A method for locating data tampering based on multi-party verification of feature bits

Abstract

A method for locating data tampering based on multi-party verification of characteristic bits, step 1: generation of an authentication watermark: dividing the original data to be protected into multiple non-overlapping blocks, extracting corresponding characteristic information from each data block; obtaining a binary characteristic sequence of the data; scrambling the binary characteristic sequence k times, and scrambling the k result sequences after each scrambling to obtain binary verification characteristic information of the same length as the binary characteristic of the data, and performing XOR on the scrambled result to obtain an authentication watermark; step 2: embedding the authentication watermark: selecting a watermark embedding algorithm, embedding the authentication watermark generated in step 1 into the original data of its corresponding carrier to obtain watermark carrier data; step 3: performing data tampering detection. The present invention can accurately locate the tampered area by scrambling the bit position and performing a multi-party verification strategy, combining the data block and its multiple mapping blocks for judgment, and shows good robustness to Gaussian noise and compression.

Description

Technical Field

[0001] The invention belongs to the technical field of multimedia information security protection, and proposes a method for tampering location by performing multi-party verification on bit positions of data suspected of tampering. Background Art

[0002] In recent years, computer network technology and multimedia information technology have developed rapidly. The Internet provides everyone with a convenient information exchange platform, enabling the rapid and convenient transmission of digital multimedia such as data, video, and audio across various networks. With the advancement of signal processing technology and the advent of multimedia editing software, people can freely modify and disseminate digital multimedia according to their will. However, while information technology has brought convenience to people's lives, it has also brought some security risks. Criminals use powerful multimedia editing software to maliciously tamper with and forge the content of digital multimedia information, making it impossible to distinguish the authenticity from the unaided eye. This has led to a continuous increase in copyright infringement and malicious tampering. In many cases, the authenticity of digital multimedia plays a decisive role in decision-making, such as in criminal investigations and court evidence collection. A piece of forged data or a modified video can affect the final judgment. The illegal copying, modification, and forgery of digital multimedia has become an increasingly serious problem. Therefore, protecting digital multimedia from unauthorized dissemination and tampering has become a top priority.

[0003] With the continuous deepening of research on digital watermarks, they can be divided into three categories according to their different application fields: robust watermarks, fragile watermarks, and semi-fragile watermarks. Robust watermarks can resist various common signal processing and have good anti-attack capabilities. They are usually used for copyright protection. By embedding secret information with product ownership into the carrier, when a copyright dispute occurs, the watermark is extracted through the watermark extraction algorithm to prove the ownership of the product. They are widely used in copyright protection. Fragile watermarks require sensitivity to any tampering and can accurately locate the tampered area. They are often used for integrity authentication of digital multimedia. However, fragile watermarks are also very sensitive to conventional signal processing operations. When attacked, it is impossible to determine whether it is accidental operation or malicious tampering, which has certain limitations. Semi-fragile watermarks combine the advantages of robust and fragile watermarks. They are resistant to operations that do not alter the primary content of the media, such as certain levels of compression, various linear filters, Gaussian noise, and data contrast adjustments. Although these operations may cause a certain degree of distortion in digital multimedia, they do not affect user experience. However, they are highly vulnerable to malicious tampering operations that seriously alter the primary content of the media, such as random collage, partial content replacement, content deletion or addition, and can locate the tampered area. Furthermore, some semi-fragile watermarking algorithms can not only locate tampering but also perform fuzzy recovery of tampered areas. Therefore, semi-fragile watermarking technology has attracted the attention of many scholars at home and abroad and has been widely used in digital multimedia content verification.

[0004] Patent publication number CN111415288 A discloses a digital watermark embedding and detection method and device. The method includes: determining a watermark embedding bit matrix and a first initial watermark corresponding to a target image based on a key, wherein the watermark embedding bit matrix includes watermark embedding bits corresponding to any pixel in the target image; determining a target watermark corresponding to the target image based on the first initial watermark; and embedding the target watermark into the target image based on the watermark embedding bit matrix. This scheme extracts n bit features and scrambles them once using a pseudorandom sequence. It compares the feature information between a single image block and a mapping block with the watermark, and is therefore limited to detecting tampering of image blocks.

[0005] Patent Publication No. 102103738A discloses a method for generating a reversible, variable-capacity watermark for digital image tampering. The method divides 2×2 image blocks into smooth and non-smooth blocks, extracts 6-bit features from smooth blocks, and 12-bit features from non-smooth blocks. The block features are encrypted to generate a block recovery watermark, which is then randomly embedded into other blocks. The authenticity of the block recovery watermark is determined by comparing the consistency of the block features and combining them with the characteristics of the block neighborhood. For image blocks determined to be tampered with, different recovery operations are performed in two steps, depending on whether the corresponding recovery watermark has been tampered with, to improve the quality of tamper recovery. The previous solution, which also involves single-shot scrambling of image blocks, is an extension of Patent Publication No. 102103738A and covers the specific methods for generating watermarks, scrambling, embedding, and extraction.

[0006] Therefore, the above method does not provide a specific method for multiple scrambling of multiple bits. If the characteristic information of a data block is simply embedded as an authentication watermark into another data block through a block mapping algorithm, it will lead to incorrect positioning of tampering and increase the false detection rate of the algorithm. Summary of the Invention

[0007] The technical problem to be solved by the present invention is: for multiple bits and multiple scrambling, how to detect the multi-party verification equation group derived from the multiple scrambling. The present invention provides a method for locating data tampering based on multi-party verification of characteristic bits.

[0008] A method for locating data tampering based on multi-party verification of characteristic bits, characterized by comprising:

[0009] Step 1: Generation of authentication watermark: Divide the original data X to be protected into multiple non-overlapping blocks, extract the corresponding feature information from each data block; convert all block features into binary representation to obtain the binary feature sequence M of the data; scramble the binary feature sequence k times, k ≥ 1, and scramble the k result sequences after each scrambling to obtain binary verification feature information A1, A2, ..., A with the same length as the binary feature of the data. k , and obtain the authentication watermark W after XORing it;

[0010] Step 2: Embedding of authentication watermark: Select a watermark embedding algorithm and embed the authentication watermark W generated in step 1 into the original data X of its corresponding carrier to obtain the watermark carrier data X. * ;

[0011] Step 3: Detect data tampering: When the watermark carrier data X * Received tampering suspicion, need to watermark carrier data X *When tampering detection and positioning is performed, the embedded watermark W' is extracted according to the corresponding watermark algorithm; then the watermark carrier data X is tampered with according to the method proposed in step 1. * Extract binary verification sequence A * , and then get the verification watermark W after XOR * According to the multi-party verification strategy, n(k+1) tampering equations are established for multi-party verification, where n is the number of bits in the binary sequence and k is the number of scrambling times. The results obtained from each equation are summed up, and the sum of the tampering equations of each data block is compared with the judgment threshold to determine whether the data block has been tampered with.

[0012] The steps of scrambling the data bit information multiple times are as follows:

[0013] Step a1: Divide the original data X into L non-overlapping data blocks X i (i=1,2,...,L), the feature information is taken from the data block X i Extract the extracted feature information and convert it into a binary sequence. Assuming that the binary feature information of each data block has n bits, the two-dimensional binary feature sequence M = {m ij |i=1,2,...,L;j=1,2,...,n},m ij Represents the information of the jth bit of the i-th data block;

[0014] Step a2: Use k groups of different random index sequences to scramble the binary feature sequence M multiple times. Each scrambling process generates a verification feature sequence A. After scrambling k times, two-dimensional verification feature sequences A1, A2, ...A are obtained. k ,

[0015] A q ={a ijq |i=1,2,...,L; j=1,2,..,n; q=1,2,...,k}

[0016] Step a3, XOR the obtained verification feature sequence to obtain the authentication watermark, that is

[0017]

[0018] The steps for detecting data tampering are as follows:

[0019] Step b1: encrypt the generated authentication watermark into the data through the watermark embedding algorithm. When the data is tampered, the same key as that used to embed the watermark is used to extract the extracted watermark W'={w' ij |i=1,2,...,L;j=1,2,...,n};

[0020] Step b2: from the watermark carrier data X containing the watermark * Get the data block regeneration verification information M * , converted into a binary sequence and then scrambled multiple times to obtain the regenerated verification feature information, and then XORed it to obtain the verification watermark

[0021] W * ={w * ij |i=1,2,...,L;j=1,2,...,n};

[0022] Step b3, according to the verification watermark W * The extracted watermark W' is used to establish equations related to tampering for multi-party verification. For the result after a single scrambling of multiple bits, two tampering equations can be listed for each bit. If a total of n bits can be obtained by extracting feature information, a total of 2n tampering equations can be established. For the result after multiple scrambling of a single bit, (k+1) tampering equations can be obtained. For the result after k-times scrambling of n bits, n(k+1) tampering equations can be obtained. The specific equation group is as follows:

[0023]

[0024] ......

[0026]

[0027] Among them, i is the data block mark, n is the number of bits, k is the number of scrambling times, A -1 q [X i ] indicates data block X i After the feature information is scrambled in A q Position in

[0028] Assume that the tampering only occurs in data block X i And all the verification feature information m of the data block * ij The extracted watermark w' of the block ij If all have changed, then the sum of n(k+1) tampered equations is equal to n(k+1);

[0029] The threshold for determining whether tampering has occurred is set to t, and the case where the sum of all equations is less than t is marked as not tampered with, and the case where the sum of the equations is greater than t is marked as the data block is tampered with.

[0030] The best choice of threshold t is half of the sum of all established equations. The best threshold for multi-bit multiple scrambling is

[0031]

[0032] The beneficial effects of the present invention are as follows: when tampering detection is required for data suspected of tampering, by scrambling the bit positions and implementing a multi-party verification strategy, combined with the data block and its multiple mapping blocks for judgment, the tampered area can be accurately located, showing good robustness to Gaussian noise and compression. Compared with the existing technology, this invention scrambles the characteristic information of the data at different bits. When the data needs to be tampered, a multi-party verification method combining watermarks with characteristic information and comparing the original data block with the mapping block is used to detect tampering, greatly improving the accuracy of locating the tampered location.

[0033] Therefore, compared to the invention patent with publication number CN111415288 A, this invention emphasizes the use of multiple scrambling of multiple bits to generate an authentication watermark, resulting in a watermark with greater robustness and resistance to malicious tampering. Furthermore, the multi-party verification technology employed in this invention is more complex: it not only compares the original data block information with the watermark, but also compares the information between multiple mapped blocks with the watermark, forming a more complete set of tampering equations. This significantly improves the accuracy of tamper detection, thus enabling a wider range of applications for various data and image tampering detection. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] Figure 1 It is the overall flow chart of the present invention.

[0035] Figure 2 It is a specific schematic diagram of the tamper detection process of the present invention. DETAILED DESCRIPTION

[0036] The present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0037] The present invention provides a method for locating data tampering based on multi-party verification of bit position disorder. The method has four parts: generating authentication watermark, embedding watermark, generating verification watermark and extracting watermark, and tamper detection. The specific flow chart is shown in the attached figure. Figure 1 shown.

[0038] To generate an authentication watermark, a suitable parameter in the data is extracted as feature information, converted into binary, quantized and the first n bits are removed. In order to improve the accuracy of tampering detection, the bit multiple scrambling method proposed in this invention is adopted. The feature information contained in different bits is scrambled multiple times and then XORed to obtain the authentication watermark.

[0039] Embed watermarks, extract watermarks, and generate verification watermarks. The authentication watermark obtained in the first step is embedded in the protected data. When the data is suspected of tampering and needs to be tampered with, it is extracted, which is called extracting the watermark. Then, the same operation is performed on the questioned data using the same method as generating the authentication watermark, and the verification information is extracted and scrambled to obtain the verification watermark.

[0040] Tamper detection uses a multi-party verification strategy to compare the extracted verification watermark with the extracted watermark. For each data block, two tampering equations used for multi-party verification can be obtained. For a data block with n bits scrambled k times, n(k+1) tampering equations can be obtained. The results of each tampering equation are summed up, and the final result is compared with the best judgment threshold t obtained through theoretical analysis and actual experiments. If it is less than the threshold, it means there is no tampering, and if it is greater than the threshold, it means it has been tampered with. After this step, it can be distinguished which data blocks have been tampered with and which data blocks have not been changed. Figure 2 This is a schematic diagram of the specific implementation method of tamper detection.

[0041] The present invention provides a method for locating data tampering based on multi-party verification of characteristic bits, the steps of which include:

[0042] Step 1: Generation of authentication watermark: Divide the original data X to be protected into multiple non-overlapping blocks, extract the corresponding feature information from each data block; convert all block features into binary representation to obtain the binary feature sequence M of the data; scramble the binary feature sequence k times, k ≥ 1, and scramble the k result sequences after each scrambling to obtain binary verification feature information A1, A2, ..., A with the same length as the binary feature of the data. k , and obtain the authentication watermark W after XORing it;

[0043] Step 2: Embed the authentication watermark: Select a watermark embedding algorithm and embed the authentication watermark W generated in step 1 into the original data X of its corresponding carrier to obtain the watermark carrier data X. * ;

[0044] Step 3: Detect data tampering: When the watermark carrier data X * Received tampering suspicion, need to watermark carrier data X * When tampering detection and positioning is performed, the embedded watermark W' is extracted according to the corresponding watermark algorithm; then the watermark carrier data X is tampered with according to the method proposed in step 1. * Extract binary verification sequence A * , and then get the verification watermark W after XOR *According to the multi-party verification strategy, n(k+1) tampering equations are established for multi-party verification, where n is the number of bits in the binary sequence and k is the number of scrambling times. The results obtained from each equation are summed up, and the sum of the tampering equations of each data block is compared with the judgment threshold to determine whether the data block has been tampered with.

[0045] In step 1, the steps for scrambling the data bit information multiple times are as follows:

[0046] Step a1: Divide the data X into L non-overlapping data blocks X i (i=1,2,...,L), the feature information is taken from the data block X i Extract the extracted feature information and convert it into a binary sequence. Assuming that the binary feature information of each data block has n bits, the two-dimensional binary feature sequence M = {m ij |i=1,2,...,L;j=1,2,...,n},m ij Represents the information of the jth bit of the i-th data block;

[0047] Step a2: Use k groups of different random index sequences to scramble the binary feature sequence M multiple times. Each scrambling process generates a verification feature sequence A. After scrambling k times, two-dimensional verification feature sequences A1, A2, ...A are obtained. k ,

[0048] A q ={a ijq |i=1,2,...,L; j=1,2,..,n; q=1,2,...,k}

[0049] Step a3, XOR the obtained verification feature sequence to obtain the authentication watermark, that is

[0050]

[0051] like Figure 2 As shown, in step 3, the steps for tampering detection of data are as follows:

[0052] Step b1: encrypt the generated authentication watermark into the data through the watermark embedding algorithm. When the data is tampered, the same key as that used to embed the watermark is used to extract the extracted watermark W'={w' ij |i=1,2,...,L;j=1,2,...,n}.

[0053] Step b2: from the data X containing the watermark * Get the data block regeneration verification information M * , converted into a binary sequence and then scrambled multiple times to obtain the regenerated verification feature information, and then XORed it to obtain the verification watermark W* ={w * ij |i=1,2,...,L;j=1,2,...,n}.

[0054] Step b3, according to the verification watermark W * The extracted watermark W' is used to establish equations related to tampering for multi-party verification. For the result of a single scrambling of multiple bits, two tampering equations can be listed for each bit. Assuming that the extracted feature information can obtain n bits in total, a total of 2n tampering equations can be established. For the result of multiple scrambling of a single bit, (k+1) tampering equations can be obtained. For the result of k-times scrambling of n bits, n(k+1) tampering equations can be obtained. The specific equation group is as follows:

[0055]

[0056] ......

[0058]

[0059] Among them, i is the data block mark, n is the number of bits, k is the number of scrambling times, A -1 q [X i ] indicates data block X i After the feature information is scrambled in A q The position in.

[0060] Through these equations, we can get the mapping block and data block X i Assume that the tampering only occurs in data block X i And all the verification feature information m of the data block * ij The extracted watermark w' of the block ij If all of the data blocks have changed, the sum of the n(k+1) tampering equations will be equal to n(k+1). However, since the feature information and the extracted watermarks are both binary sequences consisting of 0 and 1, the probability of change is 50%, and most of the locations where tampering occurs are within a large area. Not only the data block itself may change, but the corresponding mapping block may also change, so the sum of all tampering equations will be less than n(k+1). The sum of the n(k+1) tampering equations related to the data block can be used to determine whether the data block itself has been tampered with. The threshold for determining whether tampering has occurred is set to t, and the case where the sum of all equations is less than t is marked as not tampered with, and the case where the sum of the equations is greater than t is marked as the data block being tampered with.

[0061] The best choice of threshold t is half of the sum of all established equations. The best threshold for multi-bit multiple scrambling is At this time, the ratio of the detection rate to the false positive rate is the largest, that is, the difference between the tampered area and the non-tampered area is the largest, and the detection rate requirement is met.

[0062] This specification uses multiple bits being jumbled once and a single bit being jumbled multiple times as examples for specific explanation.

[0063] Example 1

[0064] Example 1 is a case where multiple bits of a data block are scrambled once

[0065] Step 1: Generation of authentication watermark

[0066] 1.1. Data segmentation: Given a set of original data X that needs to be protected, the data is divided into L non-overlapping data blocks X i (i=1,2,...,L).

[0067] 1.2. Feature extraction and scrambling, including the following steps:

[0068] Step 1.2.1: Extract any appropriate parameter of the data as feature information.

[0069] Step 1.2.2: quantize the obtained feature information and convert it into binary to obtain an n-bit binary feature sequence. The feature sequence of the original data is expressed as M = m ij (i=1,2,...,L; j=1,2,...,n), m ij Represents the information of the jth bit of the i-th data block.

[0070] Step 1.2.3, for all data blocks X of the entire original data X i (i=1,2,...,L) Perform a single scrambling within the same bit position, that is, between different data blocks. The parameter of each bit position scrambling is used as a set of keys. The binary feature sequence has n bits in total, so there are n sets of parameters in total. These n sets of different keys are recorded as Key1. Each bit position is scrambled once to obtain a new verification feature sequence A. The results after scrambling once are A1, A2,..., A n , a total of n groups of scrambled sequences, where the element in each scrambled bit is recorded as a ij (i=1,2,...,L;1,2,...,n), and then XOR the obtained verification feature information, that is, Take the elements in W as database X i Authentication watermark.

[0071] Step 2: Watermark embedding

[0072] Before tamper detection, the generated authentication watermark needs to be embedded, and the generated watermark is encrypted and embedded into the original data. The key is recorded as Key2, and the protected watermark carrier data X is obtained. * .

[0073] Step 3: Detect tampering of the data block that is suspected of tampering

[0074] Step b1: Assume that the original data block X is suspected of being tampered with. Select a suitable watermark extraction algorithm to extract the watermark, use Key2 as the key, and obtain the extracted watermark, which is recorded as W'.

[0075] Step b2: Generate a verification watermark using the same method as that used to generate the authentication watermark. Extract a parameter of each data block that has been tampered with as feature information, and convert it into binary, which is called a regeneration feature sequence.

[0076] M * ={m * ij |i=1,2,...,L;j=1,2,...,n}. The element m in the sequence * ij It is called regeneration feature information. The regeneration feature sequence is scrambled using the same scrambling method used to generate the authentication watermark. The random sequence used for scrambling uses the key Key1. After scrambling, n groups of regeneration verification feature sequences A are generated. * q (q=1,2,...,n), using A * q Mutual XOR to obtain verification watermark

[0077]

[0078] After these two steps, we obtain the regenerated signature information of the data required for tamper detection, the verification watermark, and the extracted watermark of the original watermarked data. At this point, each data block corresponds to a mapping block, and each data block stores the verification signature information of other data blocks. At the same time, each data block can also regenerate its own verification signature information.

[0079] Step b3: Use a multi-party verification strategy for the data that needs to be tested. For multiple bits and multiple scrambling, a total of n(k+1) tampering equations can be obtained. For the case of scrambling once, two tampering-related equations can be listed for each bit. The first equation compares the data block X i Extract authentication watermark and verification watermark, at this time the information contained in the two watermarks is data block X i The feature information of the inverse mapping block is XORed with its verification feature information; the second equation compares the data block X iThe extracted authentication watermark and verification watermark of the scrambled mapping block are the information contained in the two watermarks, which are the data block X i The characteristic information of the data block X is XORed with the verification characteristic information. i The relationship between the watermark and feature information contained in each data block during multi-party verification, as well as the process of verifying one of the single-bit unit information, is as follows: Figure 2 shown.

[0080] The equation for a random bit position is as follows:

[0081]

[0082] Among them, A -1 q [X i ] represents X i The characteristic information of A q Position in, w' ij and w * ij They are respectively extracted authentication watermark and verification watermark. For each bit, two tampering equations can be listed. If the tampering only occurs in the data block X i , then at this time If tampering occurs on a large scale, data block X i When the mapping block or its inverse mapping block is also within the tampering range, For data with n bits of characteristic information, 2n equations can be listed, which are:

[0083]

[0084] If only data block X i If tampering occurs, but the mapping block does not change, and each value of the authentication watermark has changed, then the sum of all equations is 2n, but for other tampering situations, the sum of the equations ranges from 0 to 2n. When using this equation to determine whether the data block has been tampered with, it is necessary to set a judgment threshold. If it is greater than the threshold, the data block is preliminarily marked as a tampered block, and if it is less than the value, it is marked as a non-tampered block. Theoretical derivation and experiments show that the ratio of detection rate to false detection rate is large and the detection rate is greater than 50%, that is, the optimal judgment threshold t is half of the total number of equations. For the 2n equations generated by multi-bit single scrambling, the optimal threshold is n. The tampering parameter is d i (i=1,2,...,L). The formula is as follows:

[0085]

[0086] d iWhen it is 1, the block is judged to be a tampered block, and when it is 0, the block is judged to be a non-tampered block.

[0087] Example 2

[0088] Example 2 is a case where a single bit of a data block is scrambled multiple times

[0089] Step 1: Generate authentication watermark

[0090] Step a1: Use the same method as the data block multi-bit single scrambling method to first divide the data into blocks to obtain X i (i=1,2,...,L), extract a parameter of each data block to get the feature sequence M={m i |i=1,2,...,L}.

[0091] Step a2: Multiple scrambling to generate authentication watermarks. The obtained feature information is scrambled multiple times using k sets of different random parameters, recorded as key Key1. Each bit position is scrambled once to obtain a new verification feature sequence A. In this case, for a single bit position, k times of scrambling can obtain k verification feature information sequences A. q ={a iq |i=1,2,...,L;q=1,2,...,k}. XOR the obtained k verification feature information sequences to obtain the authentication watermark W, that is,

[0092]

[0093] Step 2: Watermark embedding: Before tamper detection, the generated authentication watermark needs to be embedded. The generated watermark is embedded into the original data using the key Key2 to obtain the watermark carrier data X that needs to be protected. * .

[0094] Step 3: Detect tampering of suspected data

[0095] Step b1: Select a suitable watermark extraction algorithm according to the data currently to be protected to extract the watermark. The key is the same as that used to embed the authentication watermark, Key2, to obtain the extracted watermark W'.

[0096] Step b2: Generate a verification watermark. Assume that the original data block X has been tampered with. * Perform the same operation as when generating the authentication watermark, divide it into L different data blocks with no overlapping areas, and then extract a certain characteristic parameter of the data block as the characteristic information to obtain L regenerated characteristic information M * ={m * i|i=1,2,...,L}. The same scrambling parameter Key1 as used in generating the authentication watermark is used to scramble the regenerated feature information multiple times to obtain k regenerated verification feature information sequences.

[0097] A * q ={a * iq |i=1,2,...,L; q=1,2,...,k}Use A * q Get verification watermark

[0098] Step b3: Now we have the verification watermark, extract the watermark and regenerate the verification feature information. To locate the data tampering, we need to list the tampering equations. For n-bit data scrambled k times, we can get n(k+1) equations. For a single bit, there will be (k+1) tampering equations. One equation is compared with w * i with w' i ,Right now Each time the data block is shuffled, i The data in will be scrambled to other data blocks, namely X i The mapping block is then compared with the verification watermark and the extraction watermark in the mapping block. Among them A q -1 [X i ] indicates data block X i Feature information m i The verification feature sequence A after scrambling q Each time the scrambling is performed, a new mapping block is generated, corresponding to a new tampering equation.

[0099] For data block X i The tampered equations after scrambling k times are:

[0100]

[0101] Sum the results of all tampering equations and compare the result with the judgment threshold t. If the result is greater than the threshold, it is judged as a tampered block, and if it is less than the threshold, it is judged as a non-tampered block. For the case of single-bit scrambling k times, the optimal judgment threshold is Right now: If d i =1, the block is judged to be a tampered block. If d i =0, the block is judged to be a non-tampered block.

Claims

1. A method for locating data tampering based on multi-party verification of characteristic bits, characterized by: include: Step 1: Generation of authentication watermark: Divide the original data X to be protected into multiple non-overlapping blocks, extract the corresponding feature information from each data block; convert all block features into binary representation to obtain the binary feature sequence M of the data; scramble the binary feature sequence k times, k ≥ 1, and scramble the k result sequences after each scrambling to obtain binary verification feature information A1, A2, ..., A with the same length as the binary feature of the data. k , and obtain the authentication watermark W after XORing it; The steps of scrambling the data bit information multiple times are as follows: Step a1: Divide the original data X into L non-overlapping data blocks X i (i=1,2,...,L), the feature information is taken from the data block X i Extract the extracted feature information and convert it into a binary sequence. Assuming that the binary feature information of each data block has n bits, the two-dimensional binary feature sequence M = {m ij |i=1,2,...,L;j=1,2,...,n},m ij Represents the information of the jth bit of the i-th data block; Step a2: Use k groups of different random index sequences to scramble the binary feature sequence M multiple times. Each scrambling process generates a verification feature sequence A. After scrambling k times, two-dimensional verification feature sequences A1, A2, ...A are obtained. k , A q ={a ijq |i=1,2,...,L; j=1,2,..,n; q=1,2,...,k} Step a3, XOR the obtained verification feature sequence to obtain the authentication watermark, that is Step 2: Embedding of authentication watermark: Select a watermark embedding algorithm and embed the authentication watermark W generated in step 1 into the original data X of its corresponding carrier to obtain the watermark carrier data X. * ; Step 3: Detect data tampering: When the watermark carrier data X * Received tampering suspicion, need to watermark carrier data X * When tampering detection and positioning is performed, the embedded watermark W' is extracted according to the corresponding watermark algorithm; then the watermark carrier data X is tampered with according to the method proposed in step 1. * Extract binary verification sequence A * , and then get the verification watermark W after XOR * ; According to the multi-party verification strategy, n(k+1) tampering equations are established for multi-party verification, where n is the number of bits in the binary sequence and k is the number of scrambling times. The results obtained from each equation are summed up, and the sum of the tampering equations for each data block is compared with the judgment threshold to determine whether the data block has been tampered with. The steps for detecting data tampering are as follows: Step b1: encrypt the generated authentication watermark into the data through the watermark embedding algorithm. When the data is tampered, the same key as that used to embed the watermark is used to extract the extracted watermark W'={w' ij |i=1,2,...,L;j=1,2,...,n}; Step b2: from the watermark carrier data X containing the watermark * Get the data block regeneration verification information M * , converted into a binary sequence and then scrambled multiple times to obtain the regenerated verification feature information, and then XORed it to obtain the verification watermark W * ={w * ij |i=1,2,...,L;j=1,2,...,n}; Step b3, verify the watermark W * The extracted watermark W' is used to establish equations related to tampering for multi-party verification. For the result after a single scrambling of multiple bits, two tampering equations can be listed for each bit. If a total of n bits can be obtained by extracting feature information, a total of 2n tampering equations can be established. For the result after multiple scrambling of a single bit, (k+1) tampering equations can be obtained. For the result after k-times scrambling of n bits, n(k+1) tampering equations can be obtained. The specific equation group is as follows: ...... Among them, i is the data block mark, n is the number of bits, k is the number of scrambling times, A -1 q [X i ] indicates data block X i After the feature information is scrambled in A q Position in Assume that the tampering only occurs in data block X i And all the verification feature information m of the data block * ij The extracted watermark w' of the block ij If all have changed, then the sum of n(k+1) tampered equations is equal to n(k+1); The threshold for determining whether tampering has occurred is set to t, and the case where the sum of all equations is less than t is marked as not tampered with, and the case where the sum of the equations is greater than t is marked as the data block is tampered with; The best choice of threshold t is half of the sum of all established equations. The best threshold for multi-bit multiple scrambling is

Images