Medical image watermarking method based on censurestar-teblid-dct

Abstract

The application discloses a medical image watermarking method based on CenSurE-STAR-TEBLID-DCT, which comprises the following steps: extracting key points of a medical image by using a CenSurE-STAR algorithm; describing the key points of the medical image by using a TEBLID algorithm to generate a feature descriptor matrix of the medical image; performing DCT transformation on the feature descriptor matrix of the medical image to obtain a feature vector of the medical image; and performing XOR operation on the feature vector of the medical image and chaotic scrambling watermark bit by bit to embed the watermark information into the medical image. Thus, the advantages of fast CenSurE-STAR calculation speed, scale invariance and fast TEBLID calculation speed and strong robustness are taken into account, the method has strong robustness and invisibility, can effectively protect the security of original medical image information, and can avoid patient privacy leakage to a certain extent.

Description

Technical Field

[0001] This invention relates to the field of image watermarking technology, and in particular to a medical image watermarking method based on CenSurE-STAR-TEBLID-DCT. Background Technology

[0002] With the continuous advancement of medical technology, medical images are playing an increasingly important role in diagnosis and treatment. However, the protection and management of medical images have also brought about new challenges. The leakage and alteration of medical images not only compromise patient privacy but also seriously affect treatment outcomes. Therefore, protecting the integrity of medical images and preventing their copyright misappropriation is of paramount importance.

[0003] Digital image watermarking is a widely used technology that embeds a watermark into an image to protect its copyright and verify the integrity of medical images. Medical image watermarking technology needs to be highly robust and highly visible to meet the specific needs of medical images in use. However, traditional digital image watermarking methods cannot effectively protect the medical image itself, are easily altered by the data content, and have low robustness. Summary of the Invention

[0004] In view of this, the purpose of this invention is to provide a medical image watermarking method based on CenSurE-STAR-TEBLID-DCT, which can effectively protect the security of original medical image information and has strong robustness and invisibility. The specific solution is as follows:

[0005] A medical image watermarking method based on CenSurE-STAR-TEBLID-DCT includes:

[0006] The CenSurE-STAR algorithm was used to extract key points from medical images;

[0007] The TEBLID algorithm is used to describe the key points of the medical image, generating a feature descriptor matrix of the medical image;

[0008] Perform DCT transformation on the feature descriptor matrix of the medical image to obtain the feature vector of the medical image;

[0009] The feature vector of the medical image is XORed bit by bit with the chaotic scrambled watermark to embed the watermark information into the medical image.

[0010] Preferably, in the medical image watermarking method based on CenSurE-STAR-TEBLID-DCT provided in the embodiments of the present invention, the CenSurE-STAR algorithm is used to extract key points of the medical image, including:

[0011] The medical image is processed by approximate Gaussian difference, and the local maxima corresponding to the medical image are detected as feature points using a non-maximum suppression algorithm.

[0012] The Harris edge filter is used to remove unstable feature points on the edges of the medical image, and the remaining feature points are used as the key points of the medical image.

[0013] Preferably, in the above-described medical image watermarking method based on CenSurE-STAR-TEBLID-DCT provided in this embodiment of the invention, the TEBLID algorithm is used to describe the key points of the medical image and generate the feature descriptor matrix of the medical image, including:

[0014] Use binary weak descriptors as decision trees on a predefined feature extraction function;

[0015] The decision tree is used to process the key points of the medical image, and a greedy algorithm is used to minimize the selection of the corresponding target weak descriptor;

[0016] Based on the selected target weak descriptors, a feature descriptor matrix for the medical image is generated.

[0017] Preferably, in the above-described medical image watermarking method based on CenSurE-STAR-TEBLID-DCT provided in this embodiment of the invention, performing DCT transformation on the feature descriptor matrix of the medical image to obtain the feature vector of the medical image includes:

[0018] The feature descriptor matrix of the medical image is subjected to DCT transformation to obtain the coefficient matrix of the medical image;

[0019] Take a matrix of a predetermined size from the coefficient matrix of the medical image to form a new matrix of the medical image;

[0020] The new matrix of the medical image is symbolically transformed using a hash function to obtain the feature vector of the medical image.

[0021] Preferably, in the above-described medical image watermarking method based on CenSurE-STAR-TEBLID-DCT provided in this embodiment of the invention, before performing a bitwise XOR operation between the feature vector of the medical image and the chaotic scrambled watermark, the method further includes:

[0022] Obtain a binary chaotic sequence;

[0023] The original watermark is binarized to obtain a binary watermark image;

[0024] The binary chaotic sequence is used to scramble the binary watermark image to obtain a chaotic scrambled watermark.

[0025] Preferably, in the above-described medical image watermarking method based on CenSurE-STAR-TEBLID-DCT provided in the embodiments of the present invention, while embedding the watermark information into the medical image, it further includes:

[0026] Generate a logical key;

[0027] Feature extraction is performed on the medical image to be tested to obtain the feature vector of the medical image to be tested;

[0028] The feature vector of the medical image to be tested and the logical key are XORed to extract the encrypted watermark.

[0029] Preferably, in the medical image watermarking method based on CenSurE-STAR-TEBLID-DCT provided in the embodiments of the present invention, feature extraction is performed on the medical image to be tested to obtain the feature vector of the medical image to be tested, including:

[0030] The CenSurE-STAR algorithm was used to extract key points from the medical image under test.

[0031] The TEBLID algorithm is used to describe the key points of the medical image under test, and a feature descriptor matrix of the medical image under test is generated.

[0032] The feature descriptor matrix of the medical image under test is subjected to DCT transformation to obtain the feature vector of the medical image under test.

[0033] Preferably, in the medical image watermarking method based on CenSurE-STAR-TEBLID-DCT provided in the embodiments of the present invention, the CenSurE-STAR algorithm is used to extract key points of the medical image to be tested, including:

[0034] The medical image to be tested is subjected to approximate Gaussian difference processing, and the local maxima corresponding to the medical image to be tested are detected as feature points using a non-maximum suppression algorithm;

[0035] The Harris edge filter is used to remove unstable feature points on the edges of the medical image under test, and the remaining feature points are used as the key points of the medical image under test.

[0036] Preferably, in the above-described medical image watermarking method based on CenSurE-STAR-TEBLID-DCT provided in this embodiment of the invention, the TEBLID algorithm is used to describe the key points of the medical image to be tested, generating a feature descriptor matrix of the medical image to be tested, including:

[0037] The decision tree is used to process the key points of the medical image under test, and a greedy algorithm is used to minimize and select the corresponding target weak descriptor.

[0038] Based on the selected target weak descriptors, a feature descriptor matrix is ​​generated for the medical image under test.

[0039] Preferably, in the above-described medical image watermarking method based on CenSurE-STAR-TEBLID-DCT provided in the embodiments of the present invention, after extracting the encrypted watermark, the method further includes:

[0040] The extracted encrypted watermark is XORed with the binary chaotic sequence to obtain the restored watermark.

[0041] As can be seen from the above technical solution, the medical image watermarking method based on CenSurE-STAR-TEBLID-DCT provided by the present invention includes: extracting key points of the medical image using the CenSurE-STAR algorithm; describing the key points of the medical image using the TEBLID algorithm to generate a feature descriptor matrix of the medical image; performing DCT transformation on the feature descriptor matrix of the medical image to obtain the feature vector of the medical image; and performing a bitwise XOR operation between the feature vector of the medical image and the chaotic scrambled watermark to embed the watermark information into the medical image.

[0042] The medical image watermarking method provided by this invention extracts features from medical images based on CenSurE-STAR, TEBLID, and DCT. It combines the advantages of CenSurE-STAR's fast computation speed and scale invariance with TEBLID's fast computation speed and robustness. By combining the extracted feature vectors, chaotic encryption, and watermarking technology, it solves the problem of modification of the original data caused by traditional watermarking embedding technology. It has strong robustness and invisibility, effectively protects the security of the original medical image information, and avoids the leakage of patient privacy to a certain extent. Attached Figure Description

[0043] To more clearly illustrate the technical solutions in the embodiments of the present invention or related technologies, the drawings used in the description of the embodiments or related technologies will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.

[0044] Figure 1 A flowchart illustrating a medical image watermarking method based on CenSurE-STAR-TEBLID-DCT provided in an embodiment of the present invention;

[0045] Figure 2 The original medical image provided for the embodiments of the present invention;

[0046] Figure 3 A first medical image of the lung provided for an embodiment of the present invention;

[0047] Figure 4 A second lung medical image provided for an embodiment of the present invention;

[0048] Figure 5 Medical images of the femur provided for embodiments of the present invention;

[0049] Figure 6 Medical images of the neck provided for embodiments of the present invention;

[0050] Figure 7 Medical images of the liver provided in embodiments of the present invention;

[0051] Figure 8 Medical images of the tibial joint provided for embodiments of the present invention;

[0052] Figure 9 Medical images of the intestine provided in embodiments of the present invention;

[0053] Figure 10 The original watermark image provided in the embodiments of the present invention;

[0054] Figure 11 The encrypted watermark image provided in this embodiment of the invention;

[0055] Figure 12 This is a watermark extracted without interference, as provided in an embodiment of the present invention.

[0056] Figure 13 A medical image with JPEG compression quality of 10% provided in an embodiment of the present invention;

[0057] Figure 14 The watermark extracted during JPEG compression with a compression quality of 10% is provided in an embodiment of the present invention.

[0058] Figure 15 The medical image provided in this embodiment of the invention has a window size of [3*3] and is filtered 5 times by median.

[0059] Figure 16 The watermark extracted in this embodiment of the invention has a window size of [3*3] and is obtained after 5 rounds of median filtering;

[0060] Figure 17 A medical image rotated 20° counterclockwise, provided as an embodiment of the present invention;

[0061] Figure 18The watermark extracted by rotating counterclockwise by 20° is provided in an embodiment of the present invention;

[0062] Figure 19 A medical image rotated 45° counterclockwise as provided in an embodiment of the present invention;

[0063] Figure 20 The watermark extracted by rotating counterclockwise by 45° is provided in an embodiment of the present invention;

[0064] Figure 21 The image provided in this embodiment of the invention is a scaled medical image with a scaling factor of 1.3.

[0065] Figure 22 The watermark extracted when the scaling factor is 1.3, as provided in this embodiment of the invention;

[0066] Figure 23 The image provided in this embodiment of the invention is a scaled medical image with a scaling factor of 1.5.

[0067] Figure 24 The watermark extracted when the scaling factor is 1.5, as provided in this embodiment of the invention;

[0068] Figure 25 A medical image provided by an embodiment of the present invention after being horizontally shifted 20% to the left;

[0069] Figure 26 The watermark extracted after a 20% horizontal leftward shift is provided in this embodiment of the invention;

[0070] Figure 27 A medical image provided by an embodiment of the present invention after vertical upward shift of 20%;

[0071] Figure 28 The watermark extracted after vertical upward shift by 20% is provided in an embodiment of the present invention;

[0072] Figure 29 This is a medical image provided by an embodiment of the present invention after being sheared by 30% along the Y-axis;

[0073] Figure 30 The watermark extracted after cutting 30% along the Y-axis is provided in an embodiment of the present invention;

[0074] Figure 31 This is a medical image provided by an embodiment of the present invention after being sheared by 30% along the X-axis;

[0075] Figure 32 The watermark extracted after cutting 30% along the X-axis is provided in an embodiment of the present invention. Detailed Implementation

[0076] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0077] This invention provides a medical image watermarking method based on CenSurE-STAR-TEBLID-DCT, such as... Figure 1 As shown, it includes the following steps:

[0078] S101. Use the CenSurE-STAR algorithm to extract key points from medical images;

[0079] It should be noted that CenSurE-STAR is a feature detection algorithm based on the CenSurE (Center Surround Extrema) algorithm. It is mainly used in various computer vision applications, such as robot vision, autonomous driving, and intelligent security, and has good robustness and stability.

[0080] The aforementioned medical image is used as the original medical image, denoted as I(i,j). I(i,j) represents the pixel grayscale value of the original medical image. This invention uses the CenSurE-STAR algorithm to extract the key points of the medical image I(i,j).

[0081] S102. Use the TEBLID algorithm to describe the key points of the medical image and generate the feature descriptor matrix of the medical image.

[0082] It's important to note that TEBLID is a high-efficiency binary local image descriptor algorithm based on triplet loss. Its full name is "Triplet Encoded Binary Local Image Descriptor," and it's an integration of the BAD (Box Average Difference) algorithm into OpenCV. The TEBLID algorithm is a fast binary descriptor based on the Efficient-BRIEF concept, optimizing the descriptor layout through learning. It uses triplet loss, hard negative sampling, and anchor swapping to improve image matching results. Its basic idea is to use a greedy algorithm to select an irrelevant set of pixel differences to distinguish similar and different image patches. TEBLID feature descriptors are efficient, robust, compact, scale-invariant, and rotation-invariant, making them suitable for applications such as image matching and object recognition.

[0083] This invention utilizes the TEBLID algorithm to describe the key points of a medical image I(i,j) and generate a feature descriptor matrix descriptps for the medical image I(i,j).

[0084] S103. Perform DCT transformation on the feature descriptor matrix of the medical image to obtain the feature vector of the medical image;

[0085] Specifically, by performing a DCT transformation on the feature descriptor matrix descriptps of the medical image I(i,j), the feature vector V(i,j) of the medical image I(i,j) can be obtained.

[0086] S104. Perform a bitwise XOR operation between the feature vector of the medical image and the chaotic scrambled watermark to embed the watermark information into the medical image.

[0087] It should be noted that the aforementioned chaotic scrambling watermark is a watermark obtained by chaotic scrambling the original watermark. In step S104, this invention can select a meaningful binary text image as the original watermark embedded in the medical image I(i,j), denoted as W = {w(i,j)|w(i,j) = 0,1; 1≤i≤M1, 1≤j≤M2}, where M1 and M2 are the length and width dimensions of the original watermark, respectively. W(i,j) represents the pixel grayscale value of the original watermark. That is, after chaotic scrambling the original watermark W(i,j), a chaotic scrambling watermark BW(i,j) can be obtained.

[0088] Specifically, by performing a bitwise XOR operation on the feature vector V(i,j) of the medical image I(i,j) and the chaotic scrambled watermark BW(i,j), the watermark information can be embedded into the medical image I(i,j).

[0089] In the medical image watermarking method based on CenSurE-STAR-TEBLID-DCT provided in the embodiments of the present invention, features of medical images are extracted based on CenSurE-STAR, TEBLID, and DCT. This method combines the advantages of CenSurE-STAR (fast computation speed and scale invariance) and TEBLID (fast computation speed and robustness). By combining the extracted feature vectors, chaotic encryption, and watermarking technology, the method solves the problem of modification of the original data caused by traditional watermarking embedding technology. It has strong robustness and invisibility, effectively protects the security of the original medical image information, and avoids the leakage of patient privacy to a certain extent.

[0090] In a specific implementation, in the medical image watermarking method provided in the embodiments of the present invention, step S101 uses the CenSurE-STAR algorithm to extract key points of the medical image, which may specifically include: first, performing approximate Gaussian difference processing on the medical image I(i,j), and using a non-maximum suppression algorithm to detect the local maxima corresponding to the medical image I(i,j) as feature points; then, using a Harris edge filter to remove unstable feature points on the edge of the medical image I(i,j), and the remaining feature points are used as key points of the medical image I(i,j).

[0091] In a specific implementation, in the medical image watermarking method provided in the embodiments of the present invention, step S102 uses the TEBLID algorithm to describe the key points of the medical image and generate a feature descriptor matrix of the medical image. Specifically, it may include: first, using binary weak descriptors as a decision tree on a pre-defined feature extraction function; then, using the decision tree to process the key points of the medical image I(i,j) and using a greedy algorithm to minimize the selection of the corresponding target weak descriptor; finally, generating the feature descriptor matrix descriptps of the medical image I(i,j) based on the selected target weak descriptor.

[0092] Specifically, the feature extraction function mentioned above can be both discriminative and computationally fast. The expression for this feature extraction function f(x) is as follows:

[0093]

[0094] Where I(t) is the gray value at pixel t, and R(u,s) is a square image box of size s centered at pixel u.

[0095] Use binary weak descriptors

[0096]

[0097] As a decision tree on the feature extraction function f(x), the feature is calculated from the difference in average gray values ​​at pixel u, where the center of a bounding box of size s is located. It approximates the image gradient at a given orientation and scale and can be efficiently computed using integral images.

[0098] We use a greedy algorithm to minimize the selection of the i-th weak descriptor.

[0099]

[0100] in,[·] + =max(0,·) is a TRL with boundary τ, a i It's an anchor point patch, p i It is a positive sample in the same scene as the anchor point (i.e., the same scene), ni These are negative samples (i.e., different scenarios) related to 'a'. We use τ to find the K features that are minimized.

[0101] In a specific implementation, in the medical image watermarking method provided in the embodiments of the present invention, step S103 performs DCT transformation on the feature descriptor matrix of the medical image to obtain the feature vector of the medical image. Specifically, this may include: first, performing DCT transformation on the feature descriptor matrix descriptps of the medical image I(i,j) to obtain the coefficient matrix F(i,j) of the medical image I(i,j):

[0102] F(i,j)=DCT2(descripts(i,j));

[0103] Then, a matrix of a predetermined size (e.g., 2*16 for low-frequency regions) is taken from the coefficient matrix of the medical image I(i,j) to form a new matrix D(i,j) for the medical image I(i,j). A hash function is used to perform a symbol transformation on the new matrix D(i,j) of the medical image I(i,j) to obtain the feature vector V(i,j). Specifically, if an element in the new matrix D(i,j) is greater than 0, it is set to 1; otherwise, it is set to 0. This generates a 32-bit binary sequence of features for the medical image I(i,j) as the feature vector V(i,j).

[0104] In a specific implementation, the medical image watermarking method provided in the embodiments of the present invention may further include the following steps before performing the bitwise XOR operation between the feature vector of the medical image and the chaotic scrambled watermark in step S104: obtaining a binary chaotic sequence; binarizing the original watermark to obtain a binary watermark image; and scrambling the binary watermark image using the binary chaotic sequence to obtain a chaotic scrambled watermark.

[0105] Specifically, a chaotic sequence can be generated using a Logistic chaotic mapping with set parameter values, and the initial state values ​​can be set independently. The sequence length is M*N-1, where M and N are the dimensions of the host image. The parameter values ​​can range from 3.57 to 4. Preferably, a parameter value of 4 can meet the application requirements and generate a good chaotic sequence.

[0106] The original watermark W(i,j) is binarized, and the generated chaotic sequence is used to scramble the binary watermark image. The scrambling method involves rearranging the pixels of the binary watermark image according to the sorted chaotic sequence. This yields the chaotic scrambled watermark BW(i,j).

[0107] In a specific implementation, the medical image watermarking method provided in the embodiments of the present invention may further include, while performing step S104 of embedding watermark information into the medical image, generating a logical key Key(i,j):

[0108]

[0109] Save Key(i,j), which will be used later when extracting the watermark. By applying for Key(i,j) as a key from a third party, we can obtain ownership and usage rights of the medical image I(i,j), thereby achieving the purpose of protecting the medical image I(i,j).

[0110] Next, feature extraction is performed on the medical image I'(i,j) to be tested, resulting in its feature vector. This medical image I'(i,j) can be understood as the image obtained after the medical image I(i,j) has been transmitted over the network.

[0111] In specific implementation, feature extraction is performed on the medical image I'(i,j) to be tested, to obtain the feature vector of the medical image I'(i,j). Specifically, this may include: first, using the CenSurE-STAR algorithm to extract the key points of the medical image I'(i,j); then, using the TEBLID algorithm to describe the key points of the medical image I'(i,j) to generate the feature descriptor matrix descrips' of the medical image I'(i,j); finally, performing DCT transformation on the feature descriptor matrix descrips' of the medical image I'(i,j) to obtain the feature vector V'(i,j) of the medical image I'(i,j).

[0112] Finally, the feature vector V'(i,j) of the medical image I'(i,j) to be tested and the logical key Key(i,j) are XORed to extract the encrypted watermark BW'(i,j):

[0113]

[0114] It should be noted that the medical image watermarking method provided in the embodiments of the present invention only requires the key Key(i,j) when extracting the watermark, and does not require the original image to participate, which is a zero-watermark extraction algorithm.

[0115] In a specific implementation, the medical image watermarking method provided in the embodiments of the present invention uses the CenSurE-STAR algorithm to extract key points of the medical image I'(i,j) to be tested. Specifically, it may include: first, performing approximate Gaussian difference processing on the medical image I'(i,j) to be tested, and using a non-maximum suppression algorithm to detect the local maxima corresponding to the medical image I'(i,j) to be tested as feature points; then, using a Harris edge filter to remove unstable feature points on the edges of the medical image I'(i,j) to be tested, and the remaining feature points are used as key points of the medical image I'(i,j) to be tested.

[0116] In a specific implementation, the medical image watermarking method provided in the embodiments of the present invention uses the TEBLID algorithm to describe the key points of the medical image I'(i,j) to be tested, generating a feature descriptor matrix descriptps' of the medical image I'(i,j). Specifically, this may include: processing the key points of the medical image I'(i,j) to be tested using a decision tree, and minimizing the selection of the corresponding target weak descriptor using a greedy algorithm; and generating the feature descriptor matrix descriptps' of the medical image I'(i,j) to be tested based on the selected target weak descriptor.

[0117] The feature extraction method for the medical image I'(i,j) to be tested can be referred to the feature extraction method for the medical image I(i,j) mentioned above, and will not be repeated here.

[0118] In specific implementation, the above-mentioned medical image watermarking method provided in the embodiments of the present invention may further include, after extracting the encrypted watermark, the following steps: using the same method as watermark encryption to obtain the same binary chaotic sequence, performing an XOR operation between the extracted encrypted watermark BW'(i,j) and the binary chaotic sequence to obtain the restored watermark W'(i,j).

[0119] By calculating the normalized cross correlation (NC) between the original watermark W(i,j) and the restored watermark W'(i,j), the ownership of the medical image I(i,j) and the embedded watermark information are determined.

[0120] The invention will be further described below with reference to the accompanying drawings. The experimental test object is a 512*512 medical image of a hand, see... Figure 2 Let I(i,j) represent the image. First, perform CenSurE-STAR-TEBLID-DCT transformation on the hand medical image I(i,j). Considering robustness and the capacity of one-time watermark embedding, take 32 coefficients, that is, a 2*16 module.

[0121] The discriminative power of the algorithm was verified using seven different medical images. 32-bit feature vectors were extracted from each image, and the normalized correlation coefficients between them were calculated. Figures 3 to 9 Table 1 presents seven different medical images and their corresponding NC values. The data shows that the NC values ​​of the different images obtained using the feature vectors extracted by the above method are all less than 0.5, and each image has an NC value of 1.00, indicating that the algorithm can effectively distinguish between different images. These results are consistent with human visual characteristics.

[0122] Table 1. Correlation coefficients (32-bit) between seven different medical images.

[0123] Figure 3 1 0.46 0.41 0.42 0.18 0.37 0.4 Figure 4 0.46 1 0.26 0.17 0.43 0.33 0.43 Figure 5 0.41 0.26 1 0.38 0.35 0.23 0.44 Figure 6 0.42 0.17 0.38 1 0.35 0.43 0.34 Figure 7 0.18 0.43 0.35 0.35 1 0.28 0.49 Figure 8 0.37 0.33 0.23 0.43 0.28 1 0.15 Figure 9 0.4 0.43 0.44 0.34 0.49 0.15 1

[0124] Choose a meaningful binary image as the original watermark. Here, the watermark size is 32*32, denoted as W(i,j). See [link to original text]. Figure 10 The initial value of the chaos coefficient is set to 0.2, the growth parameter is 4, and the number of iterations is 32. Then, the original watermark is subjected to chaotic encryption. The encrypted watermark is shown below. Figure 11 After detecting and restoring the watermark W'(i,j) using the watermarking algorithm, the presence of an embedded watermark is determined by calculating NC. The closer the NC value is to 1, the higher the similarity, thus assessing the algorithm's robustness. PSNR represents the degree of image distortion; a higher PSNR value indicates less image distortion.

[0125] Figure 12 This is the watermark extracted without interference. As you can see, NC=1.00, indicating that the watermark can be extracted accurately.

[0126] The following experiments will determine the resistance to conventional attacks and geometric attacks of this digital watermarking method.

[0127] First, JPEG compression was applied to abdominal medical images using the percentage of image compression quality as a parameter; Table 2 shows the experimental data on watermark resistance to JPEG compression. When the compression quality was 5%, the image quality was low, but the watermark could still be extracted, NC = 0.72.

[0128] Figure 13 These are medical images with a compression quality of 10%.

[0129] Figure 14 It is a watermark extracted with a compression quality of 10%, NC=1.

[0130] Table 2. Watermark Anti-JPEG Compression Experimental Data

[0131] PSNR (dB) 39.27 38.80 37.54 35.30 31.95 27.97 NC 0.62 0.79 1 0.79 1 0.72

[0132] Table 3 shows the experimental data on watermark resistance to median filtering attacks. As can be seen from Table 3, when the median filtering window size is [3*3] and the number of filtering iterations is 20, NC=1, and the watermark can still be extracted.

[0133] Figure 15 It is a medical image with a mean filtering window size of [3*3] and a filtering repetition count of 5;

[0134] Figure 16 The watermark extracted is a mean filter with a window size of [3*3] and a filter repetition count of 5, with NC = 0.89.

[0135] Table 3. Experimental Data on Watermark Anti-Median Filtering

[0136] PSNR (dB) 37.26 36.25 35.69 NC 0.89 0.79 0.79

[0137] Table 4 shows the experimental data on watermark resistance to rotation attacks. As can be seen from Table 4, when the image is rotated 60° counterclockwise, NC = 0.89, and the watermark can still be extracted.

[0138] Figure 17 It is a medical image rotated 20° counterclockwise;

[0139] Figure 18 The watermark was extracted by rotating counterclockwise by 20°, with NC=0.79, which can accurately extract the watermark.

[0140] Figure 19 It is a medical image rotated 45° counterclockwise;

[0141] Figure 20 It is a watermark extracted by rotating counterclockwise by 45°, with NC=1, which can accurately extract the watermark.

[0142] Table 4. Experimental Data on Watermark Resistance to Rotation Attacks

[0143] PSNR (dB) 11.50 9.54 9.21 8.74 8.43 NC 0.79 0.79 1 0.79 1

[0144] Table 5 shows the experimental data on watermark resistance to scaling attacks in medical images. As can be seen from Table 5, when the scaling factor is as small as 0.9, the correlation coefficient NC = 0.79, and the watermark can be accurately extracted.

[0145] Figure 21 This is a scaled-down medical image (scaling factor 1.3);

[0146] Figure 22 It is a watermark extracted after a scaling attack, with NC=0.79, which can accurately extract the watermark.

[0147] Figure 23 This is a scaled-down medical image (scaling factor of 1.5);

[0148] Figure 24 It is the watermark extracted after a scaling attack. NC=1, which can accurately extract the watermark.

[0149] Table 5: Experimental Data on Watermark Anti-Scaling Attacks

[0150] NC 0.57 0.79 1 0.79 1 0.89 0.7

[0151] Table 6 shows the experimental data on watermark resistance to translation transformation. Table 6 shows that when the image data is vertically shifted by 40%, the NC value is always higher than 0.6, indicating accurate watermark extraction. Therefore, this watermarking method has strong resistance to translation transformation.

[0152] Figure 25 This is a medical image that has been horizontally shifted 20% to the left.

[0153] Figure 26 The watermark is extracted after shifting horizontally to the left by 20%, and the watermark can be accurately extracted. NC=1.

[0154] Figure 27 This is the image after the medical image has been vertically shifted upwards by 20%.

[0155] Figure 28 The watermark was extracted after vertically shifting upwards by 20%, and it can accurately extract the watermark, NC=0.89.

[0156] Table 6. Experimental Data on Watermark Anti-Translation Transformation

[0157]

[0158] Table 7 shows the experimental data on watermark resistance to shearing attacks. As can be seen from Table 7, when the medical image is sheared along the Y-axis with a shearing amount of 40%, NC = 0.79. When the medical image is sheared along the X-axis with a shearing amount of 40%, NC = 0.80. The watermark can still be extracted, indicating that the watermarking algorithm has strong resistance to shearing attacks.

[0159] Figure 29 This is a medical image after being cut 30% along the Y-axis;

[0160] Figure 30 The watermark is extracted after cutting 30% along the Y-axis, and the watermark can be accurately extracted with NC=0.8.

[0161] Figure 31 This is a medical image after being cut 30% along the X-axis;

[0162] Figure 32 The watermark is extracted after cutting 30% along the X-axis, and the watermark can be accurately extracted with NC=0.71.

[0163] Table 7. Experimental Data on Watermark Resistance to Shearing Attacks

[0164]

[0165] The various embodiments in this specification are described in a progressive manner. Each embodiment focuses on the differences from other embodiments. The same or similar parts between the various embodiments can be referred to each other.

[0166] Those skilled in the art will further recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of both. To clearly illustrate the interchangeability of hardware and software, the components and steps of the various examples have been generally described in terms of functionality in the foregoing description. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of this invention.

[0167] The steps of the methods or algorithms described in conjunction with the embodiments disclosed herein can be implemented directly by hardware, a software module executed by a processor, or a combination of both. The software module can be located in random access memory (RAM), main memory, read-only memory (ROM), electrically programmable ROM, electrically erasable programmable ROM, registers, hard disk, removable disk, CD-ROM, or any other form of storage medium known in the art.

[0168] Finally, it should be noted that in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0169] The above provides a detailed description of the medical image watermarking method based on CenSurE-STAR-TEBLID-DCT provided by this invention. Specific examples have been used to illustrate the principles and implementation methods of this invention. The descriptions of the above embodiments are only for the purpose of helping to understand the method and core ideas of this invention. At the same time, for those skilled in the art, there will be changes in the specific implementation methods and application scope based on the ideas of this invention. Therefore, the content of this specification should not be construed as a limitation of this invention.

Claims

1. A medical image watermarking method based on CenSurE-STAR-TEBLID-DCT, characterized in that, include: The CenSurE-STAR algorithm was used to extract key points from medical images; The TEBLID algorithm is used to describe the key points of the medical image and generate the feature descriptor matrix of the medical image. The process includes: using binary weak descriptors as a decision tree on a pre-defined feature extraction function; processing the key points of the medical image using the decision tree and minimizing the selection of the corresponding target weak descriptor using a greedy algorithm; and generating the feature descriptor matrix of the medical image based on the selected target weak descriptor. Perform DCT transformation on the feature descriptor matrix of the medical image to obtain the feature vector of the medical image; The feature vector of the medical image is XORed bit by bit with the chaotic scrambled watermark to embed the watermark information into the medical image.

2. The medical image watermarking method based on CenSurE-STAR-TEBLID-DCT according to claim 1, characterized in that, The CenSurE-STAR algorithm was used to extract key points from medical images, including: The medical image is processed by approximate Gaussian difference, and the local maxima corresponding to the medical image are detected as feature points using a non-maximum suppression algorithm. The Harris edge filter is used to remove unstable feature points on the edges of the medical image, and the remaining feature points are used as the key points of the medical image.

3. The medical image watermarking method based on CenSurE-STAR-TEBLID-DCT according to claim 2, characterized in that, Performing a DCT transform on the feature descriptor matrix of the medical image to obtain the feature vector of the medical image includes: The feature descriptor matrix of the medical image is subjected to DCT transformation to obtain the coefficient matrix of the medical image; Take a matrix of a predetermined size from the coefficient matrix of the medical image to form a new matrix of the medical image; The new matrix of the medical image is symbolically transformed using a hash function to obtain the feature vector of the medical image.

4. The medical image watermarking method based on CenSurE-STAR-TEBLID-DCT according to claim 3, characterized in that, Before performing a bitwise XOR operation between the feature vector of the medical image and the chaotic scrambled watermark, the process further includes: Obtain a binary chaotic sequence; The original watermark is binarized to obtain a binary watermark image; The binary chaotic sequence is used to scramble the binary watermark image to obtain a chaotic scrambled watermark.

5. The medical image watermarking method based on CenSurE-STAR-TEBLID-DCT according to claim 4, characterized in that, In addition to embedding watermark information into the medical image, the method also includes: Generate a logical key; Feature extraction is performed on the medical image to be tested to obtain the feature vector of the medical image to be tested; The feature vector of the medical image to be tested and the logical key are XORed to extract the encrypted watermark.

6. The medical image watermarking method based on CenSurE-STAR-TEBLID-DCT according to claim 5, characterized in that, Feature extraction is performed on the medical image to be tested to obtain the feature vector of the medical image to be tested, including: The CenSurE-STAR algorithm was used to extract key points from the medical image under test. The TEBLID algorithm is used to describe the key points of the medical image under test, and a feature descriptor matrix of the medical image under test is generated. The feature descriptor matrix of the medical image under test is subjected to DCT transformation to obtain the feature vector of the medical image under test.

7. The medical image watermarking method based on CenSurE-STAR-TEBLID-DCT according to claim 6, characterized in that, The CenSurE-STAR algorithm was used to extract key points from the medical image under test, including: The medical image to be tested is subjected to approximate Gaussian difference processing, and the local maxima corresponding to the medical image to be tested are detected as feature points using a non-maximum suppression algorithm; The Harris edge filter is used to remove unstable feature points on the edges of the medical image under test, and the remaining feature points are used as the key points of the medical image under test.

8. The medical image watermarking method based on CenSurE-STAR-TEBLID-DCT according to claim 7, characterized in that, The TEBLID algorithm is used to describe the key points of the medical image under test, generating a feature descriptor matrix of the medical image under test, including: The decision tree is used to process the key points of the medical image under test, and a greedy algorithm is used to minimize and select the corresponding target weak descriptor. Based on the selected target weak descriptors, a feature descriptor matrix is ​​generated for the medical image under test.

9. The medical image watermarking method based on CenSurE-STAR-TEBLID-DCT according to claim 8, characterized in that, After extracting the encrypted watermark, it also includes: The extracted encrypted watermark is XORed with the binary chaotic sequence to obtain the restored watermark.

Images