A multi-level piracy tracking method and system based on QR code digital fingerprint residual characteristics

Abstract

The present invention proposes a multi-level piracy tracking method and system based on the residual characteristics of QR code digital fingerprints, relating to the field of digital fingerprint technology. The method comprises: extracting the residual digital fingerprint; identifying the identity information contained therein, and targeting the purchasing user corresponding to the identity information as the target tracking object; if it is determined that the identity information cannot be identified, determining the Pearson correlation coefficient between the residual digital fingerprint and the purchasing user's digital fingerprint, and identifying the purchasing user corresponding to the digital fingerprint exceeding a preset value as an accomplice of the piracy; if it is determined that there is no accomplice exceeding the preset value, performing an evolutionary matching of the linear combination expression coefficients between the residual digital fingerprint and the purchasing user's digital fingerprint to identify the accomplice and / or the top tracking object. The present invention enhances the robustness of digital fingerprints. Through multi-level tracking, it can effectively track not only ordinary piracy but also malicious collusion piracy, improving tracking efficiency and identification accuracy.

Description

Technical Field

[0001] The present invention relates to the field of digital fingerprint technology, and in particular to a multi-level piracy tracking method and system based on QR code digital fingerprint residual features. Background Art

[0002] With the recent development of mobile networks and the widespread adoption of mobile devices, various digital products have become a widespread part of people's daily lives. This has led to an increasing severity of digital copyright protection issues, and accordingly, research on digital copyright protection technologies has received increasing attention. For certain digital works and confidential digital files, preventing legitimate users from obtaining access to these works and then illegally distributing them, as well as tracking down the disseminators, are crucial issues in digital copyright protection.

[0003] Digital fingerprinting is a digital copyright protection technology that exploits the inherent redundancy in digital works to introduce a certain amount of error into each distributed copy. Because the introduced digital fingerprint is unique to each purchaser, the source of the distribution can be traced based on the digital fingerprint in the copy. Therefore, a digital fingerprint can be considered a special type of digital watermark. In addition to meeting the requirements of digital watermarking for imperceptibility and robustness, it also requires strong collusion resistance. A collusion attack occurs when multiple users with copies of the same digital work collaborate to compare the differences between their copies of the digital work embedded with their respective digital fingerprints in order to locate the specific location of the digital fingerprint, thereby destroying the digital fingerprint embedded in the digital work and evading tracking.

[0004] The introduction of the Marking Assumption has greatly promoted the research on collusion-resistant digital fingerprint coding. Since the introduction of this assumption, the research on digital fingerprint coding has become independent from the digital fingerprint embedding algorithm. The research focus of digital fingerprints has been on how to ensure that the embedding assumption is valid, thereby resisting various collusion attacks, while the digital fingerprint embedding algorithm has focused on dealing with the attack methods of traditional digital watermarks.

[0005] In recent years, digital fingerprinting schemes based on the embedding hypothesis and utilizing combinatorial design and coding theory (Combinatorial Design-Based Anti-collusion Fingerprinting) have become a key research topic in digital fingerprinting technology. In these schemes, the collusion feasible sets obtained for different user combinations are different. By comparing the feasible sets of different users after collusion, users participating in the collusion attack—i.e., colluders among users—can be tracked. This method can also locate and track multiple colluders.

[0006] However, digital fingerprinting solutions based on combinatorial design and coding theory suffer from shortcomings such as linear tracking difficulty with user scale, high false positive rates, low tracking efficiency, poor resistance to collusion attacks, and the difficulty of establishing embedding assumptions in practical applications. Therefore, a digital fingerprinting solution that addresses these shortcomings is needed. Summary of the Invention

[0007] In view of this, the present invention proposes a multi-level piracy tracking method and system based on the residual characteristics of QR code digital fingerprints, so as to solve the problems of the current digital fingerprint schemes based on combinatorial design and coding theory, such as the tracking difficulty increases linearly with the user scale, the false detection rate is high, the tracking efficiency is low, the anti-collusion attack effect is poor, and the embedding assumptions are difficult to establish in actual application environments.

[0008] The technical solution of the present invention is achieved as follows:

[0009] According to a first aspect, an embodiment of the present invention provides a multi-level piracy tracking method based on QR code digital fingerprint residual features, the method comprising:

[0010] Extract residual digital fingerprints from the digital works to be processed that are determined to have been pirated; the digital fingerprint embedded in the digital works to be processed is a unique QR code generated based on the identity information of each purchasing user;

[0011] Using a QR decoder that matches the embedding method of the digital fingerprint to identify the identity information contained in the residual digital fingerprint, and if it is determined that the identity information can be identified, the purchasing user corresponding to the identity information is used as a target for tracking the piracy behavior;

[0012] If it is determined that the identity information cannot be identified, determining the Pearson correlation coefficient between the residual digital fingerprint and the digital fingerprints of all users who purchased the digital work to be processed stored in the user digital fingerprint database, and identifying the purchasing users whose digital fingerprints have a Pearson correlation coefficient exceeding a preset value as accomplices in the piracy;

[0013] When it is determined that there is no Pearson correlation coefficient exceeding a preset value, a linear combination expression evolution match is performed between the residual digital fingerprint and the digital fingerprints of all purchasing users of the digital work to be processed to determine the accomplices and / or top tracking targets of the piracy among the purchasing users.

[0014] In conjunction with the first aspect, in a first implementation of the first aspect, when determining that there is no Pearson correlation coefficient exceeding a preset value, performing a linear combination expression evolution match between the residual digital fingerprint and the digital fingerprints of all users who purchased the digital work to be processed to determine the accomplices and / or top tracking targets of the piracy among the purchasing users specifically includes:

[0015] When it is determined that there is no Pearson correlation coefficient exceeding a preset value, the residual digital fingerprint is used as a residual feature vector, and the residual feature vector is preprocessed to obtain a feature pattern vector;

[0016] Normalize and zero-mean the digital fingerprints of all users who purchased the digital work to be processed, and obtain a prototype pattern vector and an adjoint vector of the prototype pattern vector respectively;

[0017] Based on the evolution of matching coefficients using pseudo-inverse learning, linear combination expression coefficient evolution matching is performed to identify the accomplices and / or top tracking targets of piracy among purchasing users.

[0018] In combination with the first embodiment of the first aspect, in the second embodiment of the first aspect, evolving the matching coefficient based on the pseudo-inverse learning method, performing linear combination expression coefficient evolution matching, and determining the accomplices and / or top tracking targets of piracy among purchasing users specifically include:

[0019] Initialize the matching coefficients according to the adjoint vector;

[0020] Update the matching coefficient according to a preset dynamic ratio, and evolve the updated matching coefficient according to a preset evolution method;

[0021] After a preset number of rounds of evolution, the digital fingerprints of the purchasing users whose matching coefficients in the current round of evolution exceed 1 are determined, and the purchasing users corresponding to these digital fingerprints are regarded as candidate colluders. The digital fingerprints of the purchasing users are replaced with the candidate colluders for the next round of evolution tracking;

[0022] If it is determined that after the next round of evolution, there exists a digital fingerprint corresponding to the current round of evolution with a matching coefficient of not less than 1, the candidate conspirator corresponding to the digital fingerprint with a matching coefficient of not less than 1 in the current round of evolution is regarded as a conspirator of the piracy;

[0023] When it is determined that the matching coefficients of the current round of evolution corresponding to the digital fingerprints after the next round of evolution are all lower than 1, the evolution continues according to the candidate colluders until there is only one digital fingerprint with a matching coefficient of 1 in the current round of evolution and the remaining matching coefficients are all equal to 0, then the evolution tracking is stopped, and the candidate colluders corresponding to the digital fingerprint with a matching coefficient of 1 in the current round of evolution are taken as the number one tracking target for piracy.

[0024] In combination with the second embodiment of the first aspect, in the third embodiment of the first aspect, the calculation formula for the initialization process is:

[0025]

[0026] Among them, ξ k(0) represents the initialized matching coefficient; represents the companion vector of the prototype pattern vector of the digital fingerprint of the k-th purchasing user; w(0) represents the characteristic pattern vector of the residual digital fingerprint of the digital work to be processed.

[0027] In combination with the second embodiment of the first aspect, in the fourth embodiment of the first aspect, the calculation formula for updating the matching coefficient according to the preset dynamic ratio is:

[0028]

[0029] Among them, ξ k (i) represents the matching coefficient of the i-th round of evolution; Q k (i) represents the dynamic ratio of the i-th round of evolution; M represents the total number of users who purchased the digital works to be processed.

[0030] In combination with the second implementation of the first aspect, in the fifth implementation of the first aspect, the calculation formula of the preset evolution method is:

[0031] ξ k (i+1)=Q k (i)ξ k (i)

[0032] Among them, ξ k (i) represents the matching coefficient of the i-th round of evolution; ξ k (i+1) represents the matching coefficient of the i+1th round of evolution; Q k (i) represents the dynamic ratio of the i-th round of evolution.

[0033] In combination with the second embodiment of the first aspect, in the sixth embodiment of the first aspect, the preset rounds are set to 2 rounds.

[0034] In conjunction with the first aspect, in a seventh implementation of the first aspect, the method further comprises the following steps before the step of extracting residual digital fingerprints from the digital work to be processed that is determined to have been pirated:

[0035] When distributing digital works to purchasing users, the identity information of the purchasing user of the digital work is obtained, a unique QR code for the purchasing user is generated based on the identity information and the QR code encoding, and the QR code is embedded in the digital work based on the type of the digital work carrier.

[0036] In conjunction with the seventh implementation manner of the first aspect, in the eighth implementation manner of the first aspect, obtaining the identity information of the user who purchased the digital work, generating a QR code unique to the purchasing user based on the identity information and the QR code encoding, and embedding the QR code into the digital work based on the type of the digital work carrier specifically includes:

[0037] Encrypt the acquired identity information, and encode the encrypted identity information according to a preset QR code version number, a preset error correction level, and a QR code encoding standard to obtain a binary data codeword and an error correction codeword;

[0038] The bits corresponding to the binary data codeword and the error correction codeword are respectively filled into each module of the QR code matrix structure, and the non-functional area is masked with a preset mask pattern to obtain a QR code;

[0039] Embed the QR code into the digital work according to the type of digital work carrier.

[0040] According to a second aspect, an embodiment of the present invention provides a multi-level piracy tracking system based on QR code digital fingerprint residual features, the system comprising:

[0041] A fingerprint extraction module is used to extract residual digital fingerprints from digital works that are suspected of being pirated. The digital fingerprint embedded in the digital works is a unique QR code generated based on the identity information of each purchasing user.

[0042] A first-level tracking module is used to identify the identity information contained in the residual digital fingerprint using a QR decoder that matches the embedding method of the digital fingerprint, and if it is determined that the identity information can be identified, the purchasing user corresponding to the identity information is used as a target for tracking piracy;

[0043] a secondary tracking module for determining, when it is determined that the identity information cannot be identified, a Pearson correlation coefficient between the residual digital fingerprint and the digital fingerprints of all users who purchased the digital work to be processed stored in the user digital fingerprint library, and identifying the purchasing users whose digital fingerprints have a Pearson correlation coefficient exceeding a preset value as accomplices in the piracy;

[0044] The third-level tracking module is used to determine that there is no Pearson correlation coefficient exceeding a preset value, and then perform linear combination expression evolution matching between the residual digital fingerprint and the digital fingerprints of all purchasing users of the digital work to be processed, so as to determine the accomplices and / or top tracking targets of the piracy among the purchasing users.

[0045] The multi-level piracy tracking method and system based on the residual characteristics of QR code digital fingerprints of the present invention has the following beneficial effects compared with the prior art:

[0046] By extracting residual digital fingerprints from the digital works to be processed that are determined to have been pirated, and the digital fingerprint embedded in the digital works to be processed is a unique QR code generated based on the identity information of each purchasing user, the digital fingerprint generated based on the QR code is then used to conduct multi-level tracking of piracy, and finally identify the purchasing users involved in piracy. Since QR codes have the characteristics of easy generation and identification, large information capacity, and wide coding range, using QR codes to generate digital fingerprints greatly expands the user capacity that fingerprints can represent and enhances the robustness of digital fingerprints. QR codes themselves have certain error correction capabilities, so digital fingerprints generated based on QR codes have the ability to resist small-scale attacks, thereby achieving anti-collusion properties of digital fingerprints. After introducing QR codes to generate digital fingerprints, It is necessary to take the embedding hypothesis as a prerequisite. The QR code introduced by embedding the digital fingerprint into the digital work can effectively resist random interference in the actual application environment, can effectively resist linear and nonlinear collusion attacks, and locate the tracking target, tracking object, colluder and the number one tracking target through multi-level tracking. At the same time, the QR code is generated based on the identity information of the purchasing user, so that the digital fingerprint can use fewer feature points to represent more user information. When performing evolutionary matching tracking, there is no need to compare multiple collusion feasible sets. Even if the feasible set is large, the effect of fast tracking can be achieved, which can greatly reduce the number of evolutions, improve tracking efficiency and recognition accuracy, and through multi-level tracking, it can effectively track not only ordinary piracy but also malicious collusion piracy. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0048] Figure 1 Schematic diagram of the process of the multi-level piracy tracking method based on the residual characteristics of the QR code digital fingerprint of the present invention;

[0049] Figure 2 This is a logic diagram of the overall process of the multi-level pirated tracking method based on the residual characteristics of the QR code digital fingerprint of the present invention;

[0050] Figure 3 Schematic diagram of the process of multiple purchasing users colluding to destroy the digital fingerprint and disseminate digital works in the multi-level piracy tracking method based on the residual characteristics of the QR code digital fingerprint of the present invention;

[0051] Figure 4Schematic diagram of the process of performing evolutionary matching pursuit of collusion participants in the multi-level piracy tracking method based on QR code digital fingerprint residual features of the present invention;

[0052] Figure 5 A schematic diagram of the process of two purchasing users colluding to destroy the digital fingerprint and disseminate the digital work in the multi-level piracy tracking method based on the residual characteristics of the QR code digital fingerprint of the present invention;

[0053] Figure 6 for Figure 5 Schematic diagram of the process of evolutionary matching pursuit of two colluders in the collusion work;

[0054] Figure 7 This is a schematic diagram of the structure of the multi-level piracy tracking system based on the residual characteristics of QR code digital fingerprints of the present invention. DETAILED DESCRIPTION

[0055] The following will be combined with the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the embodiments described are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0056] In recent years, digital fingerprinting schemes based on the embedding hypothesis and utilizing combinatorial design and coding theory (Combinatorial Design-Based Anti-collusion Fingerprinting) have become a key research topic in the field of digital fingerprinting technology. In digital fingerprinting schemes based on combinatorial design and coding theory, the collusion feasible sets obtained for different user combinations are different. By comparing the feasible sets after different user collusion, users participating in the collusion attack, i.e., colluders among users, can be tracked. This method can locate and track multiple colluders, but this scheme also has at least the following disadvantages:

[0057] 1. Fingerprinting schemes based on combinatorial design and coding theory are based on the embedding assumption. The core of the embedding assumption is that collusion attacks cannot change the value of invisible bits. However, this condition is difficult to achieve in practical applications. For example, bit errors in the transmission of digital works, random noise, and attacks from target tracking objects may all violate the assumption.

[0058] 2. When the user scale is large, digital fingerprint schemes based on combinatorial design and coding theory often have the characteristics of complex codeword generation and linear growth of code length with the user scale. For example, BIBD codes with large parameters are difficult to obtain, and the tracking process consumes a lot of time, space and computational complexity. Both I codes and C-security codes have the defect that the codeword length increases linearly with the number of users.

[0059] 3. Although the digital fingerprint scheme based on combinatorial design and coding theory can locate and track multiple colluders conducting collusion attacks, it also leads to a high false positive rate, that is, detecting users with normal behavior while missing the real colluders. This is because it is difficult to simultaneously improve the code distance of the digital fingerprint and the code distance between the feasible collusion sets. If the extracted fingerprint is damaged, it will be difficult to determine which feasible collusion set the extracted fingerprint belongs to, which will lead to incorrect judgment.

[0060] 4. The digital fingerprint solution based on combinatorial design and coding theory still has the problem of low efficiency of tracking algorithm. When the number of users is n and it is set that at most m users can collude, the user digital fingerprint library responsible for storing the digital fingerprints of all buyers needs to store the feasible set of collusion: That is, the extracted digital fingerprint needs to be compared with User fingerprint and The final colluder is determined by comparing the feasible sets of collusion. The above calculation process is relatively complicated.

[0061] The multi-level piracy tracking method based on the residual characteristics of QR code digital fingerprints provided in this specification can be applied to electronic devices with digital fingerprint related processing capabilities, aiming to enhance the robustness of digital fingerprints, improve tracking efficiency and recognition accuracy. The electronic devices may include notebooks, desktop computers, smart phones, smart wearable devices (virtual reality glasses, smart watches, etc.), tablet computers, etc. Of course, the multi-level piracy tracking method based on the residual characteristics of QR code digital fingerprints provided in this specification can also be applied to applications running in the above-mentioned electronic devices. For example, the multi-level piracy tracking method based on the residual characteristics of QR code digital fingerprints can be applied to browsers with digital fingerprint related processing capabilities, and can also be applied to software with digital fingerprint related processing capabilities.

[0062] See also Figure 1 and Figure 2 , Figure 1 The flowchart of the multi-level piracy tracking method based on the residual characteristics of the QR code digital fingerprint according to the embodiment of the present invention is shown. Figure 2 The overall process logic diagram of the multi-level pirated tracking method based on the residual characteristics of QR code digital fingerprints according to an embodiment of the present invention is shown. The method may include the following steps:

[0063] S101. Extracting residual digital fingerprints from a pending digital work that is determined to have been pirated. In this embodiment, the digital fingerprint embedded in the pending digital work is a unique QR code (Quick Response Code) generated based on the identity information of each purchasing user.

[0064] In this embodiment, when a user purchases a digital work, a unique identity identification code is formed for each purchasing user through the identity information provided by the user when registering and purchasing the product, such as name, ID number, mobile phone number, email address, registration number, and other character sets of different types or lengths. The identity identification code may include a combination of character sets of different types or lengths such as name, ID number, mobile phone number, email address, registration number, etc. that can distinguish different user identities.

[0065] Preferably, the user's identity information is obtained by providing a form with a standardized format to users with purchasing intentions, and a trusted third-party center reviews and standardizes the information content and format. When the user provides identity information, the user can be required to refill the information that does not comply with the regulations.

[0066] Afterwards, a unique QR code is generated for each purchasing user using the purchasing user's identity information, and the QR code is embedded into the copy of the digital product purchased by the purchasing user, forming a digital fingerprint that can represent the purchasing user in the digital work purchased by the purchasing user.

[0067] In this embodiment, a user digital fingerprint database is also established, which is responsible for storing the digital fingerprint of each purchaser corresponding to each digital product.

[0068] Defending against collusion attacks is a key requirement for digital fingerprinting. In the case of a single user attack, existing technology can directly identify and track the target of pirated content based on the digital fingerprint extracted from digital works such as digital audio media. However, if multiple users collude to attack, the digital fingerprint may be corrupted, making it impossible to identify truly valid information, allowing the participating users to evade tracking.

[0069] Because QR codes inherently have certain error-correcting capabilities, after embedding a QR code as a digital fingerprint into a digital work, the digital fingerprint extracted from the suspected pirated work can still be quickly identified using the QR code decoding algorithm, even under certain interference conditions, to reveal the user information contained therein, making it difficult for common attack methods to erase its essential characteristics. Furthermore, QR codes possess a sufficiently large information capacity, rapid recognition capabilities, and strong error-correcting capabilities, which are beneficial for improving the tracking efficiency and robustness of digital fingerprinting schemes. Therefore, digital fingerprint encoding based on QR codes can contain a large amount of information and represent a sufficient amount of user information. The fingerprint codes of different users are sufficiently distinguishable, thus preventing users with similar identity information from generating similar digital fingerprints.

[0070] If multiple users collude to attack, they cannot directly remove the digital fingerprint from the digital work because they do not know the QR code key and other information required to remove the digital fingerprint from the digital work. Therefore, the colluders will compare their respective digital works to identify differences and manipulate the data in these different locations to destroy their respective digital fingerprints. Due to the special nature of QR codes, collusion attacks cannot completely remove the digital fingerprint embedded in the digital work. They can only blur the digital fingerprint, making it difficult for various algorithms to identify. As a result, the participating users will still leave residual digital fingerprint characteristics in the collusive work. Based on the residual digital fingerprint characteristics, through correlation coefficient and evolution tracking matching, it is possible to locate and track the colluders and even the users who have participated most in the collusion attack.

[0071] That is, QR codes have the characteristics of easy generation and recognition, large information capacity, and wide coding range. Using QR codes to generate digital fingerprints greatly expands the user capacity that fingerprints can represent and enhances the robustness of digital fingerprints.

[0072] S102: using a QR decoder that matches the embedding method of the digital fingerprint to identify the identity information contained in the residual digital fingerprint, and if it is determined that the identity information can be identified, taking the purchasing user corresponding to the identity information as a target for tracking the piracy behavior.

[0073] In this embodiment, a robust digital fingerprint embedding algorithm can be selected based on the type of digital work carrier. For example, if the digital work carrier is a digital image, the QR code digital fingerprint can be directly embedded in the image; if the digital work carrier is a digital signal such as audio, the QR code is quantized before being embedded in the digital work. A QR decoder matching the embedding method is then selected to identify the identity information contained in the residual digital fingerprint. If the identity information can be correctly identified, it indicates that the residual digital fingerprint is relatively intact and has not been damaged. The purchasing user corresponding to the identity information is the target of the piracy tracking.

[0074] S103. When it is determined that the identity information cannot be identified, determine the Pearson correlation coefficient between the residual digital fingerprint and the digital fingerprints of all purchasing users of the digital work to be processed stored in the user digital fingerprint library, and identify the purchasing users corresponding to the digital fingerprints whose Pearson correlation coefficients exceed a preset value as accomplices in the piracy behavior.

[0075] See also Figure 3 If multiple purchasing users each purchased a digital work embedded with their own encrypted digital fingerprint and participated in a conspiracy to destroy the work, they would compare their legally owned digital works and destroy the digital fingerprint by modifying pixel points with different pixel values ​​to avoid tracking. This would result in the inability to directly identify identity information from the residual digital fingerprint. In this embodiment, the Pearson correlation coefficient is used to measure the size of the residual feature. The Pearson correlation coefficient is a measure of the degree of linear correlation between variables. Specifically, the residual digital fingerprint will be compared with the digital fingerprints of all purchasing users and the Pearson correlation coefficient will be calculated. The purchasing users corresponding to the digital fingerprints with a Pearson correlation coefficient exceeding a preset value will be regarded as accomplices in the piracy.

[0076] The preset value is an empirical coefficient. Preferably, the preset value is set to 0.7. That is, if the correlation between the residual digital fingerprint and the digital fingerprint in the user digital fingerprint library is greater than 0.7, the corresponding purchasing user can be regarded as an accomplice of the piracy behavior.

[0077] S104. When it is determined that there is no Pearson correlation coefficient exceeding a preset value, a linear combination expression evolution match is performed between the residual digital fingerprint and the digital fingerprints of all purchasing users of the digital work to be processed, to determine the accomplices and / or top tracking targets of the piracy among the purchasing users.

[0078] See also Figure 4If the digital fingerprint is severely damaged after a collusion attack, the calculated Pearson correlation coefficient will not exceed the preset value, indicating that the digital fingerprint embedded in the digital work has been severely damaged by the collusion attack. The residual digital fingerprint can be considered a synthesis of the digital fingerprints already stored in the user digital fingerprint library, that is, the residual digital fingerprint is considered to be a linear combination expression of the already stored digital fingerprints. In this embodiment, if it is determined that there is no Pearson correlation coefficient exceeding the preset value, the residual digital fingerprint is matched with the linear combination expression evolution of the digital fingerprints of all users who purchased the digital work to be processed, and the accomplices and / or top tracking targets of the purchasing users who committed piracy are identified, that is, the top tracking targets with the highest degree of participation and some accomplices are identified.

[0079] See also Figure 5 and Figure 6 , taking the example of two purchasing users colluding to attack digital works, when the calculated Pearson correlation coefficient exceeds the preset value of 0.7, they will be regarded as collusion participants.

[0080] In this way, for digital works to be processed that have been pirated, the residual digital fingerprints extracted from the digital works to be processed are used, and a multi-level tracking method of digital fingerprint residual features is adopted to identify the tracking targets of the piracy behavior. The tracking process is to perform multi-level matching of the extracted fingerprint residual features with all user digital fingerprints related to the digital works to be processed in the user digital fingerprint library, and determine the tracking range based on the residual feature matching, and then determine the candidate accomplices within the tracking range each time, and the user corresponding to the user digital fingerprint matched according to the residual feature evolution tracking will be identified as the accomplice.

[0081] The multi-level piracy tracking method based on the residual characteristics of QR code digital fingerprints of the present invention extracts residual digital fingerprints from the digital works to be processed that are determined to have been pirated. The digital fingerprint embedded in the digital works to be processed is a unique QR code generated based on the identity information of each purchasing user. The digital fingerprint generated based on the QR code is then used to conduct multi-level piracy tracking, and ultimately identify the purchasing users involved in the piracy. Since QR codes have the characteristics of easy generation and identification, large information capacity, and wide encoding range, using QR codes to generate digital fingerprints greatly expands the user capacity that can be represented by fingerprints and enhances the robustness of digital fingerprints. QR codes themselves have certain error correction capabilities, so the digital fingerprints generated based on QR codes have the ability to resist small-scale attacks, thereby achieving the anti-collusion property of digital fingerprints. After the introduction of QR code to generate digital fingerprints, there is no need to use embedding assumptions as a prerequisite. The QR code introduced by embedding digital fingerprints in digital works can effectively resist random interference in actual application environments, can effectively resist linear and nonlinear collusion attacks, and locate the tracking target, tracking object, colluder and the number one tracking target through multi-level tracking. At the same time, the QR code is generated based on the identity information of the purchasing user, so that the digital fingerprint can use fewer feature points to represent more user information. When performing evolutionary matching tracking, there is no need to compare multiple collusion feasible sets. Even if the feasible set is large, the effect of fast tracking can be achieved, which can greatly reduce the number of evolutions, improve tracking efficiency and recognition accuracy, and through multi-level tracking, not only ordinary piracy can be effectively tracked, but also malicious collusion piracy can be effectively tracked.

[0082] The method may further comprise the following steps:

[0083] S201. When distributing a digital work to a purchasing user, obtain the identity information of the purchasing user, generate a unique QR code for the purchasing user based on the identity information and the QR code, and embed the QR code into the digital work based on the type of the digital work carrier.

[0084] In this embodiment, a corresponding robust digital fingerprint embedding algorithm can be selected according to the type of digital work carrier. For example, if the digital work carrier is a digital picture, the QR code digital fingerprint can be directly embedded in the picture; if the digital work carrier is a digital signal such as audio, the QR code is quantized and then embedded in the digital work.

[0085] More specifically, the acquired identity information is first encrypted, and then the encrypted identity information is encoded according to a preset QR code version number, a preset error correction level, and a QR code encoding standard, such as the GB / T18284-2000 quick response matrix code, to obtain a binary data codeword and an error correction codeword.

[0086] For example, chaotic stream cipher encryption can be used with a random key generator to generate a unique key for the user. The user's key is used to perform an XOR operation on the chaotic sequence generated by Chen's hyperchaotic system and the user's information. The encrypted information is then scrambled using Anorld transformation to finally obtain the encrypted result of the user's information, which is used to generate the binary data codeword of the QR code.

[0087] Since QR codes with low error correction levels have relatively weak anti-interference capabilities, and QR codes with high error correction levels have smaller data codeword areas, the amount of information that can be represented is also smaller, in this embodiment, a moderate error correction level of H (30%) is selected for encoding. When the error correction level is H (30%), based on the character type and length of the user information, for example, the maximum length contains 3 Chinese characters, 17 numbers, and 1 English letter (such as "Wang 189419432565671824"), QR code version 3 is used for encoding.

[0088] In order to further improve the robustness of the fingerprint, the unit module needs to be expanded. The larger the unit module, the larger the fingerprint embedding amount, and thus the stronger the robustness. However, the size of the module is limited by the size of the carrier. In this embodiment, a unit module size of 2×2 is selected according to the carrier size to generate the QR code.

[0089] Afterwards, the bits corresponding to the binary data codeword and the error correction codeword are filled into each module of the QR code matrix structure, and the non-functional area is masked using a preset mask pattern to obtain a QR code, which serves as the digital fingerprint of the purchasing user.

[0090] It should be noted that, when filling in the QR code matrix, separators, image-finding graphics, and positioning graphics can also be filled in.

[0091] The digital fingerprint embedding process based on QR code can use any spatial domain or transform domain watermark embedding algorithm. For example, a watermark embedding algorithm based on discrete wavelet transform can be used. First, a Haar-based wavelet transform is performed on the digital work to convert the pixel value of the digital fingerprint image into 2 x It is then superimposed on the low-frequency band of the digital work after wavelet decomposition, and finally an inverse wavelet transform is performed to obtain the digital work with embedded digital fingerprints.

[0092] S202: Extract residual digital fingerprints from the digital work to be processed that has been identified as pirated. In this embodiment, the digital fingerprint embedded in the digital work to be processed is a unique QR code generated based on the identity information of each purchasing user. For details, refer to step S101.

[0093] S203: Using a QR decoder that matches the embedding method of the digital fingerprint, the identity information contained in the residual digital fingerprint is identified. If the identity information can be identified, the purchasing user corresponding to the identity information is identified as a target for tracking piracy. For details, refer to step S102.

[0094] If the identity information cannot be identified, S204 determines the Pearson correlation coefficient between the residual digital fingerprint and the digital fingerprints of all users who purchased the digital work to be processed, stored in the user digital fingerprint database. The purchasing users whose fingerprints have a Pearson correlation coefficient exceeding a preset value are designated as accomplices in the piracy. For details, refer to step S103.

[0095] S205: If it is determined that there is no Pearson correlation coefficient exceeding a preset value, a linear combination expression evolution match is performed between the residual digital fingerprint and the digital fingerprints of all users who purchased the digital work to be processed, thereby identifying accomplices and / or top tracking targets among the users who purchased the work. For details, refer to step S104.

[0096] The method may further comprise the following steps:

[0097] S301: Extract residual digital fingerprints from the digital work to be processed that has been identified as pirated. In this embodiment, the digital fingerprint embedded in the digital work to be processed is a unique QR code generated based on the identity information of each purchasing user. For details, refer to step S101.

[0098] S302: Using a QR decoder that matches the embedding method of the digital fingerprint, the identity information contained in the residual digital fingerprint is identified. If the identity information can be identified, the purchasing user corresponding to the identity information is targeted for piracy tracking. For details, refer to step S102.

[0099] S303: If the identity information cannot be identified, the Pearson correlation coefficient between the residual digital fingerprint and the digital fingerprints of all users who purchased the digital work to be processed, stored in the user digital fingerprint database, is determined. The purchasing users whose fingerprints have a Pearson correlation coefficient exceeding a preset value are designated as accomplices in the piracy. For details, refer to step S103.

[0100] S3041. When it is determined that there is no Pearson correlation coefficient exceeding a preset value, the residual digital fingerprint is used as a residual feature vector w, and the residual feature vector w is preprocessed to obtain a feature pattern vector w(0).

[0101] In this embodiment, the preprocessing is normalization and zero-mean processing performed sequentially.

[0102] S3042, the digital fingerprints of all users who purchased the digital works to be processed are k , k=1,2,……,M} are normalized and zero-mean processed respectively to obtain the prototype pattern vector {v k ,k=1,2,……,M} and the adjoint vector of the prototype pattern vector

[0103] Among them, w k represents the digital fingerprint of the kth purchasing user; v k Represents the prototype pattern vector of the k-th purchasing user; represents the accompanying vector of the k-th purchasing user, and v k The generalized inverse of MP; M represents the total number of users who purchased the digital works to be processed.

[0104] In this embodiment, the residual digital fingerprint is considered to be a composite of the digital fingerprints in the user's digital fingerprint set, that is, the residual digital fingerprint is considered to be a linear combination of the digital fingerprints in the user's digital fingerprint set. Therefore, there exists an optimal linear expression of the characteristic pattern vector w(0) in the least squares sense.

[0105] Among them, ξ k Represents the matching coefficient, whose size describes the corresponding prototype pattern vector v k Represents the number of components of the characteristic mode vector w(0).

[0106] In particular, k =1 and other coefficients are 0, v k It is w(0) itself.

[0107] The normalization and zero-mean processing in step S3042 are the same as those in step S3041 and are not described in detail here.

[0108] S3043, Evolution of matching coefficient ξ based on pseudo-inverse learning k , perform linear combination expression evolution matching to identify the accomplices and / or top tracking targets of piracy among purchasing users.

[0109] Specifically, the implementation process of step S3043 is as follows:

[0110] S30431, according to the matching coefficient ξ k Perform initialization processing. The calculation formula for this process is:

[0111]

[0112] Among them, ξ k (0) represents the initial matching coefficient.

[0113] S30432. Continuously update the matching coefficient according to the preset dynamic ratio, and evolve the updated matching coefficient according to the preset evolution method. The calculation formula of this process is:

[0114]

[0115] Among them, Q k (i) represents the dynamic ratio of the i-th round of evolution; ξ k (i) represents the matching coefficient of the i-th round of evolution.

[0116] ξ k (i+1)=Q k (i)ξ k (i) (3)

[0117] Among them, ξ k (i+1) represents the matching coefficient of the i+1th round of evolution.

[0118] S30433. After the preset rounds of evolution, determine the digital fingerprints of the purchasing users whose matching coefficients in the current round of evolution are greater than 1, and use the purchasing users corresponding to these digital fingerprints as candidate colluders. Use the candidate colluders to replace the digital fingerprints of the purchasing users for the next round of evolution tracking, that is, replace the candidate colluders {w k ', k=1,2,……,M} replace the digital fingerprint of the purchasing user {w k ,k=1,2,……,M}.

[0119] Among them, w' k Represents the digital fingerprint of the kth candidate colluder.

[0120] For example, 8 candidate colluders {"Jean 207255596426323875","Amber462743513158202540","Carl 142367862683747805","Modi 870628587793456648","Louise 458222409251858545","Jack 189419432565671824","C hris724349031514069069","Jessica 695275441671173273"} are found, and the set of candidate colluders is used to replace the previous set of digital fingerprints of purchasing users, and the set of candidate colluders is used for the next round of evolution tracking.

[0121] S30434. When it is determined that after the next round of evolution, there exists a digital fingerprint corresponding to the current round of evolution with a matching coefficient not less than 1, the candidate conspirator corresponding to the digital fingerprint with a matching coefficient not less than 1 in the current round of evolution is regarded as a conspirator of the piracy.

[0122] S30435. When it is determined that the matching coefficients of the current round of evolution corresponding to the digital fingerprints after the next round of evolution are all lower than 1, the evolution continues according to the candidate colluders until there is only one digital fingerprint with a matching coefficient of 1 in the current round of evolution and the remaining matching coefficients are all equal to 0, then the evolution tracking is stopped, and the candidate colluders corresponding to the digital fingerprint with a matching coefficient of 1 in the current round of evolution are taken as the number one tracking target for piracy.

[0123] In this embodiment, the preset round is set to 2, that is, after 2 rounds of evolution, all ξ k (2) The purchasing users corresponding to no less than 1 are regarded as candidate colluders, and the candidate colluders {w' k ,k=1,2,……,M} replace the digital fingerprint of the purchasing user {w k , k=1,2,……,M} and then repeat the next round of evolution, that is, the third round, to determine ξ k (3) The set H2 of purchasing users corresponding to no less than 1 as collusion participants, that is, all the conspirators.

[0124] For example, four collusion participants are found: {"Jean 207255596426323875", "Modi870628587793456648", "Jack 189419432565671824", "Jessica 695275441671173273"}. Since this set is not empty, no further tracking evolution is required, and the evolutionary matching process ends. These four collusion participants are all the conspirators involved in the conspiracy attack.

[0125] If k (3) are all less than 1, then the evolution process continues until a certain round of evolution, there exists ξ k (i)=1, and the remaining ξ k’ (i) = 0, k' ≠ k, then the purchasing user / candidate colluder corresponding to k is the top tracking target of the piracy behavior.

[0126] The following describes a system provided by an embodiment of the present invention. The system described below and the method described above can refer to each other.

[0127] See also Figure 7 , Figure 7The following is a schematic diagram showing the structure of a multi-level piracy tracking system based on QR code digital fingerprint residual features according to an embodiment of the present invention. The system may include:

[0128] The fingerprint extraction module 10 is used to extract residual digital fingerprints from the digital works to be processed that are determined to have been pirated. In this embodiment, the digital fingerprint embedded in the digital works to be processed is a unique QR code generated based on the identity information of each purchasing user.

[0129] Preferably, the user's identity information is obtained by providing a form with a standardized format to users with purchasing intentions, and a trusted third-party center reviews and standardizes the information content and format. When the user provides identity information, the user can be required to refill the information that does not comply with the regulations.

[0130] Afterwards, a unique QR code is generated for each purchasing user using the purchasing user's identity information, and the QR code is embedded into the copy of the digital product purchased by the purchasing user, forming a digital fingerprint that can represent the purchasing user in the digital work purchased by the purchasing user.

[0131] In this embodiment, a user digital fingerprint database is also established, which is responsible for storing the digital fingerprint of each purchaser corresponding to each digital product.

[0132] Defending against collusion attacks is a key requirement for digital fingerprinting. In the case of a single user attack, existing technology can directly identify and track the target of pirated content based on the digital fingerprint extracted from digital works such as digital audio media. However, if multiple users collude to attack, the digital fingerprint may be corrupted, making it impossible to identify truly valid information, allowing the participating users to evade tracking.

[0133] Because QR codes inherently have certain error-correcting capabilities, after embedding a QR code as a digital fingerprint into a digital work, the digital fingerprint extracted from the suspected pirated work can still be quickly identified using the QR code decoding algorithm, even under certain interference conditions, to reveal the user information contained therein, making it difficult for common attack methods to erase its essential characteristics. Furthermore, QR codes possess a sufficiently large information capacity, rapid recognition capabilities, and strong error-correcting capabilities, which are beneficial for improving the tracking efficiency and robustness of digital fingerprinting schemes. Therefore, digital fingerprint encoding based on QR codes can contain a large amount of information and represent a sufficient amount of user information. The fingerprint codes of different users are sufficiently distinguishable, thus preventing users with similar identity information from generating similar digital fingerprints.

[0134] If multiple users collude to attack, they cannot directly remove the digital fingerprint from the digital work because they do not know the QR code key and other information required to remove the digital fingerprint from the digital work. Therefore, the colluders will compare their respective digital works to identify differences and manipulate the data in these different locations to destroy their respective digital fingerprints. Due to the special nature of QR codes, collusion attacks cannot completely remove the digital fingerprint embedded in the digital work. They can only blur the digital fingerprint, making it difficult for various algorithms to identify. As a result, the participating users will still leave residual digital fingerprint characteristics in the collusive work. Based on the residual digital fingerprint characteristics, through correlation coefficient and evolution tracking matching, it is possible to locate and track the colluders and even the users who have participated most in the collusion attack.

[0135] That is, QR codes have the characteristics of easy generation and recognition, large information capacity, and wide coding range. Using QR codes to generate digital fingerprints greatly expands the user capacity that fingerprints can represent and enhances the robustness of digital fingerprints.

[0136] The first-level tracking module 20 is used to use a QR decoder that matches the embedding method of the digital fingerprint to identify the identity information contained in the residual digital fingerprint, and when it is determined that the identity information can be identified, the purchasing user corresponding to the identity information is used as the target tracking object of the piracy behavior.

[0137] In this embodiment, a robust digital fingerprint embedding algorithm can be selected based on the type of digital work carrier. For example, if the digital work carrier is a digital image, the QR code digital fingerprint can be directly embedded in the image; if the digital work carrier is a digital signal such as audio, the QR code is quantized before being embedded in the digital work. A QR decoder matching the embedding method is then selected to identify the identity information contained in the residual digital fingerprint. If the identity information can be correctly identified, it indicates that the residual digital fingerprint is relatively intact and has not been damaged. The purchasing user corresponding to the identity information is the target of the piracy tracking.

[0138] The secondary tracking module 30 is used to determine the Pearson correlation coefficient between the residual digital fingerprint and the digital fingerprints of all purchasing users of the digital work to be processed stored in the user digital fingerprint library when it is determined that the identity information cannot be identified, and to regard the purchasing users corresponding to the digital fingerprints whose Pearson correlation coefficients exceed the preset value as accomplices in the piracy.

[0139] If the identity information cannot be directly identified from the residual digital fingerprint, in this embodiment, the Pearson correlation coefficient is used to measure the size of the residual feature. The Pearson correlation coefficient is a measure of the degree of linear correlation between research variables. Specifically, the residual digital fingerprint will be compared with the digital fingerprints of all purchasing users and the Pearson correlation coefficient will be calculated. The purchasing users corresponding to the digital fingerprints whose Pearson correlation coefficient exceeds the preset value will be regarded as accomplices in the piracy.

[0140] The preset value is an empirical coefficient. Preferably, the preset value is set to 0.7. That is, if the correlation between the residual digital fingerprint and the digital fingerprint in the user digital fingerprint library is greater than 0.7, the corresponding purchasing user can be regarded as an accomplice of the piracy behavior.

[0141] The third-level tracking module 30 is used to determine that there is no Pearson correlation coefficient exceeding a preset value, and then perform linear combination expression evolution matching between the residual digital fingerprint and the digital fingerprints of all purchasing users of the digital work to be processed, so as to determine the accomplices of piracy and / or the number one tracking target among the purchasing users.

[0142] If the calculated Pearson correlation coefficients do not exceed the preset value, it indicates that the digital fingerprint embedded in the digital work has been severely damaged by the collusion attack. The residual digital fingerprint can be considered a synthesis of the digital fingerprints already stored in the user digital fingerprint library, that is, the residual digital fingerprint is considered to be a linear combination expression of the already stored digital fingerprints. In this embodiment, if it is determined that no Pearson correlation coefficient exceeds the preset value, the residual digital fingerprint is matched with the linear combination expression evolution of the digital fingerprints of all users who purchased the digital work to be processed. The accomplices and / or top tracking targets of the purchasing users in the piracy behavior are identified, that is, the top tracking targets with the highest degree of participation and some of the accomplices are identified.

[0143] In this way, for digital works to be processed that have been pirated, the residual digital fingerprints extracted from the digital works to be processed are used, and a multi-level tracking method of digital fingerprint residual features is adopted to identify the tracking targets of the piracy behavior. The tracking process is to perform multi-level matching of the extracted fingerprint residual features with all user digital fingerprints related to the digital works to be processed in the user digital fingerprint library, and determine the tracking range based on the residual feature matching, and then determine the candidate accomplices within the tracking range each time, and the user corresponding to the user digital fingerprint matched according to the residual feature evolution tracking will be identified as the accomplice.

[0144] The multi-level piracy tracking system based on the residual characteristics of QR code digital fingerprints of the present invention extracts residual digital fingerprints from the digital works to be processed that are determined to have been pirated. The digital fingerprint embedded in the digital works to be processed is a unique QR code generated based on the identity information of each purchasing user. The digital fingerprint generated based on the QR code is then used to conduct multi-level tracking of piracy behavior, and ultimately identify the purchasing users involved in the piracy. Since QR codes are easy to generate and identify, have a large information capacity, and a wide encoding range, using QR codes to generate digital fingerprints greatly expands the user capacity that can be represented by fingerprints and enhances the robustness of digital fingerprints. QR codes themselves have certain error correction capabilities, so digital fingerprints generated based on QR codes have the ability to resist small-scale attacks, thereby achieving anti-collusion properties of digital fingerprints. After the introduction of QR code to generate digital fingerprints, there is no need to use embedding assumptions as a prerequisite. The QR code introduced by embedding digital fingerprints in digital works can effectively resist random interference in actual application environments, can effectively resist linear and nonlinear collusion attacks, and locate the tracking target, tracking object, colluder and the number one tracking target through multi-level tracking. At the same time, the QR code is generated based on the identity information of the purchasing user, so that the digital fingerprint can use fewer feature points to represent more user information. When performing evolutionary matching tracking, there is no need to compare multiple collusion feasible sets. Even if the feasible set is large, the effect of fast tracking can be achieved, which can greatly reduce the number of evolutions, improve tracking efficiency and recognition accuracy, and through multi-level tracking, not only ordinary piracy can be effectively tracked, but also malicious collusion piracy can be effectively tracked.

[0145] Through the description of the above embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus a necessary general hardware platform, or of course, by hardware. Based on this understanding, the essence of the above technical solution or the part that contributes to the existing technology can be embodied in the form of a software product. The computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, a magnetic disk, an optical disk, etc., and includes a number of instructions for enabling a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment or certain parts of the embodiments.

[0146] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the various embodiments of the present invention.

Claims

1. A multi-level piracy tracking method based on QR code digital fingerprint residual features, characterized by: The method comprises: Extract residual digital fingerprints from the digital works to be processed that are determined to have been pirated; the digital fingerprint embedded in the digital works to be processed is a unique QR code generated based on the identity information of each purchasing user; Using a QR decoder that matches the embedding method of the digital fingerprint to identify the identity information contained in the residual digital fingerprint, and if it is determined that the identity information can be identified, the purchasing user corresponding to the identity information is used as a target for tracking the piracy behavior; If it is determined that the identity information cannot be identified, determining the Pearson correlation coefficient between the residual digital fingerprint and the digital fingerprints of all users who purchased the digital work to be processed stored in the user digital fingerprint database, and identifying the purchasing users whose digital fingerprints have a Pearson correlation coefficient exceeding a preset value as accomplices in the piracy; If it is determined that there is no Pearson correlation coefficient exceeding a preset value, a linear combination expression evolution match is performed between the residual digital fingerprint and the digital fingerprints of all users who purchased the digital work to be processed, to determine the accomplices and / or top tracking targets of the purchasing users in the piracy behavior; When it is determined that there is no Pearson correlation coefficient exceeding a preset value, performing linear combination expression evolution matching between the residual digital fingerprint and the digital fingerprints of all users who purchased the digital work to be processed, and determining the accomplices and / or top tracking targets of the purchasing users in the piracy behavior, specifically includes: When it is determined that there is no Pearson correlation coefficient exceeding a preset value, the residual digital fingerprint is used as a residual feature vector, and the residual feature vector is preprocessed to obtain a feature pattern vector; Normalize and zero-mean the digital fingerprints of all users who purchased the digital work to be processed, and obtain a prototype pattern vector and an adjoint vector of the prototype pattern vector respectively; Evolving matching coefficients based on pseudo-inverse learning, performing evolutionary matching of linear combination expression coefficients, and identifying accomplices and / or top tracking targets of piracy among purchasing users; The method of evolving matching coefficients based on pseudo-inverse learning and performing evolutionary matching of linear combination expression coefficients to determine the accomplices and / or top tracking targets of piracy among purchasing users specifically includes: Initialize the matching coefficients according to the adjoint vector; Update the matching coefficient according to a preset dynamic ratio, and evolve the updated matching coefficient according to a preset evolution method; After a preset number of rounds of evolution, the digital fingerprints of the purchasing users whose matching coefficients in the current round of evolution exceed 1 are determined, and the purchasing users corresponding to these digital fingerprints are regarded as candidate colluders. The digital fingerprints of the purchasing users are replaced with the candidate colluders for the next round of evolution tracking; If it is determined that after the next round of evolution, there exists a digital fingerprint corresponding to the current round of evolution with a matching coefficient of not less than 1, the candidate conspirator corresponding to the digital fingerprint with a matching coefficient of not less than 1 in the current round of evolution is regarded as a conspirator of the piracy; When it is determined that the matching coefficients of the current round of evolution corresponding to the digital fingerprints after the next round of evolution are all lower than 1, the evolution continues according to the candidate colluders until there is only one digital fingerprint with a matching coefficient of 1 in the current round of evolution and the remaining matching coefficients are all equal to 0, then the evolution tracking is stopped, and the candidate colluders corresponding to the digital fingerprint with a matching coefficient of 1 in the current round of evolution are taken as the number one tracking target for piracy.

2. The multi-level piracy tracking method based on QR code digital fingerprint residual features as claimed in claim 1, characterized in that: The calculation formula for the initialization process is: ； in, Represents the initialized matching coefficient; Indicates the The adjoint vector of the prototype pattern vector of the digital fingerprint of the purchasing user; A feature pattern vector representing the residual digital fingerprint of the digital work to be processed.

3. The multi-level piracy tracking method based on QR code digital fingerprint residual features as claimed in claim 1, characterized in that: The calculation formula for updating the matching coefficient according to the preset dynamic ratio is: ； in, Indicates the Matching coefficient of round evolution; Indicates the Dynamic proportion of round evolution; Indicates the total number of users who purchased the pending digital works.

4. The multi-level piracy tracking method based on QR code digital fingerprint residual features as claimed in claim 1, characterized in that: The calculation formula of the preset evolution method is: ； in, Indicates the Matching coefficient of round evolution; Indicates the +1 round of evolution matching coefficient; Indicates the Dynamic proportion of round evolution.

5. The multi-level piracy tracking method based on QR code digital fingerprint residual features as claimed in claim 1, characterized in that: The method further comprises the following steps before the step of extracting residual digital fingerprints from the digital work to be processed that is determined to have been pirated: When distributing digital works to purchasing users, the identity information of the purchasing user of the digital work is obtained, a unique QR code for the purchasing user is generated based on the identity information and the QR code encoding, and the QR code is embedded in the digital work based on the type of the digital work carrier.

6. The multi-level piracy tracking method based on QR code digital fingerprint residual features as claimed in claim 5, characterized in that: The step of obtaining the identity information of the user who purchased the digital work, generating a QR code unique to the user based on the identity information and the QR code, and embedding the QR code into the digital work based on the type of the digital work carrier specifically includes: Encrypt the acquired identity information, and encode the encrypted identity information according to a preset QR code version number, a preset error correction level, and a QR code encoding standard to obtain a binary data codeword and an error correction codeword; The bits corresponding to the binary data codeword and the error correction codeword are respectively filled into each module of the QR code matrix structure, and the non-functional area is masked with a preset mask pattern to obtain a QR code; Embed the QR code into the digital work according to the type of digital work carrier.

7. A multi-level piracy tracking system based on QR code digital fingerprint residual features, characterized by: The system comprises: A fingerprint extraction module is used to extract residual digital fingerprints from digital works that are suspected of being pirated. The digital fingerprint embedded in the digital works is a unique QR code generated based on the identity information of each purchasing user. A first-level tracking module is used to identify the identity information contained in the residual digital fingerprint using a QR decoder that matches the embedding method of the digital fingerprint, and if it is determined that the identity information can be identified, the purchasing user corresponding to the identity information is used as a target for tracking piracy; a secondary tracking module for determining, when it is determined that the identity information cannot be identified, a Pearson correlation coefficient between the residual digital fingerprint and the digital fingerprints of all users who purchased the digital work to be processed stored in the user digital fingerprint library, and identifying the purchasing users whose digital fingerprints have a Pearson correlation coefficient exceeding a preset value as accomplices in the piracy; The third-level tracking module is used to determine that there is no Pearson correlation coefficient exceeding a preset value, and then perform a linear combination expression evolution match between the residual digital fingerprint and the digital fingerprints of all users who purchased the digital work to be processed, so as to identify the accomplices of the piracy and / or the top tracking targets among the purchasing users; When it is determined that there is no Pearson correlation coefficient exceeding a preset value, performing linear combination expression evolution matching between the residual digital fingerprint and the digital fingerprints of all users who purchased the digital work to be processed, and determining the accomplices and / or top tracking targets of the purchasing users in the piracy behavior, specifically includes: When it is determined that there is no Pearson correlation coefficient exceeding a preset value, the residual digital fingerprint is used as a residual feature vector, and the residual feature vector is preprocessed to obtain a feature pattern vector; Normalize and zero-mean the digital fingerprints of all users who purchased the digital work to be processed, and obtain a prototype pattern vector and an adjoint vector of the prototype pattern vector respectively; Evolving matching coefficients based on pseudo-inverse learning, performing evolutionary matching of linear combination expression coefficients, and identifying accomplices and / or top tracking targets of piracy among purchasing users; The method of evolving matching coefficients based on pseudo-inverse learning and performing evolutionary matching of linear combination expression coefficients to determine the accomplices and / or top tracking targets of piracy among purchasing users specifically includes: Initialize the matching coefficients according to the adjoint vector; Update the matching coefficient according to a preset dynamic ratio, and evolve the updated matching coefficient according to a preset evolution method; After a preset number of rounds of evolution, the digital fingerprints of the purchasing users whose matching coefficients in the current round of evolution exceed 1 are determined, and the purchasing users corresponding to these digital fingerprints are regarded as candidate colluders. The digital fingerprints of the purchasing users are replaced with the candidate colluders for the next round of evolution tracking; If it is determined that after the next round of evolution, there exists a digital fingerprint corresponding to the current round of evolution with a matching coefficient of not less than 1, the candidate conspirator corresponding to the digital fingerprint with a matching coefficient of not less than 1 in the current round of evolution is regarded as a conspirator of the piracy; When it is determined that the matching coefficients of the current round of evolution corresponding to the digital fingerprints after the next round of evolution are all lower than 1, the evolution continues according to the candidate colluders until there is only one digital fingerprint with a matching coefficient of 1 in the current round of evolution and the remaining matching coefficients are all equal to 0, then the evolution tracking is stopped, and the candidate colluders corresponding to the digital fingerprint with a matching coefficient of 1 in the current round of evolution are taken as the number one tracking target for piracy.

8. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the program, the steps of the multi-level piracy tracking method based on the residual features of the QR code digital fingerprint are implemented as described in any one of claims 1 to 6.

Images