Three-dimensional medical volume data zero-watermarking method based on hexad discrete cosine transform

Abstract

The application discloses a three-dimensional medical body data zero watermarking method based on hexad number discrete cosine transform, which comprises the following steps: carrying out slice processing and grouping on original three-dimensional medical body data, then carrying out Tucker decomposition on each medical image slice group to obtain a first energy feature map; carrying out two-dimensional non-subsampled discrete wavelet transform on each obtained first energy feature map of each group to obtain a corresponding low-frequency subband, then carrying out non-overlapping block division on each group of low-frequency subbands and constructing a hexad number image block, calculating a direct current component of the image block after hexad number discrete cosine transform in a space domain, and finally obtaining a binary image feature matrix according to L2 norms of the direct current components and using the binary image feature matrix for zero watermarking authentication. The application has strong robustness to various conventional attacks in digital watermarking attacks, satisfies high fidelity requirements in image data copyright protection in the medical field, and has high security.

Description

Technical Field

[0001] This invention relates to the field of image processing technology, and in particular to a zero-watermarking method for three-dimensional medical volume data based on hexadecimal discrete cosine transform. Technical Background

[0002] With the rapid development of multimedia technology and network applications in recent years, digitalization has become widespread in the medical field. Applying computer information network technology to the entire medical process has gradually become a new model of modern medicine, as well as a development direction and management goal of public healthcare. Some new medical technologies are also rapidly developing and being widely applied in medical practice, such as digital medical testing, telemedicine, and medical digital imaging technology, accelerating the process of medical digitalization. The medical field is rapidly transforming from traditional medicine to telemedicine, with large amounts of medical information, including numerous medical images, being transmitted quickly and efficiently over public networks.

[0003] Currently, medical equipment such as computed tomography (CT) and magnetic resonance imaging (MRI) outputs three-dimensional medical volume data, which is inevitably subject to risks of tampering and theft during transmission. Furthermore, in the medical field, even the slightest change can indicate a disease, so any alteration to the original image can distort the medical image, thus affecting the doctor's diagnosis. To address these issues, we employ zero-watermarking technology to protect the copyright information of medical volume data. Zero-watermarking technology typically involves XORing extracted features of the original image with a binary watermark image to generate a unique key, thereby achieving watermark embedding without altering the pixel information in the original image. This technology not only balances the imperceptibility and robustness of traditional embedded watermarks but also causes no loss to the original image, thus preserving its normal usability. Zero-watermarking technology is one of the important research directions in the field of information hiding.

[0004] However, there are relatively few zero-watermarking algorithms currently applied to medical body data, and these algorithms are not very robust to common image attacks. With the increasing transmission of medical information over the internet, improving the robustness of zero-watermarking algorithms for medical body data to reduce the risk of content tampering and theft has become particularly important and urgent. Summary of the Invention

[0005] This invention provides a zero-watermarking method for 3D medical volume data based on hexadecimal discrete cosine transform, which can effectively resist various conventional image processing attacks to protect the copyright information of 3D medical volume data and solve the problem of poor robustness of existing 3D medical volume data zero-watermarking algorithms.

[0006] This invention provides a zero-watermarking method for three-dimensional medical volume data based on hexadecimal discrete cosine transform, comprising:

[0007] Embed zero watermarks in the original 3D medical body data and extract zero watermarks from the 3D medical body data to be authenticated;

[0008] The process of embedding zero watermarks into the original three-dimensional medical volume data includes:

[0009] The Fibonacci Q matrix is ​​used as the scrambling matrix to scramble the original N×N binary watermark image W, resulting in a scrambled binary watermark image W. A ;

[0010] The original three-dimensional medical volume data V of size M×M×P is sliced ​​and grouped, and the processed medical image slice group is denoted as U. i , i = 1, 2, 3, ..., 15; for each medical image slice group U i All samples underwent Tucker decomposition, and the first energy feature map after each Tucker decomposition was extracted and labeled as E. i , i=1,2,3,...,15; for E i Each component undergoes a k-level two-dimensional non-subsampled discrete wavelet transform to obtain the corresponding low-frequency subband L. i , i = 1, 2, 3, ..., 15;

[0011] For low-frequency subband L i Each component is divided into non-overlapping sub-blocks, and then the L-shaped components are used. i Construct hexagonal image blocks from the corresponding sub-blocks in each component and calculate the DC coefficient D of the image block after hexagonal discrete cosine transform in the spatial domain. st D st The magnitudes of the real and imaginary parts are represented by a one-dimensional vector O. st This indicates that O was finally used. st The L2 norm yields the binary robust characteristic matrix T;

[0012] The binary robust feature matrix T is scrambled using a two-dimensional Sinusoidal-Chebyshev chaotic system and a Spatial Pyramid Matching chaotic system to obtain the binary image feature matrix J.

[0013] The scrambled binary watermark image W corresponding to the original binary watermark image W. A The binary image feature matrix J is XORed with the binary image feature matrix J to obtain the certified zero-watermark image H. H is saved to the watermark database of a third-party registration agency, and the control parameters and process parameters used in the binary zero-watermark embedding process are saved as keys.

[0014] The process of extracting watermarks from the 3D medical body data to be authenticated includes:

[0015] The three-dimensional medical volume data V′ of size M×M×P, which is to be authenticated, is sliced ​​and grouped. The processed medical image slice group is denoted as U. i ′, i = 1, 2, 3, ..., 15; for each medical image slice group U i All samples were subjected to Tucker decomposition, and the first energy feature map after each Tucker decomposition was extracted and labeled as E. i ′, i=1,2,3,...,15; for E i Each component in the vector is subjected to a k-level two-dimensional non-subsampled discrete wavelet transform to obtain the corresponding low-frequency subband L. i ′,i=1,2,3,……,15;

[0016] For low-frequency subband L i Each component in ' is divided into non-overlapping sub-blocks, and then the L-shaped sub-blocks are used. i Construct hexagonal image blocks from the corresponding sub-blocks in each component of the image block and calculate the DC coefficient D of the image block after hexagonal discrete cosine transform in the spatial domain. st ′, will D st The magnitudes of the real and imaginary parts of ′ are represented by a one-dimensional vector O. st ′ indicates that O is used last. st The L2 norm of T' yields the binary robust characteristic matrix T';

[0017] The binary robust feature matrix T′ is scrambled using a two-dimensional Sinusoidal-Chebyshev chaotic system and a Spatial Pyramid Matching chaotic system to obtain the binary image feature matrix J′.

[0018] Retrieve the certified zero-watermark image H stored in the watermark database of a third-party registration agency, and XOR it with the binary image feature matrix J′ to obtain the binary watermark image W to be descrambled. A ′;

[0019] The binary watermark image W is treated as a descrambling matrix using the Fibonacci Q inverse matrix for descrambling operations. A By performing an unscramble operation, the final watermark information W′ can be extracted. Finally, the copyright ownership of the three-dimensional medical body data V′ to be authenticated is determined based on the content information displayed by W′.

[0020] This invention provides a zero-watermarking method for 3D medical volume data based on hexadecimal discrete cosine transform. It uses the Fibonacci Q matrix as the scrambling matrix to scramble the original watermark image, and simultaneously employs a two-dimensional Sinusoidal-Chebyshev chaotic system and a Spatial Pyramid Matching chaotic system to encrypt and scramble the feature matrix, thus improving the security of the zero-watermarking algorithm. The Tucker decomposition and hexadecimal discrete cosine transform are used to construct the zero-watermark by utilizing the correlation between each slice of the 3D medical volume data and its low-frequency components. Furthermore, by directly calculating the DC coefficients of the hexadecimal image blocks after hexadecimal discrete cosine transform in the spatial domain, the computational cost is effectively reduced, enhancing the algorithm's timeliness. Simultaneously, before constructing the zero-watermark, a two-dimensional non-subsampled discrete wavelet transform is used to extract low-frequency components, effectively improving the robustness and watermark embedding capacity of this method. Attached Figure Description

[0021] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings required in the description of the embodiments will be briefly described below. Obviously, the drawings described below are only a part of the embodiments of the present invention, and those skilled in the art can obtain other drawings based on these drawings without creative effort.

[0022] Figure 1 This is a flowchart illustrating the embedding method in a three-dimensional medical volume data zero-watermarking method based on hexadecimal discrete cosine transform provided in an embodiment of the present invention.

[0023] Figure 2 This is a flowchart illustrating the method for extracting watermarks from three-dimensional medical volume data based on the discrete cosine transform of hexadecimals, as provided in an embodiment of the present invention.

[0024] Figure 3 This is a schematic diagram of a single slice of MRI and CT brain three-dimensional medical volume data in a zero-watermarking method for three-dimensional medical volume data based on hexadecimal discrete cosine transform provided in an embodiment of the present invention;

[0025] Figure 4 This is a schematic diagram of the original binary watermark image in a three-dimensional medical volume data zero-watermarking method based on hexadecimal discrete cosine transform provided in an embodiment of the present invention;

[0026] Figure 5 This invention provides a test NC value table for MRI and CT brain three-dimensional medical volume data under various non-geometric attacks in a zero-watermarking method for three-dimensional medical volume data based on hexadecimal discrete cosine transform.

[0027] Figure 6This invention provides a test NC value table for MRI and CT brain three-dimensional medical volume data under various geometric attacks in a zero-watermarking method for three-dimensional medical volume data based on hexadecimal discrete cosine transform. Detailed Implementation

[0028] To enable those skilled in the art to better understand the present invention, the technical solutions of the embodiments of the present invention will be described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are merely some, not all, of the embodiments of the present invention. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort should fall within the scope of protection of the present invention.

[0029] Figure 1 This is a flowchart illustrating the embedding method in a three-dimensional medical volume data zero-watermarking method based on hexadecimal discrete cosine transform provided in an embodiment of the present invention. Figure 2 This is a flowchart illustrating a zero-watermark extraction method for 3D medical volume data based on hexadecimal discrete cosine transform, provided in an embodiment of the present invention. This embodiment is applicable to situations where zero-watermarking is used for copyright protection of 3D medical volume data.

[0030] like Figure 1 As shown in the figure, the embedding method in the three-dimensional medical volume data zero-watermarking method based on hexadecimal discrete cosine transform provided in this embodiment specifically includes the following steps:

[0031] S110. Using the Fibonacci Q matrix as the scrambling matrix, the original N×N binary watermark image W is scrambled to obtain the scrambled binary watermark image W. A .

[0032] Figure 4 This is a schematic diagram of the original binary watermark image in a three-dimensional medical volume data zero-watermarking method based on hexadecimal discrete cosine transform provided in an embodiment of the present invention. The image is N×N in size. In this embodiment, the original binary watermark image provided above can be used to achieve binary watermark scrambling.

[0033] The scrambling operation on the original binary watermark image W using the Fibonacci Q matrix as the scrambling matrix is ​​expressed by the following formula:

[0034]

[0035] In the formula, Q p Let N be the p-order Fibonacci Q-matrix, i.e., the scrambling matrix. Let N be the length and width of the original binary watermark image W, x and y be the pixel coordinates in the original binary watermark image W before scrambling, and x′ and y′ be the scrambled binary watermark image W after scrambling.A The position coordinates of the middle pixel, mod(·) is the remainder function.

[0036] The p-order Fibonacci Q matrix is ​​achieved by the following formula:

[0037]

[0038] In the formula, F p+1 F p F p-1 All are the terms in the Fibonacci sequence. The order p is denoted as the key Key1 and stored.

[0039] In this embodiment, the Fibonacci Q matrix is ​​used as the scrambling matrix to scramble the original binary watermark image W. This method has low computational cost and high degree of chaos, which can enhance the timeliness and security of the overall watermarking algorithm.

[0040] S120. Slice and group the original three-dimensional medical volume data V of size M×M×P, and group each medical image slice group U. i All samples underwent Tucker decomposition, and the first energy feature map E of each Tucker decomposition was extracted. i For E i Each component undergoes a k-level two-dimensional non-subsampled discrete wavelet transform to obtain the corresponding low-frequency subband L. i .

[0041] For example, the original three-dimensional medical volume data V of size M×M×P is sliced ​​and divided into 15 medical image slice groups, denoted as U. i For i = 1, 2, 3, ..., 15, the following method can be used:

[0042]

[0043] In the formula, P is the total number of slices after processing the three-dimensional medical body data slices, floor(·) is the function to round down to the nearest integer, and mod(·) is the function to take the remainder.

[0044] Will U i Converting to tensor form and performing Tucker decomposition, the medical image slice group U i It can be represented as:

[0045] U i =G i ×1V i (1) ×2V i (2) ×3V i (3)

[0046] In the formula, i=1,2,3,...,15, U i G represents the tensor representation of each medical image slice group. i For each group's core tensor, V (1)i V (2)i V (3)i This represents the factor matrix of each group of tensors in different directions.

[0047] Let tensor X i =G i ×1V i (1) ×2V i (2) For tensor U i Extract the subtensors X from each group. i The first sub-image, namely the medical image slice group U i The first energy characteristic map after Tucker decomposition is denoted as E. i , i = 1, 2, 3, ..., 15.

[0048] For E i Perform a k-level two-dimensional non-subsampled discrete wavelet transform to obtain E i The low-frequency subband L corresponding to each component i , i = 1, 2, 3, ..., 15.

[0049] Three-dimensional medical volume data is a typical high-dimensional medical image data. Processing typically requires slicing. By slicing and grouping the 3D medical volume data before performing Tucker decomposition and extracting the first energy feature map, the correlation between each slice in each medical image slice group can be effectively preserved. Furthermore, it retains most of the energy of the 3D medical volume data as a whole, which helps enhance the overall robustness of the zero-watermarking algorithm for medical volume data. Simultaneously, using grouping operations combined with Tucker decomposition can improve the algorithm's timeliness without affecting robustness, avoiding processing all slices. Using two-dimensional non-subsampled discrete wavelet transform to obtain low-frequency sub-band images can preserve most of the information in the first energy feature map after Tucker decomposition without reducing the size. Based on this low-frequency sub-band, a robust feature matrix can be constructed, which can enhance the algorithm's robustness while ensuring the watermark embedding capacity.

[0050] S130, for low-frequency subband L i Each component is divided into non-overlapping sub-blocks, and then the L-shaped components are used. i Construct hexagonal image blocks from the sub-blocks corresponding to each component position and calculate the DC coefficient D of the image block after hexagonal discrete cosine transform in the spatial domain. st D st The magnitudes of the real and imaginary parts are represented by a one-dimensional vector O.st This indicates that O was finally used. st The L2 norm yields the binary robust characteristic matrix T.

[0051] The L i Perform non-overlapping sub-block partitioning of size r×r, and then use L i Construct hexagonal image blocks from the sub-blocks corresponding to each component position and calculate the DC coefficient D of the image block after hexagonal discrete cosine transform in the spatial domain. st This can be achieved in the following way:

[0052]

[0053] In the formula, u g It is a unit pure hexadecimal, and satisfies

[0054] L st (x,y) represents the L after block partitioning. i The hexadecimal image block constructed from the sub-blocks corresponding to each component position is represented as:

[0055]

[0056] C st (u,v) is L st The spectrum of (x,y) in the hexadecimal discrete cosine transform domain can be expressed as:

[0057]

[0058] In the formula, s = 1, 2, 3, ..., M / r, t = 1, 2, 3, ..., M / r. C st The distribution of (u,v) in the hexadecimal space.

[0059] From the above formula, it can be seen that when u and v are both equal to 0, that is, when the hexadecimal image block L is equal to 0, the hexadecimal image block L is equal to 0. st The DC coefficient D of (x,y) after hexadecimal discrete cosine transform is... st It can be represented in the spatial domain as:

[0060]

[0061] Then L st The definition of a sixteen-ary (x, y) can be obtained as follows:

[0062]

[0063]

[0064] Other relevant parameters are:

[0065]

[0066] After the above calculations, the hexagonal image block L can be obtained. st The DC coefficient D of (x,y) after hexadecimal discrete cosine transform is... st The real and imaginary parts of , s = 1, 2, 3, ..., M / r, t = 1, 2, 3, ..., M / r.

[0067] Using a one-dimensional vector O st D represents st The magnitudes of the real and imaginary parts:

[0068]

[0069] The last passage through O st The coefficient eigenma matrix S is obtained by taking the L2 norm, using the following formula:

[0070] S(s,t)=||O st ||2

[0071] In the formula, ||·||2 is the symbol for calculating the L2 norm, s=1,2,3,……,M / r, t=1,2,3,……,M / r.

[0072] The binary robust feature matrix T is obtained by binarizing the matrix using the mean of the coefficient feature matrix S, as shown in the following formula:

[0073]

[0074] Utilizing low-frequency subband L i Constructing hexadecimal image blocks from the sub-blocks corresponding to each component after non-overlapping partitioning preserves the correlation between low-frequency components to the greatest extent. Simultaneously, directly calculating the DC coefficients of the hexadecimal image blocks after hexadecimal discrete cosine transform in the spatial domain effectively reduces computational cost and improves algorithm efficiency. Constructing a binary robust feature matrix based on the robustness of the relationship between the L2 norm and the overall mean of the DC coefficients of each hexadecimal image block after hexadecimal discrete cosine transform further enhances the algorithm's robustness.

[0075] S140. The binary robust feature matrix T is scrambled using a two-dimensional Sinusoidal-Chebyshev chaotic system and a Spatial Pyramid Matching chaotic system to obtain the binary image feature matrix J.

[0076] The binary robust feature matrix T is expanded into an N×N binary extended feature matrix R, which is achieved as follows:

[0077]

[0078] In the formula, N is the length and width of the original binary watermark image W, M is the length and width of a single slice in the original three-dimensional medical body data V, r is the length and width of the sub-block, and repmat(·) is the function for repeating the array copy.

[0079] A two-dimensional Sinusoidal-Chebyshev system is used to generate an N×N one-dimensional chaotic sequence, and the mean of this chaotic sequence is used for binarization to obtain a one-dimensional binary sequence Y. Then, the reshape function is used to reconstruct Y into a two-dimensional matrix, and XORed with the binary extended feature matrix R to obtain the encrypted binary feature matrix B. The implementation method is as follows:

[0080] B = xor(R, reshape(Y, [NN]))

[0081] In the formula, xor(·) is the function to perform the XOR operation, and reshape(·) is the function to reshape the array.

[0082] The two-dimensional Sinusoidal-Chebyshev system is represented by the following formula:

[0083]

[0084] Where a, b, c, and d are system control parameters, m i n i m is an intermediate variable in the iterative process of a two-dimensional Sinusoidal-Chebyshev chaotic system. i+1 n i+1 m i n i The next state is mod(·), where mod(·) is the remainder function.

[0085] An N×N one-dimensional chaotic sequence Z is generated using the Spatial Pyramid Matching chaotic system. The index vector of the one-dimensional chaotic sequence Z in descending order is used to scramble the encrypted binary feature matrix B, resulting in the binary image feature matrix J. The following method is used:

[0086]

[0087] In the formula, sort(·) is the sorting function, and index is the index vector obtained by sorting the one-dimensional chaotic sequence Z in descending order.

[0088] The Spatial Pyramid Matching chaotic system is represented by the following formula:

[0089]

[0090] In the formula, η and μ are system control parameters, h is a fixed parameter, q(t) is an intermediate variable in the iterative process of the Spatial Pyramid Matching chaotic system, and q(t+1) is the next state of q(t).

[0091] When using a two-dimensional Sinusoidal-Chebyshev chaotic system to obtain a one-dimensional binary sequence Y of length N×N, the system control parameters a, b, c, d and the initial system values ​​m0, n0 are used as key Key2. When using a Spatial PyramidMatching chaotic system to generate a one-dimensional chaotic sequence Z of length N×N, the fixed parameter h, the system control parameters η and μ and the initial system value q(0) are used as key Key3. Key2 and Key3 are saved.

[0092] Employing a two-dimensional Sinusoidal-Chebyshev chaotic system and a Spatial Pyramid Matching chaotic system to encrypt and scramble the binary robust feature matrix T improves both the security of the algorithm and its robustness through the randomness of the generated binary sequence. Expanding the binary robust feature matrix T using the repmat function resolves the size mismatch issue when the feature matrix XORs with the scrambled watermark, thus improving the overall stability of the algorithm.

[0093] S150, scramble the binary watermark image W corresponding to the original binary watermark image W. A The binary image feature matrix J is XORed with the binary image feature matrix J to obtain the certified zero-watermark image H. H is then saved to the watermark database of a third-party registration agency. The control parameters and process parameters used in the binary zero-watermark embedding process are saved as keys, thus completing the zero-watermark embedding process.

[0094] H = xor(W) A ,J)

[0095] In the formula, xor(·) is the function that performs the XOR operation.

[0096] Through the above steps, the final certified zero-watermark image H and the relevant keys Key1, Key2 and Key3 used in the zero-watermark embedding process are obtained, which are used to determine the copyright ownership of the three-dimensional medical body data to be certified.

[0097] Furthermore, this embodiment also provides an extraction method for zero-watermarking of three-dimensional medical volume data based on hexadecimal discrete cosine transform, such as... Figure 2 As shown, the specific steps include the following:

[0098] S210. Slice and group the three-dimensional medical volume data V′ of size M×M×P that is to be authenticated. Then, group each medical image slice group U. i All samples underwent Tucker decomposition, and the first energy feature map E after each Tucker decomposition was extracted. i ′, for E i Each component in the ' ' model undergoes a k-level two-dimensional non-subsampled discrete wavelet transform to obtain the corresponding low-frequency subband L. i ′.

[0099] The three-dimensional medical volume data V′ of size M×M×P, which is to be authenticated, is sliced ​​and divided into 15 medical image slice groups, denoted as U. i Let i = 1, 2, 3, ..., 15, and implement it in the following way:

[0100]

[0101] Will U i Converting the medical image slices to tensor form and performing Tucker decomposition yields the medical image slice group U. i ′ can be represented as:

[0102]

[0103] Let tensor For tensor U i Extract the subtensors X from each group of subtensors. i The first sub-image of ′, namely the medical image slice group U i The first energy characteristic diagram after Tucker decomposition is denoted as E. i ′,i=1,2,3,……,15。

[0104] For E i Perform a k-level two-dimensional non-subsampled discrete wavelet transform to obtain E. i The low-frequency subband L corresponding to each component in ′ i ′,i=1,2,3,……,15。

[0105] S220, for low-frequency subband L i Perform non-overlapping sub-block partitioning, and then utilize the partitioned L i The sub-blocks corresponding to each component in the image are used to construct a hexagonal image block, and the DC coefficient D′ of the image block after hexagonal discrete cosine transform is calculated in the spatial domain. st , D′st The magnitudes of the real and imaginary parts are represented by a one-dimensional vector O′. st This means that O′ is used finally. st The L2 norm yields the binary robust characteristic matrix T′.

[0106] For L i Perform non-overlapping sub-block partitioning of size r×r, and then use L... i The sub-blocks corresponding to each component in the image are used to construct a hexagonal image block, and the DC coefficient D′ of the image block after hexagonal discrete cosine transform is calculated in the spatial domain. st , D′ st The magnitudes of the real and imaginary parts are represented by a one-dimensional vector O′. st This means that O′ is used finally. st The coefficient eigenvalue matrix S′ is obtained by taking the L2 norm, using the following formula:

[0107] S'(s,t)=||O' st ||2

[0108] The matrix is ​​binarized using the mean of the coefficient feature matrix S′ to obtain the binary robust feature matrix T′, which is achieved using the following formula:

[0109]

[0110] S230. The binary robust feature matrix T′ is scrambled using a two-dimensional Sinusoidal-Chebyshev chaotic system and a Spatial Pyramid Matching chaotic system to obtain the binary image feature matrix J′.

[0111] The binary robust feature matrix T′ is expanded into an N×N binary extended feature matrix R′, which is achieved as follows:

[0112]

[0113] Using key Key2, a one-dimensional chaotic sequence of length N×N is generated using a two-dimensional Sinusoidal-Chebyshev system. The mean of this chaotic sequence is then used for binarization to obtain a one-dimensional binary sequence Y. The reshape function is then used to reconstruct Y into a two-dimensional matrix, and an XOR operation is performed with the binary extended feature matrix R′ to obtain the encrypted binary feature matrix B′. The implementation method is as follows:

[0114] B' = xor(R', reshape(Y, [NN]))

[0115] Using key Key3, a one-dimensional chaotic sequence Z is generated using the Spatial Pyramid Matching chaotic system. The index vectors of the one-dimensional chaotic sequence Z in descending order are then used to scramble the encrypted binary feature matrix B′, yielding the binary image feature matrix J′, as shown in the following formula:

[0116]

[0117] S240. Retrieve the certified zero-watermark image H stored in the watermark database of a third-party registration agency, and perform an XOR operation with the binary image feature matrix J′ to obtain the binary watermark image W to be descrambled. A ′.

[0118] W A = xor(H,J)

[0119] S250. Using the Fibonacci Q inverse matrix as the descrambling matrix, the binary watermark image W to be descrambled is treated as the descrambling matrix. A By performing an unscramble operation, the final watermark information W′ can be extracted. Finally, the copyright ownership of the three-dimensional medical body data V′ to be authenticated is determined based on the content information displayed by W′.

[0120] The binary watermark image W, which uses the Fibonacci Q inverse matrix as the descrambling matrix for the descrambling operation, is then subjected to the descrambling operation. A By performing a descrambling operation, the final watermark information W′ can be extracted. Finally, the copyright ownership of the 3D medical body data V′ to be authenticated is determined based on the content information displayed by W′, including:

[0121] Using key Key1, the binary watermark image W to be descrambled is treated with the Fibonacci Q inverse matrix as the descrambling matrix. A Performing a descrambling operation yields the descrambled watermark image, from which the final watermark image W′ can be extracted. The implementation method is as follows:

[0122]

[0123] In the formula, Q -p Let x1 be the p-order Fibonacci Q-inverse matrix, i.e., the descrambling matrix, and y1 be the binary watermark image W to be descrambled. A x1′ and y1′ are the position coordinates of the pixels in the final watermark image W′ extracted after descrambling, and x1′ and y1′ are the position coordinates of the pixels in the final watermark image W′ after descrambling.

[0124] The p-order Fibonacci Q-inverse matrix is ​​achieved by the following formula:

[0125]

[0126] The copyright ownership of the three-dimensional medical body data V′ to be authenticated is determined based on the content displayed in the final extracted watermark image W′.

[0127] This embodiment provides a zero-watermarking method for 3D medical volume data based on hexadecimal discrete cosine transform. The method involves grouping and processing the 3D medical volume data into slices and performing Tucker decomposition. The first energy feature map of each group after Tucker decomposition is then extracted, maximizing the preservation of correlation between slices and improving the overall robustness of the algorithm. Next, a two-dimensional non-subsampled discrete wavelet transform is performed on the first energy feature map of each group after Tucker decomposition to obtain corresponding low-frequency sub-bands of the same size, which helps improve the watermark embedding capacity. Hexadecimal image blocks are constructed using the sub-blocks corresponding to each component after non-overlapping division of the low-frequency sub-bands. The DC coefficients of these image blocks after hexadecimal discrete cosine transform are directly calculated in the spatial domain, resulting in a binary robust feature matrix. This not only preserves the correlation between low-frequency components but also effectively reduces computational costs. Simultaneously, the Fibonacci Q matrix and a dual chaotic system are used to scramble and encrypt the original watermark image and the binary robust feature matrix, respectively, enhancing the overall security of the algorithm. It exhibits excellent robustness against various image processing attacks, such as noise attacks, filtering attacks, JPEG compression attacks, cropping attacks, rotation attacks, scaling attacks, and translation attacks.

[0128] The following specific examples illustrate the process and effects of a zero-watermarking method for three-dimensional medical volume data based on hexadecimal discrete cosine transform provided by this invention.

[0129] To verify the effectiveness of this invention, simulation experiments were conducted using MRI and CT three-dimensional brain medical volume data, each with a size of 128×128×99. A schematic diagram of a single slice of the MRI and CT three-dimensional brain medical volume data is shown below. Figure 3 As shown. The original binary watermark image is a 64×64 "CAUC" image, see... Figure 4 .

[0130] This invention uses normalized cross-correlation (NC) to evaluate the similarity between the final extracted binary watermark image W′ and the original binary watermark image W. NC is defined as follows:

[0131]

[0132] In the formula, N is the size of the original binary watermark image, W(i,j) represents the pixel value of the original binary watermark image at point (i,j), and W′(i,j) represents the pixel value of the finally extracted binary watermark image at point (i,j). The closer the NC value is to 1, the more similar the finally extracted binary watermark image W′ is to the original binary watermark image W, indicating that the method is more robust.

[0133] Figure 5 This is a test NC value table of MRI and CT brain 3D medical volume data under several non-geometric attacks in a zero-watermarking method for 3D medical volume data based on hexadecimal discrete cosine transform provided in an embodiment of the present invention. Figure 5 It can be seen that the NC value of watermark extraction from MRI and CT brain 3D medical volume data is close to 1 when facing noise, filtering and compression attacks, indicating that the method has a very high ability to resist non-geometric attacks and has strong robustness.

[0134] Figure 6 This invention provides a test NC value table for MRI and CT brain 3D medical volume data under geometric attacks such as shearing, scaling, translation, and rotation, using a 3D medical volume data zero-watermarking method based on hexadecimal discrete cosine transform. From... Figure 6 As can be seen, the zero-watermarking method for 3D medical volume data provided in this embodiment exhibits good robustness against common geometric attacks, particularly for MRI and CT brain 3D medical volume data.

[0135] The specific embodiments described above do not constitute a limitation on the scope of protection of this invention. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this invention should be included within the scope of protection of this invention.

Claims

1. A zero-watermarking method for three-dimensional medical volume data based on hexadecimal discrete cosine transform, characterized in that, include: Embed zero watermarks in the original 3D medical body data and extract zero watermarks from the 3D medical body data to be authenticated; The process of embedding zero watermarks into the original three-dimensional medical volume data includes: The Fibonacci Q matrix is ​​used as the scrambling matrix to scramble the original N×N binary watermark image W, resulting in a scrambled binary watermark image W. A ; The original three-dimensional medical volume data V of size M×M×P is sliced ​​and grouped, and the processed medical image slice group is denoted as U. i , i = 1, 2, 3, ..., 15; for each medical image slice group U i All samples underwent Tucker decomposition, and the first energy feature map after each Tucker decomposition was extracted and labeled as E. i , i=1,2,3,...,15; for E i Each component undergoes a k-level two-dimensional non-subsampled discrete wavelet transform to obtain the corresponding low-frequency subband L. i , i = 1, 2, 3, ..., 15; For low-frequency subband L i Each component is divided into non-overlapping sub-blocks, and then the L-shaped components are used. i Construct hexagonal image blocks from the corresponding sub-blocks in each component and calculate the DC coefficient D of the image block after hexagonal discrete cosine transform in the spatial domain. st D st The magnitudes of the real and imaginary parts are represented by a one-dimensional vector O. st This indicates that O was finally used. st The L2 norm yields the binary robust characteristic matrix T; The binary robust feature matrix T is scrambled using a two-dimensional Sinusoidal-Chebyshev chaotic system and a Spatial Pyramid Matching chaotic system to obtain the binary image feature matrix J. The scrambled binary watermark image W corresponding to the original binary watermark image W. A The binary image feature matrix J is XORed with the binary image feature matrix J to obtain the certified zero-watermark image H. H is saved to the watermark database of a third-party registration agency, and the control parameters and process parameters used in the binary zero-watermark embedding process are saved as keys. The zero-watermark extraction from the three-dimensional medical body data to be authenticated includes: The three-dimensional medical volume data V′ of size M×M×P, which is to be authenticated, is sliced ​​and grouped. The processed medical image slice group is denoted as U. i ′, i = 1, 2, 3, ..., 15; for each medical image slice group U i All samples were subjected to Tucker decomposition, and the first energy feature map after each Tucker decomposition was extracted and labeled as E. i ′, i=1,2,3,...,15; for E i Each component in the vector is subjected to a k-level two-dimensional non-subsampled discrete wavelet transform to obtain the corresponding low-frequency subband L. i ′,i=1,2,3,……,15; For low-frequency subband L i Each component in ' is divided into non-overlapping sub-blocks, and then the L-shaped sub-blocks are used. i Construct hexagonal image blocks from the corresponding sub-blocks in each component of the image block and calculate the DC coefficient D of the image block after hexagonal discrete cosine transform in the spatial domain. st ′, will D st The magnitudes of the real and imaginary parts of ′ are represented by a one-dimensional vector O. st ′ indicates that O is used last. st The L2 norm of T' yields the binary robust characteristic matrix T'; The binary robust feature matrix T′ is scrambled using a two-dimensional Sinusoidal-Chebyshev chaotic system and a Spatial Pyramid Matching chaotic system to obtain the binary image feature matrix J′. Retrieve the certified zero-watermark image H stored in the watermark database of a third-party registration agency, and XOR it with the binary image feature matrix J′ to obtain the binary watermark image W to be descrambled. A ′; The binary watermark image W is treated as a descrambling matrix using the Fibonacci Q inverse matrix for descrambling operations. A By performing an unscramble operation, the final watermark information W′ can be extracted. Finally, the copyright ownership of the three-dimensional medical body data V′ to be authenticated is determined based on the content information displayed by W′.

2. The method according to claim 1, characterized in that, The original N×N binary watermark image W is scrambled using the Fibonacci Q matrix as the scrambling matrix to obtain the scrambled binary watermark image W. A ,include: The original binary watermark image W is scrambled using the Fibonacci Q matrix as the scrambling matrix, expressed by the following formula: In the formula, Q p Let N be the p-order Fibonacci Q-matrix, i.e., the scrambling matrix. Let N be the length and width of the original binary watermark image W, x and y be the pixel coordinates in the original binary watermark image W before scrambling, and x′ and y′ be the scrambled binary watermark image W after scrambling. A The position coordinates of the middle pixel, mod(·) is the remainder function; The p-order Fibonacci Q matrix is ​​achieved by the following formula: In the formula, F p+1 F p F p-1 All are the terms in the Fibonacci sequence. The order p is treated as the key Key1 and stored.

3. The method according to claim 1, characterized in that, The original three-dimensional medical volume data V of size M×M×P is sliced ​​and grouped, and each medical image slice group U is... i All samples underwent Tucker decomposition, and the first energy feature map E of each Tucker decomposition was extracted. i For E i Each component undergoes a k-level two-dimensional non-subsampled discrete wavelet transform to obtain the corresponding low-frequency subband L. i ,include: The original three-dimensional medical volume data V of size M×M×P is sliced ​​and divided into 15 medical image slice groups, denoted as U. i For i = 1, 2, 3, ..., 15, the following method can be used: In the formula, P is the total number of slices after processing the three-dimensional medical body data slices, floor(·) is the function to round down to the nearest integer, and mod(·) is the function to take the remainder; Will U i Converting to tensor form and performing Tucker decomposition, the medical image slice group U i Represented as: U i =G i ×1V i (1) ×2V i (2) ×3V i (3) In the formula, i=1,2,3,...,15, U i G represents the tensor representation of each medical image slice group. i V is the core tensor in each group. i (1) V i (2) V i (3) These are the factor matrices of each group of tensors in different directions; Let tensor X i =G i ×1V i (1) ×2V i (2) For tensor U i Extract the subtensors from each group of subtensors X. i The first sub-image, namely the medical image slice group U i The first energy characteristic map after Tucker decomposition is denoted as E. i , i = 1, 2, 3, ..., 15; For E i Perform a k-level two-dimensional non-subsampled discrete wavelet transform to obtain E i The low-frequency subband L corresponding to each component i , i = 1, 2, 3, ..., 15.

4. The method according to claim 1, characterized in that, The low-frequency subband L i Perform non-overlapping sub-block partitioning, and then utilize the L-shaped sub-blocks. i Construct hexagonal image blocks from the sub-blocks corresponding to each component position and calculate the DC coefficient D of the image block after hexagonal discrete cosine transform in the spatial domain. st D st The magnitudes of the real and imaginary parts are represented by a one-dimensional vector O. st This indicates that O was finally used. st The L2 norm yields the binary robust characteristic matrix T, which includes: For L i Perform non-overlapping sub-block partitioning of size r×r, and then utilize the partitioned L... i Construct hexagonal image blocks from the sub-blocks corresponding to each component position and calculate the DC coefficient D of the image block after hexagonal discrete cosine transform in the spatial domain. st D st The magnitudes of the real and imaginary parts are represented by a one-dimensional vector O. st This indicates that O was finally used. st The coefficient eigenma matrix S is obtained by taking the L2 norm, using the following formula: S(s,t)=||O st ||2 In the formula, ||·||2 is the symbol for calculating the L2 norm, s=1,2,3,……,M / r, t=1,2,3,……,M / r; The binary robust feature matrix T is obtained by binarizing the matrix using the mean of the coefficient feature matrix S, as shown in the following formula:

5. The method according to claim 1, characterized in that, The binary robust feature matrix T is scrambled using a two-dimensional Sinusoidal-Chebyshev chaotic system and a Spatial Pyramid Matching chaotic system to obtain a binary image feature matrix J, including: The binary robust feature matrix T is expanded into an N×N binary extended feature matrix R, which is achieved as follows: In the formula, N is the length and width of the original binary watermark image W, M is the length and width of a single slice in the original three-dimensional medical body data V, r is the length and width of the sub-block, and repmat(·) is the function for repeating the array copy. A two-dimensional Sinusoidal-Chebyshev system is used to generate an N×N one-dimensional chaotic sequence, and the mean of this chaotic sequence is used for binarization to obtain a one-dimensional binary sequence Y. Then, the reshape function is used to reconstruct Y into a two-dimensional matrix, and XORed with the binary extended feature matrix R to obtain the encrypted binary feature matrix B. The implementation method is as follows: B = xor(R, reshape(Y, [NN])) In the formula, xor(·) is the function to perform the XOR operation, and reshape(·) is the function to reshape the array. The two-dimensional Sinusoidal-Chebyshev system is represented by the following formula: In the formula, a, b, c, and d are system control parameters, and m i n i m is an intermediate variable in the iterative process of a two-dimensional Sinusoidal-Chebyshev chaotic system. i+1 n i+1 m i n i The next state, mod(·) is the remainder function; An N×N one-dimensional chaotic sequence Z is generated using the Spatial Pyramid Matching chaotic system. The index vector of the one-dimensional chaotic sequence Z in descending order is used to scramble the encrypted binary feature matrix B, resulting in the binary image feature matrix J. The following method is used: In the formula, sort(·) is the sorting function, and index is the index vector obtained by sorting the one-dimensional chaotic sequence Z in descending order; The Spatial PyramidMatching chaotic system is represented by the following formula: In the formula, η and μ are system control parameters, h is a fixed parameter, q(t) is an intermediate variable in the iterative process of the Spatial Pyramid Matching chaotic system, and q(t+1) is the next state of q(t). When using a two-dimensional Sinusoidal-Chebyshev chaotic system to obtain a one-dimensional binary sequence Y of length N×N, the system control parameters a, b, c, d and the initial system values ​​m0, n0 are used as key Key2. When using a Spatial Pyramid Matching chaotic system to generate a one-dimensional chaotic sequence Z of length N×N, the fixed parameter h, the system control parameters η and μ and the initial system value q(0) are used as key Key3. Key2 and Key3 are saved.

6. The method according to claim 1, characterized in that, The process involves slicing and grouping the three-dimensional medical volume data V′ of size M×M×P that is to be authenticated, and then grouping each medical image slice group U. i All samples underwent Tucker decomposition, and the first energy feature map E after each Tucker decomposition was extracted. i ′, for E i Each component in the ' ' model undergoes a k-level two-dimensional non-subsampled discrete wavelet transform to obtain the corresponding low-frequency subband L. i ',include: The three-dimensional medical volume data V′ of size M×M×P, which is to be authenticated, is sliced ​​and divided into 15 medical image slice groups, denoted as U. i Let i = 1, 2, 3, ..., 15, and implement it in the following way: Will U i Converting the medical image slices to tensor form and performing Tucker decomposition yields the medical image slice group U. i ′ is represented as: Let tensor For tensor U i Extract the subtensors X from each group of subtensors. i The first sub-image of ′, namely the medical image slice group U i The first energy characteristic diagram after Tucker decomposition is denoted as E. i ′,i=1,2,3,……,15; For E i Perform a k-level two-dimensional non-subsampled discrete wavelet transform to obtain E. i The low-frequency subband L corresponding to each component in ′ i ′,i=1,2,3,……,15。 7. The method according to claim 1, characterized in that, The low-frequency subband L i Perform non-overlapping sub-block partitioning, and then utilize the partitioned L i The sub-blocks corresponding to each component in the image are used to construct a hexagonal image block, and the DC coefficient D′ of the image block after hexagonal discrete cosine transform is calculated in the spatial domain. st , D′ st The magnitudes of the real and imaginary parts are represented by a one-dimensional vector O′. st This means that O′ is used finally. st The L2 norm yields the binary robust characteristic matrix T′, which includes: For L i Perform non-overlapping sub-block partitioning of size r×r, and then use L... i The sub-blocks corresponding to each component in the image are used to construct a hexagonal image block, and the DC coefficient D′ of the image block after hexagonal discrete cosine transform is calculated in the spatial domain. st , D′ st The magnitudes of the real and imaginary parts are represented by a one-dimensional vector O′. st This means that O′ is used finally. st The coefficient eigenvalue matrix S′ is obtained by taking the L2 norm, using the following formula: S'(s,t)=||O' st ||2 The matrix is ​​binarized using the mean of the coefficient feature matrix S′ to obtain the binary robust feature matrix T′, which is achieved using the following formula:

8. The method according to claim 1, characterized in that, The binary watermark image W to be descrambled A By performing descrambling, the final watermark image W′ can be extracted, including: The binary robust feature matrix T′ is expanded into an N×N binary extended feature matrix R′, which is achieved as follows: Using key Key2, a one-dimensional chaotic sequence of length N×N is generated using a two-dimensional Sinusoidal-Chebyshev system. The mean of this chaotic sequence is then used for binarization to obtain a one-dimensional binary sequence Y. The reshape function is then used to reconstruct Y into a two-dimensional matrix, and an XOR operation is performed with the binary extended feature matrix R′ to obtain the encrypted binary feature matrix B′. The implementation method is as follows: B' = xor(R', reshape(Y, [NN])) Using key Key3, a one-dimensional chaotic sequence Z is generated using the Spatial Pyramid Matching chaotic system. The index vectors of the one-dimensional chaotic sequence Z in descending order are then used to scramble the encrypted binary feature matrix B′, yielding the binary image feature matrix J′, as shown in the following formula: Extract the certified zero-watermark image H stored in the watermark database of a third-party registration agency, and XOR it with the binary image feature matrix J′ to obtain the binary watermark image W to be descrambled. A ′; Using key Key1, the binary watermark image W to be descrambled is treated with the Fibonacci Q inverse matrix as the descrambling matrix. A Perform a descrambling operation to obtain the descrambled watermark image, i.e., extract the final watermark image W′. The implementation method is as follows: In the formula, Q -p Let x1 be the p-order Fibonacci Q-inverse matrix, i.e., the descrambling matrix, and y1 be the binary watermark image W to be descrambled. A x1′ and y1′ are the position coordinates of the pixels in the final watermark image W′ extracted after descrambling; The p-order Fibonacci Q-inverse matrix is ​​achieved by the following formula: The copyright ownership of the three-dimensional medical body data V′ to be authenticated is determined based on the content displayed in the final extracted watermark image W′.

Images