on digital watermarking for two-dimensional code printing and scanning attacks has been ongoing, existing work is still in the initial exploration stage, and its research focus is mostly focused on improving the robustness of two-dimensional codes under physical damage, or developing fully fragile watermarks for content tamper detection, rather than the semi-fragile properties required by the application scenario.

[0006] The limitations of existing technical solutions are obvious: methods focusing on robustness overemphasize damage resistance, which weakens the anti-counterfeiting sensitivity to fine copying behavior; while fully fragile watermarks cannot effectively distinguish the subtle signal differences between legal operations and malicious imitation. More critically, current algorithms generally ignore the special constraints of the unmanned aerial vehicle regulation scenario, such as the need to strictly control the embedded data volume of the watermark within the error correction redundancy of the two-dimensional code itself, the need for seamless compatibility with general-purpose devices such as standard smartphone cameras during the verification process, and the need to control the response time of on-site authentication within the extremely short time of 500 milliseconds. The lack of these key parameters makes it difficult to directly transplant and apply existing solutions to the unmanned aerial vehicle identity authentication system.

[0007] In view of this, in the low-altitude security field, a new defense framework capable of effectively resisting physical replication attacks is urgently needed. The core of this framework is to develop a new type of digital watermark with precise controllable semi-fragile characteristics. By precisely setting the signal sensitivity threshold, that is, seeking the best balance point between the damage tolerance caused by the first printing and the distortion sensitivity caused by the second printing, the "readability and watermark effectiveness" dual factor verification of the identity authenticity of the unmanned aerial vehicle is finally achieved. In summary, the existing digital watermark anti-counterfeiting method for two-dimensional code carriers cannot meet the complex requirements of this specific application scenario. Therefore, research and implementation of a high-efficiency anti-replication technology dedicated to unmanned aerial vehicle real-name registration two-dimensional code not only has great theoretical innovation value, but also has urgent practical application significance. SUMMARY

[0008] The present application is directed to the security defects of the real-name registration two-dimensional code of civil unmanned aerial vehicles, and solves the following technical problems:

[0009] (1) Static two-dimensional code is easy to be physically replicated: the current "one machine one code" identification is based on plaintext two-dimensional code, and attackers can achieve identity forgery by scanning, recovering and printing, and the authenticity of the two-dimensional code cannot be verified.

[0010] (2) Limitations of existing anti-counterfeiting technologies: traditional physical tags are high in cost, and encryption algorithm verification is complex. Existing technologies focus on the robustness against compression / noise, and cannot balance the sensitivity to replication attacks.

[0011] (3) Lack of mature solutions for printing-scanning scenarios: two-dimensional codes need to resist noise from the first scanning and remain sensitive to tampering from the second replication after being printed and pasted. There is currently no mature solution.

[0012] The present application proposes a semi-fragile digital watermarking algorithm based on intermediate frequency DCT coefficient energy enhancement, including the following steps:

[0013] Stage 1: input stage, input the original two-dimensional code image to be embedded with a size of 256x256 and the watermark image with a size of 16x16.

[0014] Stage 2: processing stage, DCT domain statistical property analysis and enhancement.

[0015] (1) The two-dimensional code image is divided into 16x16 pixel sub-blocks for discrete cosine transform.

[0016] (2) Analysis of the disturbance law of printing / copying on DCT coefficients, it is found that the sign accuracy of medium frequency coefficients is significantly different between the first printing and copying, the sign accuracy of DC and low frequency coefficients in DCT coefficients is higher after QR code passes through the printing and copying process, there is a relatively obvious gap in the accuracy of printing and copying in the medium frequency position, and embedding steganographic information in DC and low frequency coefficients will cause obvious visual changes in the image, while the medium frequency changes little, so we can improve the energy of the coefficients with smaller absolute values in the medium frequency coefficients to improve the effect of the medium frequency coefficient sign on the anti-printing scanning attack.

[0017] (3) Medium frequency coefficient energy enhancement: for the a medium frequency coefficients with the smallest absolute value in each block, energy enhancement is performed according to formula (1), k is the enhancement strength, which improves the robustness to the first scanning noise, after the medium frequency coefficient energy enhancement, the accuracy of the first printing and copying is significantly improved, and there is a large gap between some positions, we can embed watermark in these positions to obtain semi-fragile watermark robust to the first printing and fragile to copying.

[0018] I i (u, v) = sign [I i (u, v)] x (|I i (u, v) | + 255k), i e a (1)

[0019] Stage 3: QR code watermark embedding stage.

[0020] (1) The binary watermark image is encrypted by Logistic chaotic mapping to generate a pseudo-random sequence, which enhances the security.

[0021] (2) Select p medium frequency coefficient positions in the energy enhanced DCT block, and count the number of signs (n + , n - ).

[0022] (3) Dynamic embedding strategy: adjust the coefficient sign distribution according to the watermark bit w e {0, 1}: if w = 0 and n + ≥ n - , or w = 1 and n + < n - , keep the coefficient unchanged. Otherwise, make n + and n - satisfy the target relationship by sign flipping to realize watermark embedding, as shown in formula (2).

[0023]

[0024] Where d is the minimum difference value parameter.

[0025] Stage 4: QR code verification stage

[0026] n is the same position coefficient statistics + With n - , according to the rule of formula (3) to decode watermark. Calculate the normalized correlation coefficient of the extracted watermark and the original watermark, set the decision threshold to distinguish the first printing and the copy attack,

[0027]

[0028] Compared with the prior art, the present application has the following outstanding advantages:

[0029] (1) High-precision copy prevention authentication:

[0030] The present application realizes accurate distinction between printing and copying by the characteristics of the intermediate frequency coefficient symbol distribution, and helps to solve the problem of identity forgery in the "one machine one code" scenario.

[0031] (2) Deployment cost and efficiency optimization:

[0032] The present application is based on the general printing and scanning process, and the watermark embedding and verification only need standard image processing algorithm, which avoids the high cost of physical label, and the algorithm complexity is low, which is suitable for mobile terminal code scanning fast authentication, and ensures the readability of two-dimensional code.

[0033] (3) Technology compatibility expansion:

[0034] The energy enhancement module of the present application can be independently applied to other digital watermarking schemes, providing a general framework for printed matter anti-counterfeiting. BRIEF DESCRIPTION OF DRAWINGS

[0035] The accompanying drawings are used to provide a further understanding of the present application, and constitute a part of the specification, together with the embodiments of the present application, to explain the present application, and do not constitute a limitation of the present application, wherein:

[0036] Figure 1 The flow chart of the present application for the attacker to copy the real-name registration two-dimensional code of the unmanned aerial vehicle.

[0037] Figure 2 The two-dimensional code watermark embedding process diagram of the present application.

[0038] Figure 3 The two-dimensional code verification process diagram of the present application. DETAILED DESCRIPTION

[0039] The preferred embodiments of the present application are described below in conjunction with the accompanying drawings, and it should be understood that the preferred embodiments described herein are only used to illustrate and explain the present application, and do not limit the present application.

[0040] Example 1:

[0041] Figure 2 is the watermark embedding process schematic diagram of the unmanned aerial vehicle real-name registration two-dimensional code anti-copy method provided by the embodiment one of the application, including the following steps:

[0042] (1) input stage, input the original image to be embedded and the watermark image.

[0043] (2) processing stage, the original image is divided into 16x16 blocks, then the discrete cosine transform is carried out on each block to obtain DCT coefficients.

[0044] (3) embedding stage, p coefficients c1, c2…cp are selected in the middle frequency sub-band of each DCT block. p Calculate n + The number of positive numbers in the p coefficients, n - The number of negative numbers in the p coefficients, according to the relationship between the watermark to be embedded and n + , n - , the DCT coefficients are changed to embed the watermark, then the modified DCT coefficients are reconstructed into spatial domain images through inverse transform, and after the watermark embedding is completed, the relationship between n + , n - needs to be verified, if it does not meet the requirements, then re-enter the embedding stage. If it meets the requirements, it means that the embedding is successful, and the process ends.

[0045] Figure 3 is the verification process schematic diagram of the unmanned aerial vehicle real-name registration two-dimensional code anti-copy method provided by the embodiment one of the application, including the following steps:

[0046] (1) input stage, input the original image to be embedded and the watermark image.

[0047] (2) processing stage, the original image is divided into 16x16 blocks, then the discrete cosine transform is carried out on each block to obtain DCT coefficients.

[0048] (3) extraction stage, p coefficients c1, c2…cp are selected in the middle frequency sub-band of each DCT block. p Calculate n + The number of positive numbers in the p coefficients, n - The number of negative numbers in the p coefficients, according to the relationship between n + , n - , the embedded watermark bit of each DCT block is obtained, and the watermark of each DCT block is extracted to obtain the final watermark.

[0049] (4) verification stage, according to the correlation degree of the extracted watermark and the standard watermark judges the similarity degree of the extracted watermark picture and the original watermark, so as to verify whether the picture is copied, if the correlation degree is less than the set threshold, the unmanned plane real name authentication code is considered to be copied, otherwise it is normal.

[0050] The method provided by the application shows excellent performance on the fragile digital watermark of the two-dimensional code, and provides strong support for promoting the progress and application of the unmanned plane real name authentication.

[0051] Finally, it should be noted that: the above only for the preferred embodiments of the application, and not for limiting the application, although the application has been described in detail with reference to the foregoing embodiments, for those skilled in the art, it still can modify the technical scheme recorded in the foregoing embodiments, or equivalent replacement for part of the technical features. Any modification, equivalent replacement, improvement, etc. within the spirit and principles of the application shall be included in the protection scope of the application.

Claims

1. A method for preventing the duplication of a QR code for real-name registration of drones, characterized by: Follow these steps: Stage 1: Input stage, input the original QR code image to be embedded with a size of 256×256 and the watermark image with a size of 16×16; Phase 2: Processing phase, the QR code image is divided into 256 non-overlapping 16×16 pixel sub-blocks and discrete cosine transform (DCT) is performed. DCT domain statistical characteristics analysis and enhancement are performed. The perturbation patterns of printing and copying on DCT coefficients are analyzed respectively. It is found that the locations with higher energy are less affected. Based on this, the energy of intermediate frequency coefficients is enhanced. Phase 3: Watermark embedding. Watermark embedding is performed according to the dynamic watermark embedding strategy. The DCT coefficients are dynamically adjusted according to the watermark bits and the number of positive and negative signs of the selected DCT coefficients to achieve watermark embedding. After the watermark embedding is completed, the relationship between the parameters is verified. The embedded watermark image is decoded according to the watermark extraction rules and the result is compared with the embedded watermark bits. If the overall correlation coefficient is greater than the threshold and all watermark bits can be decoded, the verification is considered successful and the watermark is embedded successfully. Otherwise, the verification is considered failed and the watermark needs to be re-embedded.

2. A method for preventing duplication of a drone real-name registration QR code according to claim 1, characterized in that: By analyzing the perturbation rules of printing and copying on DCT coefficients, it is found that the positions with larger energy in the DCT block are less affected. After the QR code has been printed and copied respectively, the positive and negative signs of the DC and low-frequency coefficients in the DCT coefficients are more accurate. There is a significant difference in the accuracy of printing and copying at the intermediate frequency position. In addition, embedding steganographic information in the DC and low-frequency coefficients will cause obvious visual changes to the image, while the influence of the intermediate frequency position is smaller. The energy of the coefficients with smaller absolute values ​​in the intermediate frequency coefficients can be increased to improve the effect of the intermediate frequency coefficients against printing and scanning attacks. For the a intermediate frequency coefficients with the smallest absolute values ​​in each block, as shown in formula (1): I i (u,v)=sign[I i (u,v)]×(|I i (u,v)|+255k),i∈α (1) Where k is the enhancement strength.

3. A method for preventing duplication of a drone real-name registration QR code according to claim 1, characterized in that: Perform energy enhancement of intermediate frequency coefficients: Energy enhancement is performed on some intermediate frequency coefficients with the smallest absolute values ​​in each block to improve their robustness to the first scanning noise. After the energy enhancement of the intermediate frequency coefficients, the accuracy of the first printing and copying is significantly improved. There is a large gap between some positions between the two, and watermark embedding can be performed at these positions, thus obtaining a semi-fragile watermark that is robust to the first printing and fragile to copying.

4. A method for preventing duplication of a drone real-name registration QR code according to claim 1, characterized in that: The binary watermark image is encrypted by Logistic chaotic mapping to generate a pseudo-random sequence to enhance security. In the DCT block after energy enhancement, p intermediate frequency coefficient positions are selected and the number of positive and negative signs is counted as n. + ,n - , adjust the coefficient sign distribution according to the watermark bit w∈{0,1}: if w=0 and n + ≥n - , or w=1 and n + <n - , keep the coefficient unchanged, otherwise, flip the sign to make n + With n - Satisfy the target relationship and realize watermark embedding, as shown in formula (2): Where d is the minimum difference value. If the watermark to be embedded is 0, then the lowest DCT coefficient value is modified to make n + -n - <d, and when the watermark to be embedded is 1, then n - -n + <d.

5. A method for preventing duplication of a drone real-name registration QR code according to claim 1, characterized in that: After the watermark is dynamically embedded, it is necessary to decode the watermark according to certain rules, as shown in formula (3): where n + , n - Indicates the number of positive and negative signs of the p intermediate frequency coefficients selected in the DCT block, d is the minimum difference value, according to n + , n - The relationship between the difference between and d, if n + -n - ≥d, the watermark obtained is 0, n - -n + If ≥d, the extracted watermark is 1. The normalized correlation coefficient between the extracted watermark and the original watermark is calculated, and a decision threshold is set to distinguish between first printing and copying attacks. If the normalized correlation coefficient is less than the decision threshold or the watermark cannot be decoded according to formula (3), the watermark embedding steps are repeated until the requirements are met.

Images