CN113012020A - Image watermarking method, system and electronic equipment - Google Patents
Image watermarking method, system and electronic equipment Download PDFInfo
- Publication number
- CN113012020A CN113012020A CN202110443175.XA CN202110443175A CN113012020A CN 113012020 A CN113012020 A CN 113012020A CN 202110443175 A CN202110443175 A CN 202110443175A CN 113012020 A CN113012020 A CN 113012020A
- Authority
- CN
- China
- Prior art keywords
- image
- feature
- feature matrix
- host
- detected
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T1/00—General purpose image data processing
- G06T1/0021—Image watermarking
-
- G—PHYSICS
- G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
- G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
- G16H30/00—ICT specially adapted for the handling or processing of medical images
- G16H30/40—ICT specially adapted for the handling or processing of medical images for processing medical images, e.g. editing
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02D—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN INFORMATION AND COMMUNICATION TECHNOLOGIES [ICT], I.E. INFORMATION AND COMMUNICATION TECHNOLOGIES AIMING AT THE REDUCTION OF THEIR OWN ENERGY USE
- Y02D10/00—Energy efficient computing, e.g. low power processors, power management or thermal management
Landscapes
- Engineering & Computer Science (AREA)
- Health & Medical Sciences (AREA)
- Medical Informatics (AREA)
- Radiology & Medical Imaging (AREA)
- Epidemiology (AREA)
- General Health & Medical Sciences (AREA)
- Nuclear Medicine, Radiotherapy & Molecular Imaging (AREA)
- Primary Health Care (AREA)
- Public Health (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Editing Of Facsimile Originals (AREA)
- Image Processing (AREA)
Abstract
The invention relates to an image watermarking method, a system and electronic equipment, which can not modify any content of a host image and effectively protect the integrity of the host image on one hand, and can construct a first characteristic image and a second characteristic image by combining the advantages of two-dimensional empirical mode decomposition (BEMD) and Singular Value Decomposition (SVD) algorithm for further copyright authentication and tampering detection on the other hand. A large number of experimental results and comparison with the existing watermarking algorithm prove that the method has good performance in tampering detection and also has good performance in resisting various attacks, particularly shearing attack, Gaussian noise, median filtering, image enhancement and the like, and the robustness is improved.
Description
Technical Field
The present invention relates to the field of image watermarking technologies, and in particular, to an image watermarking method, an image watermarking system, and an electronic device.
Background
Images often store private information of an individual, such as: digital medical images store private information of patients and play an important role in the process of disease diagnosis. Therefore, it is very important to protect private information of the patient. The digital medical image may be copied and tampered by unauthorized persons in the process of transmission and the like, and the medical value of the medical image is greatly damaged and the personal rights and interests of patients are greatly infringed. Watermarking algorithms for digital medical images have been widely used in order to better protect ownership of patient medical images from human exploitation and abuse.
Conventional watermarking algorithms generally provide copyright protection by embedding a watermark image into a host image, which may present a variety of problems due to the particularities of digital medical images, in particular:
1) inserting watermark information, i.e. watermark images, into a host image, i.e. digital medical image, inevitably modifies the host image, i.e. the digital medical image needs to be modified, which may cause misdiagnosis of a doctor, because any modification of content may mask the actual condition of a patient;
2) the direct insertion of watermark information, i.e., a watermark image, into a host image, i.e., a digital medical image, may make it difficult for conventional watermarking algorithms to balance robustness and imperceptibility.
Disclosure of Invention
The invention aims to solve the technical problem of the prior art and provides an image watermarking method, an image watermarking system and electronic equipment.
The technical scheme of the image watermarking method of the invention is as follows:
decomposing a host image by using a two-dimensional empirical mode decomposition algorithm to obtain a first intrinsic mode function and a margin corresponding to the host image, decomposing the first intrinsic mode function corresponding to the host image by using a singular value decomposition algorithm with a preset window to obtain a plurality of first feature matrices, and decomposing the margin corresponding to the host image by using a singular value decomposition algorithm with a preset window to obtain a plurality of second feature matrices;
respectively obtaining elements corresponding to the same preset position from each first feature matrix and carrying out binarization to obtain a third feature matrix, and respectively obtaining elements corresponding to the same preset position from each second feature matrix and carrying out binarization to obtain a fourth feature matrix;
performing Arnold scrambling on the watermark image corresponding to the host image and repeating to obtain a fifth feature matrix with the size consistent with that of the third feature matrix;
and performing exclusive-or operation on the fifth feature matrix and the third feature matrix to obtain the first feature image, performing exclusive-or operation on the fifth feature matrix and the fourth feature matrix to obtain the second feature image, and performing copyright authentication and/or tampering detection according to the first feature image and the second feature image.
The image watermarking method has the following beneficial effects:
on one hand, the host image is not modified, the integrity of the host image is effectively protected, on the other hand, the advantages of the two-dimensional empirical mode decomposition BEMD and the singular value decomposition SVD are combined, firstly, the good characteristics of the BEMD are fully exerted, and the host image is decomposed into a limited number of inherent modal functions and margins in a self-adaptive mode. Secondly, respectively using a singular value decomposition algorithm from a first inherent mode function and the allowance corresponding to the host image, and performing binarization operation to obtain a third feature matrix and a fourth feature matrix; the watermark image is then secured using an Arnold transformation, a first feature image and the second feature image are constructed by an exclusive OR operation (XOR), and the first feature image and the second feature image may be securely stored in a copyright authentication database for further use in copyright authentication and tamper detection. A large number of experimental results and comparison with the existing watermarking algorithm prove that the method has good performance in tampering detection and also has good performance in resisting various attacks, particularly shearing attack, Gaussian noise, median filtering, image enhancement and the like, and the robustness is improved.
The technical scheme of the image watermarking system is as follows:
the device comprises a decomposition module, a binarization module, a scrambling repetition module and an exclusive OR operation module;
the decomposition module is configured to: decomposing a host image by using a two-dimensional empirical mode decomposition algorithm to obtain a first intrinsic mode function and a margin corresponding to the host image, decomposing the first intrinsic mode function corresponding to the host image by using a singular value decomposition algorithm with a preset window to obtain a plurality of first feature matrices, and decomposing the margin corresponding to the host image by using a singular value decomposition algorithm with a preset window to obtain a plurality of second feature matrices;
the binarization module is used for: respectively obtaining elements corresponding to the same preset position from each first feature matrix and carrying out binarization to obtain a third feature matrix, and respectively obtaining elements corresponding to the same preset position from each second feature matrix and carrying out binarization to obtain a fourth feature matrix;
the scrambling repetition module is to: performing Arnold scrambling on the watermark image corresponding to the host image and repeating to obtain a fifth feature matrix with the size consistent with that of the third feature matrix;
the XOR operation module is to: and performing exclusive-or operation on the fifth feature matrix and the third feature matrix to obtain the first feature image, performing exclusive-or operation on the fifth feature matrix and the fourth feature matrix to obtain the second feature image, and performing copyright authentication and/or tampering detection according to the first feature image and the second feature image.
The image watermarking system has the following beneficial effects:
on one hand, the host image is not modified, the integrity of the host image is effectively protected, on the other hand, the advantages of the two-dimensional empirical mode decomposition BEMD and the singular value decomposition SVD are combined, firstly, the good characteristics of the BEMD are fully exerted, and the host image is decomposed into a limited number of inherent modal functions and margins in a self-adaptive mode. Secondly, respectively using a singular value decomposition algorithm from a first inherent mode function and the allowance corresponding to the host image, and performing binarization operation to obtain a third feature matrix and a fourth feature matrix; the watermark image is then secured using an Arnold transformation, a first feature image and the second feature image are constructed by an exclusive OR operation (XOR), and the first feature image and the second feature image may be securely stored in a copyright authentication database for further use in copyright authentication and tamper detection. A large number of experimental results and comparison with the existing watermarking algorithm prove that the method has good performance in tampering detection and also has good performance in resisting various attacks, particularly shearing attack, Gaussian noise, median filtering, image enhancement and the like, and the robustness is improved.
The technical scheme of the electronic equipment is as follows:
comprising a memory, a processor and a program stored on the memory and running on the processor, the processor implementing the steps of an image watermarking method as described in any one of the preceding claims when executing the program.
Drawings
Fig. 1 is a schematic flowchart of an image watermarking method according to an embodiment of the present invention;
fig. 2 is a schematic flow chart of acquiring a watermark image to be detected;
FIG. 3 is a schematic illustration of a host image;
FIG. 4 is a schematic diagram of a watermark image;
FIG. 5 is the result of a median filtering attack;
FIG. 6 shows the results of salt and pepper noise attack, speckle noise attack, and Gaussian noise attack;
FIG. 7 shows one of the results of comparison with the literature;
FIG. 8 shows the second comparison result with the reference;
fig. 9 is a schematic structural diagram of an image watermarking system according to an embodiment of the present invention;
Detailed Description
As shown in fig. 1, an image watermarking method according to an embodiment of the present invention includes the following steps:
s1, sequentially decomposing the host image by using a two-dimensional empirical mode decomposition algorithm and a singular value decomposition algorithm, specifically:
decomposing a host image by using a two-dimensional empirical mode decomposition algorithm to obtain a first intrinsic mode function and a margin corresponding to the host image, decomposing the first intrinsic mode function corresponding to the host image by using a singular value decomposition algorithm with a preset window to obtain a plurality of first feature matrices, and decomposing the margin corresponding to the host image by using a singular value decomposition algorithm with a preset window to obtain a plurality of second feature matrices;
s2, obtaining elements corresponding to the same preset position and carrying out binarization, specifically:
respectively obtaining elements corresponding to the same preset position from each first feature matrix and carrying out binarization to obtain a third feature matrix, and respectively obtaining elements corresponding to the same preset position from each second feature matrix and carrying out binarization to obtain a fourth feature matrix;
s3, scrambling and repeating the watermark image, specifically: performing Arnold scrambling on the watermark image corresponding to the host image and repeating to obtain a fifth feature matrix with the size consistent with that of the third feature matrix;
s4, performing an exclusive or operation to obtain a first feature image and a second feature image, specifically:
and performing exclusive-or operation on the fifth feature matrix and the third feature matrix to obtain the first feature image, performing exclusive-or operation on the fifth feature matrix and the fourth feature matrix to obtain the second feature image, and performing copyright authentication and/or tampering detection according to the first feature image and the second feature image.
On one hand, the host image is not modified, the integrity of the host image is effectively protected, on the other hand, the advantages of the two-dimensional empirical mode decomposition BEMD and the singular value decomposition SVD are combined, firstly, the good characteristics of the BEMD are fully exerted, and the host image is decomposed into a limited number of inherent modal functions and margins in a self-adaptive mode. Secondly, respectively using a singular value decomposition algorithm from a first inherent mode function and the allowance corresponding to the host image, and performing binarization operation to obtain a third feature matrix and a fourth feature matrix; the watermark image is then secured using an Arnold transformation, a first feature image and the second feature image are constructed by an exclusive OR operation (XOR), and the first feature image and the second feature image may be securely stored in a copyright authentication database for further use in copyright authentication and tamper detection. A large number of experimental results and comparison with the existing watermarking algorithm prove that the method has good performance in tampering detection and also has good performance in resisting various attacks, particularly shearing attack, Gaussian noise, median filtering, image enhancement and the like, and the robustness is improved.
Taking a grayscale image in which the host image H has a pixel size of m × n and a binary image in which the watermark image W has a pixel size of p × q as examples, specifically:
s10, decomposing by using a two-dimensional empirical mode decomposition algorithm: specifically, the method comprises the following steps:
decomposing the host image H by using a two-dimensional Empirical Mode Decomposition (Bi-dimensional Empirical Mode Decomposition, abbreviated as BEMD) algorithm to obtain a plurality of intrinsic Mode functions, setting a Decomposition termination condition when using the two-dimensional Empirical Mode Decomposition algorithm, for example, setting a margin termination threshold, stopping Decomposition when a margin res obtained by decomposing the host image H reaches the margin termination threshold, or directly setting the total number of the obtained intrinsic Mode functions as the termination condition, for example, setting 5 intrinsic Mode functions, stopping Decomposition, at this time
Where k represents the total number of resulting eigenmode functions, IMFiAn ith natural mode function corresponding to the host image H is represented, i is a positive integer, so that a first natural mode function corresponding to the host image H and a margin res are obtained, and then:
s11, decomposing by using a singular value decomposition algorithm, specifically:
setting a preset window of a singular value decomposition algorithm as [ L, L ], wherein L represents the number of pixels, decomposing a first inherent mode function corresponding to a host image H by using the singular value decomposition algorithm with the preset window to obtain a plurality of first characteristic matrixes, and decomposing the surplus corresponding to the host image by using the singular value decomposition algorithm with the preset window to obtain a plurality of second characteristic matrixes;
s12, obtaining a third feature matrix
And a fourth feature matrix
Specifically, the method comprises the following steps:
1) obtaining elements corresponding to the same preset position from each first feature matrix respectively, and generally setting the preset position as the position of the first element at the upper left corner, where the matrix is, for example:
the position of the element 1 is the position of the first element at the upper left corner, so as to obtain a first intermediate feature matrix I, and the size of the first intermediate feature matrix I is m/L multiplied by n/L;
each element in the first intermediate feature matrix I is binarized, specifically: comparing each element in the first intermediate feature matrix I with an average value of all elements in the first intermediate feature matrix I, respectively, recording elements larger than the average value as 1, and recording elements smaller than the average value as 0; thereby obtaining a third feature matrix
2) Respectively acquiring elements corresponding to the same preset position from each second feature matrix, thereby obtaining a second intermediate feature matrix R, wherein the size of the second intermediate feature matrix R is also m/L multiplied by n/L;
each element in the second intermediate feature matrix R is binarized, specifically: comparing each element in the second intermediate feature matrix R with an average value of all elements in the second intermediate feature matrix R, respectively, recording the elements larger than the average value as 1, and recording the elements smaller than the average value as 0; thereby obtaining a fourth feature matrix
S13, processing the watermark image, specifically:
firstly, performing Amold scrambling on watermark image W to obtain scrambled imageThen, the scrambled image W' is repeated m/(Lp) × n/(Lq) times to obtain a third feature matrix
A fifth feature matrix WA of uniform size, wherein the third feature matrix WA
And a fourth feature matrix
Consistent in size, it can also be described here as: obtaining a fourth feature matrix
The fifth feature matrix WA of uniform size;
s14, carrying out exclusive OR operation to obtain a first feature image F'IAnd a second feature image F'RSpecifically:
combining the fifth feature matrix WA with the third feature matrix WA
Performing XOR operation, i.e. comparing the fifth feature matrix WA with the third feature matrix WA
Is equal, the result of the exclusive or operation is 0 if equal, and is 1 if not equal, thereby obtaining a first feature image F'I(ii) a Similarly, and the fifth feature matrix WA is compared with the fourth feature matrix
Performing exclusive OR operation to obtain a second feature image F'RExtracting the first feature image F'IAnd the second feature image F'RSecurely stored in a copyright authentication database to be in accordance with the first feature image F'IAnd the second characteristic diagramLike F'RCopyright authentication and/or tamper detection is performed.
It is understood that S10-S14 corresponds to the watermark embedding process, but in this process, no content modification is caused to the host image H, which can be regarded as a zero-watermark algorithm, and the integrity of the host image H is effectively protected, and then:
1) as shown in fig. 2, the process of detecting whether the host image to be detected is tampered includes S5-S9, specifically:
s5, sequentially decomposing the host image to be detected by using a two-dimensional empirical mode decomposition algorithm and a singular value decomposition algorithm, specifically:
decomposing the host image H' to be detected corresponding to the host image H by using a two-dimensional empirical mode decomposition algorithm to obtain a plurality of inherent mode functions, wherein at the moment
res 'represents a margin, IMF', corresponding to the host image H 'to be detected'1Representing the ith inherent modal function corresponding to the host image H 'to be detected, wherein i is a positive integer, and obtaining the first inherent modal function IMF' corresponding to the host image H 'to be detected'1And the balance res',
and using a window having a predetermined window, [ L, L]The singular value decomposition algorithm is used for carrying out the IMF 'on the first intrinsic mode function corresponding to the host image to be detected'1Decomposing to obtain multiple sixth feature matrices, and using the predetermined window [ L, L ]]Decomposing the residual res' corresponding to the host image to be detected by using a singular value decomposition algorithm to obtain a plurality of seventh feature matrices;
it can be understood that the host image H' to be detected corresponds to the host image H, so that various pre-stored parameters of the host image H are used to verify whether the host image to be detected is tampered, and the various parameters include: the two-dimensional empirical mode decomposition algorithm is used for decomposing the host image H' under the same termination condition, and the preset windows in the singular value decomposition algorithm are the same.
S6, acquiring elements and carrying out binarization, specifically:
1) respectively acquiring elements corresponding to the same preset position from each sixth feature matrix, which is referred to in the process of "acquiring elements corresponding to the same preset position from each first feature matrix", and also respectively selecting elements at the first position at the upper left corner from each sixth feature matrix to obtain a third intermediate feature matrix I'; and binarizing each element in the third intermediate feature matrix I', specifically: comparing each element in the third intermediate feature matrix I 'with an average value of all elements in the third intermediate feature matrix I', respectively, and recording elements larger than the average value as 1 and elements smaller than the average value as 0; thereby obtaining an eighth feature matrix I ";
2) respectively acquiring elements corresponding to the same preset position from each seventh feature matrix, which is referred to in the process of "acquiring elements corresponding to the same preset position from each first feature matrix", and also respectively selecting elements at the first position at the upper left corner from each seventh feature matrix to obtain a fourth intermediate feature matrix R'; and binarizing each element in the fourth intermediate feature matrix R', specifically: comparing each element in the fourth intermediate feature matrix R 'with an average value of all elements in the fourth intermediate feature matrix R', respectively, and recording elements greater than the average value as 1 and elements less than the average value as 0; thereby obtaining a ninth feature matrix R';
s7, performing an exclusive or operation to obtain a third feature image and a fourth feature image, specifically:
a first feature image F 'of the host image'IAnd the eighth feature matrix I ' are subjected to exclusive-OR operation, namely, the feature images F ' are respectively compared 'IWhether the values of the elements at the same positions in the eighth feature matrix I ″ are equal, if so, the result of the exclusive-or operation is 0, and if not, the result of the exclusive-or operation is 1, thereby obtaining a third feature image WF; similarly, and a second feature image F 'of the host image'RCarrying out exclusive or operation on the fourth characteristic matrix R ' and the ninth characteristic matrix R ' to obtain a fourth characteristic image W ';
s8, acquiring the watermark image to be detected, specifically:
dividing the fourth feature image W ″ according to the size of a preset window to obtain a plurality of tenth feature matrices, where the total number of the tenth feature matrices is m/(Lp) × n/(Lq), which can be understood as: the fourth feature image W ″ is divided into m/(Lp) × n/(Lq) blocks, and elements at the same position in each tenth feature matrix are added, specifically: taking the sum of the elements at the first position at the upper left corner in each tenth feature matrix as an element, placing the element at the first position at the upper left corner of a newly generated fifth intermediate feature matrix, repeating the above steps to obtain a fifth intermediate feature matrix, and binarizing each element in the fifth intermediate feature matrix, specifically: comparing each element in the fifth intermediate characteristic matrix with the average value of all elements in the fifth intermediate characteristic matrix, marking the element larger than the average value as 1, and marking the element smaller than the average value as 0, so as to obtain an image W' with the same size as the watermark image W corresponding to the host image Hr；
Then, for image W ″)rPerforming Amold inverse transformation to obtain a watermark image W' to be detected;
s9, detecting whether tampering has been received, specifically:
dividing the third feature image WF according to a preset window size [ L, L ] to obtain a plurality of eleventh feature matrices, where the total number of the eleventh feature matrices is m/(Lp) × n/(Lq), and can be understood as: dividing the third feature image WF into m/(Lp) × n/(Lq) blocks;
and then, respectively calculating an NC (numerical control) value between each eleventh feature matrix and the watermark image W ' to be detected, judging that the host image H ' to be detected is tampered when the NC value smaller than a preset threshold exists, and judging that the host image H ' to be detected is not tampered when the NC value smaller than the preset threshold does not exist.
Specifically, the NC value between each eleventh feature matrix and the watermark image W' to be detected can be calculated by a first formula, where the first formula is:
where K (x, y) is the pixel value of the x-th row and the y-th column in each eleventh feature matrix, and K '(x, y) is the pixel value of the x-th row and the y-th column of the extracted watermark image W' to be detected. The larger the NC value is, the higher the similarity between the eleventh feature matrix and the watermark image W' to be detected is. When the NC value of the extracted watermark image is greater than 0.9, the quality of the extracted watermark is better, and the calculation process of the NC value of the normalized correlation coefficient is known to those skilled in the art, and is not described herein, then:
and when the NC value smaller than the preset threshold value does not exist, judging that the host image H' to be detected is not tampered.
Preferably, in the above technical solution, the method further comprises: when an NC value smaller than a preset threshold exists, dividing the NC value into two types by using a K-means algorithm, acquiring the type with a smaller average value of the NC value, and determining the position of the type corresponding to the host image H' to be detected as a tampered position, specifically:
the class with the smaller average value of the NC values is recorded as the first class, the class with the larger average value of the NC values is recorded as the second class, and then each NC value in the first class can be associated with a corresponding eleventh feature matrix, the corresponding eleventh feature matrix is associated with a third feature image WF, the third feature image WF is associated with an eighth feature matrix I ", and so on, the position of each NC value in the first class corresponding to the host image H 'to be detected can be determined, that is, the position is inverted from S9 to S5, and the position of each NC value in the first class corresponding to the host image H' to be detected can be determined, so that subsequent targeted processing can be performed conveniently.
Preferably, in the above technical solution, the method further comprises: and obtaining an NC value for representing the strong and weak robustness according to the watermark image W' to be detected and the watermark image. The larger the NC value is, the higher the similarity between the watermark image W' to be detected and the watermark image is, and the robustness is strong.
The following experiment is also carried out by carrying out attack experiments of various different parameters to verifyThe application is robust against attacks, in particular: the evaluation criteria were: the similarity measurement of the two images can be carried out by using data to carry out objective evaluation more accurately besides a visual observation-based method. The detection result of the watermark is generally measured by using a Normalized Correlation Coefficient (NC) to measure the extracted watermark image
The similarity with the original embedded watermark image K, the NC value calculation formula is referred to above and is not described herein,
the main parameters of the invention include termination conditions of BEMD decomposition, decomposition layer number and the like, the termination conditions of the BEMD decomposition process of the invention are SD, namely the margin is less than a certain threshold value, the value range of SD is [0.1, 0.3], generally, the smaller the value of SD, the more the number of inherent mode functions obtained by screening, but the longer the time consumption of the algorithm; therefore, in order to balance the algorithm duration with the number of natural mode functions, the termination condition of the present invention is: the default value of SD is 0.2, or the number of the intrinsic mode functions is uniformly set to 2. In the step of detecting tampering, the preset threshold may be set to 0.98.
The invention mainly uses 8 gray level images with the size of 512 multiplied by 512 pixels as host images, which are respectively named as brain, chest, cervical vertebra, abdominal cavity, brain kernel, skull, spine and brainstem, as shown in fig. 3, the used watermark images are 4 binary images with the size of 32 multiplied by 32 pixels, which are respectively named as information, elephant, medicine and CAD, as shown in fig. 4, then:
in order to check the robustness of the invention, the section attacks images containing embedded watermarks before watermark extraction, including filtering attack, noise attack and the like, wherein table 1 shows the NC values of different host images under different attacks, and the data in table 1 shows that the NC values of different host images under Gaussian filtering [3,3], median filtering [3,3], wiener filtering [3,3], mean filtering [3,3], Gaussian noise 0.005 and 0.01, sharpening attack 3, histogram equalization 64, JPEG compression 60 and scaling attack are all 1, which shows that the NC values of different host images under the attack types and corresponding strengths are all 1;
table 1:
fig. 5 shows the results of a large-windowed median filtering attack on a host image, comprising four different windows 7, 9, 11 and 13. In the figure, a host image is attacked to a certain degree, the pixel value tends to be smooth, and the result shows that all NC values are above 0.97 and the highest NC value is 1.000 in four filter attack experiments on the image, which shows that the algorithm of the invention has good effect on resisting filter attack. And the extracted watermark image can be clearly distinguished.
Fig. 6 shows the extraction results of salt and pepper noise, speckle noise, and gaussian noise attacks at σ ═ 0.1 and σ ═ 0.15. It can be seen that the watermark image can be completely extracted under the attack of speckle noise, and the corresponding NC values are all 1. Under the attack of Gaussian noise and salt and pepper noise, although a small amount of spots appear on the extracted watermark, the extracted watermark image can be well identified, and the NC value is larger than 0.99.
FIG. 7 shows the results of comparison of the present application with the literature (Qin F, Li J, Li H, et al. A Robust Zero-watermark for Medical Images Using Curvelet-Dct and RSA Pseudo-random Sequences [ C ]// International Conference on scientific understanding and security. Springer, Cham,2020: 179. 190.) under Gaussian noise attack. The algorithm of the invention and the algorithm of the comparison document can realize that NC is 1 when the intensity of 1% Gaussian noise is low. However, as the strength increases, the numerical value of NC of the reference decreases, and the result of the reference is lower than the algorithm result of the present invention after the strength is 10%. When the intensity increases to 75%, the NC value is below 0.5. In contrast, the algorithm of the present invention can still maintain a higher NC value in the following cases. A strong gaussian noise attack with NC values higher than 0.96 even at 75% intensity.
FIG. 8 shows the comparison of the NC values under shear attack in this application and the literature (Qian F, Li J, Li H, et al. A Robust Zero-Water marking Algorithm for Medical Images Using Curvelet-Dct and RSA Pseudo-random Sequences [ C ]// International Conference on Artificial Intelligence insight and security. Springer, Cham,2020: 179-190.). The shearing attack mode is to shear the rectangular region from the top left corner of the host image and replace the pixel value of the rectangular region with 255. It can be clearly seen that the inventive algorithm is not very different from the algorithm of this document at 9% and 15% shear attacks. However, at 20% shear strength, the inventive algorithm is clearly superior to that of the document, achieving an NC value of 0.95 at a maximum shear strength of 56%, while at a comparative algorithm that does not exceed 0.5, illustrates that the inventive algorithm has a strong resistance against gaussian noise attacks and shear attacks.
In the foregoing embodiments, although the steps are numbered as S1, S2, etc., but only the specific embodiments are given in this application, and those skilled in the art may adjust the execution order of S1, S2, etc. according to the actual situation, which is also within the protection scope of the present invention, and it is understood that some embodiments may include some or all of the above embodiments.
As shown in fig. 9, an image watermarking system 200 according to an embodiment of the present invention includes a decomposition module 210, a binarization module 220, a scrambling and repeating module 230, and an exclusive or operation module 240;
the decomposition module 210 is configured to: decomposing a host image by using a two-dimensional empirical mode decomposition algorithm to obtain a first intrinsic mode function and a margin corresponding to the host image, decomposing the first intrinsic mode function corresponding to the host image by using a singular value decomposition algorithm with a preset window to obtain a plurality of first feature matrices, and decomposing the margin corresponding to the host image by using a singular value decomposition algorithm with a preset window to obtain a plurality of second feature matrices;
the binarization module 220 is configured to: respectively obtaining elements corresponding to the same preset position from each first feature matrix and carrying out binarization to obtain a third feature matrix, and respectively obtaining elements corresponding to the same preset position from each second feature matrix and carrying out binarization to obtain a fourth feature matrix;
the scrambling repetition module 230 is configured to: performing Arnold scrambling on the watermark image corresponding to the host image and repeating to obtain a fifth feature matrix with the size consistent with that of the third feature matrix;
the exclusive-or operation module 240 is configured to: and performing exclusive-or operation on the fifth feature matrix and the third feature matrix to obtain the first feature image, performing exclusive-or operation on the fifth feature matrix and the fourth feature matrix to obtain the second feature image, and performing copyright authentication and/or tampering detection according to the first feature image and the second feature image.
On one hand, the host image is not modified, the integrity of the host image is effectively protected, on the other hand, the advantages of the two-dimensional empirical mode decomposition BEMD and the singular value decomposition SVD are combined, firstly, the good characteristics of the BEMD are fully exerted, and the host image is decomposed into a limited number of inherent modal functions and margins in a self-adaptive mode. Secondly, respectively using singular value decomposition algorithm from the first inherent mode function and the margin corresponding to the host image, and carrying out binarization operation to obtain a third feature matrix
And a fourth feature matrix; the watermark image is then secured using an Arnold transform, the first feature image F 'being constructed by an exclusive OR operation (XOR)'IAnd the second characteristic image, the first characteristic image and the second characteristic image can be safely stored in a copyright authentication database for further copyright authentication and tampering detection. The application is proved to be not only usurped through a large number of experimental results and comparison with the existing watermarking algorithmThe method has good performance in detection and also has good performance in resisting various attacks, particularly shearing attack, Gaussian noise, median filtering, image enhancement and the like, and the robustness is improved.
Preferably, in the above technical solution, the apparatus further includes a detection module, and the decomposition module is further configured to: decomposing a host image to be detected corresponding to a host image by using a two-dimensional empirical mode decomposition algorithm to obtain a first intrinsic mode function and a margin corresponding to the host image to be detected, decomposing the first intrinsic mode function corresponding to the host image to be detected by using a singular value decomposition algorithm with a preset window to obtain a plurality of sixth feature matrices, and decomposing the margin corresponding to the host image to be detected by using a singular value decomposition algorithm with a preset window to obtain a plurality of seventh feature matrices;
the binarization module 220 is further configured to: respectively obtaining elements corresponding to the same preset position from each sixth feature matrix and carrying out binarization to obtain an eighth feature matrix, and respectively obtaining elements corresponding to the same preset position from each seventh feature matrix and carrying out binarization to obtain a ninth feature matrix;
the exclusive-or operation module 240 is further configured to: performing exclusive-or operation on the first characteristic image and the eighth characteristic matrix of the host image to obtain a third characteristic image; performing exclusive-or operation on the second characteristic image of the host image and the ninth characteristic matrix to obtain a fourth characteristic image;
the binarization module 220 is further configured to: dividing the fourth characteristic image according to the size of a preset window to obtain a plurality of tenth characteristic matrixes, adding elements at the same position in each tenth characteristic matrix, carrying out binarization to obtain an image with the same size as the watermark image corresponding to the host image, carrying out Arnold inverse transformation to obtain a watermark image to be detected, and dividing the third characteristic image according to the size of the preset window to obtain a plurality of eleventh characteristic matrixes;
the detection module is used for: and respectively calculating an NC (numerical control) value between each eleventh feature matrix and the watermark image to be detected, judging that the host image to be detected is tampered when the NC value smaller than the preset threshold exists, and judging that the host image to be detected is not tampered when the NC value smaller than the preset threshold does not exist.
Preferably, in the above technical solution, the detecting module is further configured to: and when the NC value smaller than the preset threshold exists, dividing the NC value into two types by using a K-means algorithm, obtaining the type with a smaller average value of the NC value, and determining the position of the type corresponding to the host image H' to be detected as the tampered position.
Preferably, in the above technical solution, the detecting module is further configured to: and obtaining an NC value for representing the strong and weak robustness according to the watermark image to be detected and the watermark image.
The above-mentioned steps for realizing the corresponding functions of each parameter and each unit module in the image watermarking system 200 of the present invention may refer to each parameter and step in the above-mentioned embodiment of an image watermarking method, which are not described herein again.
An electronic device according to an embodiment of the present invention includes a memory, a processor, and a program stored in the memory and running on the processor, where the processor implements any of the steps of the image watermarking method implemented in the foregoing when executing the program.
The electronic device may be a computer, a mobile phone, or the like, and correspondingly, the program is computer software or a mobile phone APP, and the parameters and the steps in the electronic device of the present invention may refer to the parameters and the steps in the above embodiment of the image watermarking method, which are not described herein again.
As will be appreciated by one skilled in the art, the present invention may be embodied as a system, method or computer program product.
Accordingly, the present disclosure may be embodied in the form of: may be embodied entirely in hardware, entirely in software (including firmware, resident software, micro-code, etc.) or in a combination of hardware and software, and may be referred to herein generally as a "circuit," module "or" system. Furthermore, in some embodiments, the invention may also be embodied in the form of a computer program product in one or more computer-readable media having computer-readable program code embodied in the medium.
Any combination of one or more computer-readable media may be employed. The computer readable medium may be a computer readable signal medium or a computer readable storage medium. A computer readable storage medium may be, for example, but not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination of the foregoing. More specific examples (a non-exhaustive list) of the computer-readable storage medium include an electrical connection having one or more wires, a portable computer diskette, a hard disk, a Random Access Memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing. In the context of this document, a computer readable storage medium may be any tangible medium that can contain, or store a program for use by or in connection with an instruction execution system, apparatus, or device.
Although embodiments of the present invention have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting the present invention, and that variations, modifications, substitutions and alterations can be made to the above embodiments by those of ordinary skill in the art within the scope of the present invention.
Claims (9)
1. An image watermarking method, comprising:
decomposing a host image by using a two-dimensional empirical mode decomposition algorithm to obtain a first intrinsic mode function and a margin corresponding to the host image, decomposing the first intrinsic mode function corresponding to the host image by using a singular value decomposition algorithm with a preset window to obtain a plurality of first feature matrices, and decomposing the margin corresponding to the host image by using a singular value decomposition algorithm with a preset window to obtain a plurality of second feature matrices;
respectively obtaining elements corresponding to the same preset position from each first feature matrix and carrying out binarization to obtain a third feature matrix, and respectively obtaining elements corresponding to the same preset position from each second feature matrix and carrying out binarization to obtain a fourth feature matrix;
performing Arnold scrambling on the watermark image corresponding to the host image and repeating to obtain a fifth feature matrix with the size consistent with that of the third feature matrix;
and performing exclusive-or operation on the fifth feature matrix and the third feature matrix to obtain the first feature image, performing exclusive-or operation on the fifth feature matrix and the fourth feature matrix to obtain the second feature image, and performing copyright authentication and/or tampering detection according to the first feature image and the second feature image.
2. An image watermarking method according to claim 1, further comprising:
decomposing a host image to be detected corresponding to a host image by using a two-dimensional empirical mode decomposition algorithm to obtain a first inherent mode function and a margin corresponding to the host image to be detected;
decomposing the first inherent mode function corresponding to the host image to be detected by using a singular value decomposition algorithm with a preset window to obtain a plurality of sixth feature matrices, and decomposing the surplus corresponding to the host image to be detected by using the singular value decomposition algorithm with the preset window to obtain a plurality of seventh feature matrices;
respectively obtaining elements corresponding to the same preset position from each sixth feature matrix, and performing binarization to obtain an eighth feature matrix;
respectively obtaining elements corresponding to the same preset position from each seventh feature matrix, and performing binarization to obtain a ninth feature matrix;
performing exclusive-or operation on the first characteristic image and the eighth characteristic matrix of the host image to obtain a third characteristic image;
performing exclusive-or operation on the second characteristic image of the host image and the ninth characteristic matrix to obtain a fourth characteristic image;
dividing the fourth characteristic image according to the size of a preset window to obtain a plurality of tenth characteristic matrixes, adding elements at the same position in each tenth characteristic matrix, carrying out binarization to obtain an image with the same size as the watermark image corresponding to the host image, and carrying out Arnold inverse transformation to obtain a watermark image to be detected;
and segmenting the third characteristic image according to the size of a preset window to obtain a plurality of eleventh characteristic matrixes, respectively calculating an NC (numerical control) value between each eleventh characteristic matrix and the watermark image to be detected, judging that the host image to be detected is tampered when the NC value smaller than a preset threshold exists, and judging that the host image to be detected is not tampered when the NC value smaller than the preset threshold does not exist.
3. An image watermarking method according to claim 2, further comprising: and when the NC value smaller than the preset threshold exists, dividing the NC value into two types by using a K-means algorithm, acquiring the type with a smaller average value of the NC value, and determining the position of the type corresponding to the host image to be detected as the tampered position.
4. An image watermarking method according to claim 2, further comprising: and obtaining an NC value for representing the strong and weak robustness according to the watermark image to be detected and the watermark image.
5. An image watermarking system is characterized by comprising a decomposition module, a binarization module, a scrambling repetition module and an exclusive OR operation module;
the decomposition module is configured to: decomposing a host image by using a two-dimensional empirical mode decomposition algorithm to obtain a first intrinsic mode function and a margin corresponding to the host image, decomposing the first intrinsic mode function corresponding to the host image by using a singular value decomposition algorithm with a preset window to obtain a plurality of first feature matrices, and decomposing the margin corresponding to the host image by using a singular value decomposition algorithm with a preset window to obtain a plurality of second feature matrices;
the binarization module is used for: respectively obtaining elements corresponding to the same preset position from each first feature matrix and carrying out binarization to obtain a third feature matrix, and respectively obtaining elements corresponding to the same preset position from each second feature matrix and carrying out binarization to obtain a fourth feature matrix;
the scrambling repetition module is to: performing Arnold scrambling on the watermark image corresponding to the host image and repeating to obtain a fifth feature matrix with the size consistent with that of the third feature matrix;
the XOR operation module is to: and performing exclusive-or operation on the fifth feature matrix and the third feature matrix to obtain the first feature image, performing exclusive-or operation on the fifth feature matrix and the fourth feature matrix to obtain the second feature image, and performing copyright authentication and/or tampering detection according to the first feature image and the second feature image.
6. The image watermarking system of claim 5, further comprising a detection module, wherein the decomposition module is further configured to: decomposing a host image to be detected corresponding to a host image by using a two-dimensional empirical mode decomposition algorithm to obtain a first intrinsic mode function and a margin corresponding to the host image to be detected, decomposing the first intrinsic mode function corresponding to the host image to be detected by using a singular value decomposition algorithm with a preset window to obtain a plurality of sixth feature matrices, and decomposing the margin corresponding to the host image to be detected by using a singular value decomposition algorithm with a preset window to obtain a plurality of seventh feature matrices;
the binarization module is further configured to: respectively obtaining elements corresponding to the same preset position from each sixth feature matrix and carrying out binarization to obtain an eighth feature matrix, and respectively obtaining elements corresponding to the same preset position from each seventh feature matrix and carrying out binarization to obtain a ninth feature matrix;
the XOR operation module is further to: performing exclusive-or operation on the first characteristic image and the eighth characteristic matrix of the host image to obtain a third characteristic image; performing exclusive-or operation on the second characteristic image of the host image and the ninth characteristic matrix to obtain a fourth characteristic image;
the binarization module is further configured to: dividing the fourth characteristic image according to the size of a preset window to obtain a plurality of tenth characteristic matrixes, adding elements at the same position in each tenth characteristic matrix, carrying out binarization to obtain an image with the same size as the watermark image corresponding to the host image, carrying out Arnold inverse transformation to obtain a watermark image to be detected, and dividing the third characteristic image according to the size of the preset window to obtain a plurality of eleventh characteristic matrixes;
the detection module is used for: and respectively calculating an NC (numerical control) value between each eleventh feature matrix and the watermark image to be detected, judging that the host image to be detected is tampered when the NC value smaller than the preset threshold exists, and judging that the host image to be detected is not tampered when the NC value smaller than the preset threshold does not exist.
7. The image watermarking system of claim 6, wherein the detection module is further configured to: and when the NC value smaller than the preset threshold exists, dividing the NC value into two types by using a K-means algorithm, acquiring the type with a smaller average value of the NC value, and determining the position of the type corresponding to the host image to be detected as the tampered position.
8. The image watermarking system of claim 7, wherein the detection module is further configured to: and obtaining an NC value for representing the strong and weak robustness according to the watermark image to be detected and the watermark image.
9. An electronic device comprising a memory, a processor and a program stored on the memory and running on the processor, wherein the steps of an image watermarking method according to any of claims 1 to 4 are implemented when the program is executed by the processor.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110443175.XA CN113012020B (en) | 2021-04-23 | 2021-04-23 | Image watermarking method, system and electronic equipment |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110443175.XA CN113012020B (en) | 2021-04-23 | 2021-04-23 | Image watermarking method, system and electronic equipment |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN113012020A true CN113012020A (en) | 2021-06-22 |
| CN113012020B CN113012020B (en) | 2023-07-21 |
Family
ID=76389079
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110443175.XA Active CN113012020B (en) | 2021-04-23 | 2021-04-23 | Image watermarking method, system and electronic equipment |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113012020B (en) |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102129659A (en) * | 2011-03-17 | 2011-07-20 | 浙江工商大学 | Robust zero-watermarking algorithm based on singular value decomposition, Harr wavelet transformation and mean value calculation |
| CN103106636A (en) * | 2013-02-28 | 2013-05-15 | 大连民族学院 | Gray image watermark embedding and extracting method based on singular value decomposition |
| US20170109858A1 (en) * | 2015-10-15 | 2017-04-20 | Ningbo University | Zero-watermarking registration and detection method for HEVC video streaming against requantization transcoding |
| CN110659651A (en) * | 2019-09-05 | 2020-01-07 | 淮阴工学院 | Tampering positioning reversible image watermarking method based on SIFT |
-
2021
- 2021-04-23 CN CN202110443175.XA patent/CN113012020B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102129659A (en) * | 2011-03-17 | 2011-07-20 | 浙江工商大学 | Robust zero-watermarking algorithm based on singular value decomposition, Harr wavelet transformation and mean value calculation |
| CN103106636A (en) * | 2013-02-28 | 2013-05-15 | 大连民族学院 | Gray image watermark embedding and extracting method based on singular value decomposition |
| US20170109858A1 (en) * | 2015-10-15 | 2017-04-20 | Ningbo University | Zero-watermarking registration and detection method for HEVC video streaming against requantization transcoding |
| CN110659651A (en) * | 2019-09-05 | 2020-01-07 | 淮阴工学院 | Tampering positioning reversible image watermarking method based on SIFT |
Also Published As
| Publication number | Publication date |
|---|---|
| CN113012020B (en) | 2023-07-21 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Ansari et al. | Multipurpose image watermarking in the domain of DWT based on SVD and ABC | |
| Gangadhar et al. | An evolutionary programming approach for securing medical images using watermarking scheme in invariant discrete wavelet transformation | |
| Qi et al. | A singular-value-based semi-fragile watermarking scheme for image content authentication with tamper localization | |
| Lei et al. | Reversible watermarking scheme for medical image based on differential evolution | |
| Ernawan et al. | An improved image watermarking by modifying selected DWT-DCT coefficients | |
| CN104599225A (en) | Image watermarking embedding and extracting method based on singular value decomposition and principal component analysis | |
| CN110599388B (en) | Blind robust digital watermark embedding and detecting method based on positioning point assistance | |
| Liu et al. | Robust blind image watermarking based on chaotic mixtures | |
| CN112700363B (en) | Self-adaptive visual watermark embedding method and device based on region selection | |
| CN113538197B (en) | Watermark extraction method, watermark extraction device, storage medium and electronic equipment | |
| Ansari et al. | Secured and optimized robust image watermarking scheme | |
| Chang et al. | Robust image authentication using content based compression | |
| Yamato et al. | Digital image watermarking method using between-class variance | |
| CN106600516B (en) | Image embedding method based on digital fingerprint | |
| CN113012020B (en) | Image watermarking method, system and electronic equipment | |
| Huynh-The et al. | Robust image watermarking framework powered by convolutional encoder-decoder network | |
| CN116051349A (en) | BRISK-FREAK-based medical image watermarking method | |
| Sarkar et al. | Comparison of various Edge Detection Techniques for maximum data hiding using LSB Algorithm | |
| Nikbakht et al. | Targeted watermark removal of a SVD-based image watermarking scheme | |
| Chrysochos et al. | Robust watermarking of digital images based on chaotic mapping and DCT | |
| Che et al. | Semi-fragile image watermarking algorithm based on visual features | |
| Pal et al. | A comparative study between LSB and modified bit replacement (MBR) watermarking technique in spatial domain for biomedical image security | |
| Takore et al. | Robust image watermarking scheme using population-based stochastic optimization technique | |
| Ghazali et al. | Review of watermarking techniques for medical images | |
| Naseem et al. | Novel technique for capacity maximizing in digital watermarking using fuzzy rule base |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |