CN108648130B - Totally-blind digital watermarking method with copyright protection and tampering positioning functions - Google Patents
Totally-blind digital watermarking method with copyright protection and tampering positioning functions Download PDFInfo
- Publication number
- CN108648130B CN108648130B CN201810299604.9A CN201810299604A CN108648130B CN 108648130 B CN108648130 B CN 108648130B CN 201810299604 A CN201810299604 A CN 201810299604A CN 108648130 B CN108648130 B CN 108648130B
- Authority
- CN
- China
- Prior art keywords
- watermark
- cosine transform
- discrete cosine
- sub
- region
- 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.)
- Active
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
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2201/00—General purpose image data processing
- G06T2201/005—Image watermarking
- G06T2201/0065—Extraction of an embedded watermark; Reliable detection
Landscapes
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Editing Of Facsimile Originals (AREA)
- Image Processing (AREA)
Abstract
The invention discloses a totally blind digital watermarking method with copyright protection and tampering positioning functions, which divides an approximation sub-image of an original gray level image into non-overlapping 8 x 8 sub-blocks, and divides each sub-block into 4 x 4 areas; performing discrete cosine transform on each region; creating a characteristic watermark and a self-embedded characteristic watermark according to the discrete cosine transform coefficient matrixes of different areas; during extraction, discrete cosine transform is carried out on each region in each sub-block in an approximate subgraph of the watermark image in the same mode; secondly, blind extracting a characteristic watermark and an authentication watermark according to discrete cosine transform coefficient matrixes of different areas; calculating a normalized correlation coefficient between the characteristic watermark and the authentication watermark for copyright protection, and comparing the characteristic watermark and the authentication watermark bit by bit to realize totally blind tampering and positioning; the method has the advantages that the method can realize the double functions of copyright protection and tampering positioning only by embedding one digital watermark, does not need any information of the original image and the original digital watermark when extracting the watermark, and can realize the blind detection of the watermark.
Description
Technical Field
The invention relates to a digital media information security technology, in particular to a totally blind digital watermarking method with copyright protection and tampering positioning functions.
Background
With the rapid development of digital media and network technologies, digital watermarking technology has become an effective means for media copyright protection and content integrity authentication. However, most of the existing digital watermarking technologies are single watermarking algorithms, and the existing digital watermarking technologies often have the limitation of single function, namely only copyright protection or only content authentication can be realized. Obviously, digital watermarking algorithms with multiple functions have stronger requirements in practical applications.
The multifunctional digital watermark is characterized in that watermarks with different properties are embedded in the same digital image so as to achieve different application purposes. For example: a visible watermark and an invisible watermark are embedded in a digital image, wherein the visible watermark is used for copyright notification, and the invisible watermark is used for copyright protection. For another example: a fragile watermark and a robust watermark are embedded in a digital image, wherein the fragile watermark is used for tamper indication and content authentication, and the robust watermark is used for copyright protection and the like.
In 2009, leaf astronomy provides a multifunctional double watermark algorithm on electronic and information science, and the algorithm embeds robust watermarks in singular values of image blocks and embeds fragile watermarks in least significant bits of airspace pixels of images containing the robust watermarks, so that double functions of copyright protection and content authentication are realized. In 2012, the forest and military sea proposed a multifunctional watermarking algorithm based on error correction coding and space-frequency domain combination on the basis of Hangzhou electronic technology university school newspaper, and the algorithm performs discrete wavelet transform on a carrier image, embeds a robust watermark into a diagonal sub-band of a low-frequency sub-band, then performs Hamming code coding by using a bit plane of the image and the robust watermark, and embeds a generated check code into the least significant bit of the image to realize embedding of a fragile watermark. In 2013, Yuxiaoqing provides a multifunctional color image dual watermark algorithm based on lifting wavelets in 'computer and application software', the algorithm utilizes different characteristics of three color channels of a color RGB image, and embeds robust watermarks in a low-frequency coefficient of a blue component and semi-fragile watermarks in an intermediate-frequency coefficient of a green component, so that multiple functions of image copyright protection, content authentication and tampering positioning are realized. In 2015, the Wang Jue in the university of Master Min and Min provides a color image multifunctional dual watermark algorithm, the algorithm quantitatively embeds copyright watermarks into discrete wavelet low-frequency coefficient block singular values of a blue component by extracting three channels of RGB of a color image, and then adaptively and quantitatively embeds authentication watermarks generated from a red component into green component block singular values, and experimental results show that the algorithm has dual functions of copyright protection and content authentication.
It can be seen that in order to achieve the multi-functional purpose of digital watermarking, the above proposed digital watermarking algorithm with multiple functions needs to embed both a robust watermark and a fragile watermark (or authentication watermark) into an image carrier. Obviously, compared with a single watermark algorithm, the simultaneous embedding of dual watermarks inevitably leads to a reduction in image quality; moreover, because the double watermarks are embedded in the same image carrier, certain interference must exist between the robust watermark and the fragile watermark, and the extraction effect of the subsequent watermark is influenced. In addition, in order to realize functions such as copyright protection and content authentication, correlation measurement and comparison between the extracted robust watermark and fragile watermark and the original digital watermark are required, and blind detection of the digital watermark cannot be realized.
Therefore, in order to overcome the related limitations of the above digital watermarking algorithm with multiple functions, the research on a digital watermarking algorithm which can realize the multiple functions only by embedding a single watermark and can realize the blind detection has practical application significance and is a necessary development direction in the technical field of information security.
Disclosure of Invention
The technical problem to be solved by the invention is to provide a totally blind digital watermarking method with copyright protection and tampering positioning functions, wherein the double functions of copyright protection and tampering positioning can be realized only by embedding one digital watermark at an embedding end, and on the premise of ensuring the quality of a watermark image, the method has ideal robustness for conventional image processing, has good sensibility to malicious attack, does not need any information of an original image and the original digital watermark when the digital watermark is extracted at a detection end, and can realize the totally blind detection of the digital watermark.
The technical scheme adopted by the invention for solving the technical problems is as follows: a totally blind digital watermarking method with copyright protection and tampering positioning functions is characterized by comprising the following main processing procedures:
at a digital watermark embedding end, performing one-level discrete wavelet decomposition on an original 8-bit gray level image, dividing an obtained approximation subgraph into a plurality of non-overlapping subblocks with the size of 8 multiplied by 8, and dividing each subblock in the approximation subgraph into 4 non-overlapping regions with the size of 4 multiplied by 4; then, performing discrete cosine transform on each region in each sub-block in the approximation subgraph; then, according to discrete cosine transform direct current coefficients of upper left corner regions and upper right corner regions of all sub-blocks in the approximation subgraph, a characteristic watermark is created and obtained; according to the discrete cosine transform alternating current coefficient at the same position of the left lower corner region and the right lower corner region of each sub-block in the approximation subgraph and each bit in the created characteristic watermark, self-embedding one bit in the characteristic watermark in each sub-block in the approximation subgraph; finally, obtaining a watermark image embedded with the digital watermark through inverse discrete cosine transform and first-level inverse discrete wavelet transform;
at a digital watermark extraction and detection end, performing one-level discrete wavelet decomposition on a watermark image embedded with a digital watermark, dividing an obtained approximation subgraph into a plurality of non-overlapping subblocks with the size of 8 multiplied by 8, and dividing each subblock in the approximation subgraph into 4 non-overlapping regions with the size of 4 multiplied by 4; then, performing discrete cosine transform on each region in each sub-block in the approximation subgraph; secondly, blind extraction is carried out according to discrete cosine transform direct current coefficients of upper left corner regions and upper right corner regions of all sub-blocks in the approximation subgraph to obtain a characteristic watermark; blind extraction is carried out according to discrete cosine transform alternating current coefficients at the same positions of a lower left corner region and a lower right corner region of each sub-block in the approximation subgraph to obtain an authentication watermark; and finally, performing copyright protection by calculating a normalized correlation coefficient between the characteristic watermark obtained by blind extraction and the authentication watermark obtained by blind extraction, and realizing totally blind tampering and positioning by comparing the characteristic watermark obtained by blind extraction and the authentication watermark obtained by blind extraction bit by bit.
The totally blind digital watermarking method with copyright protection and tampering positioning functions comprises two parts, namely digital watermarking embedding, digital watermarking extracting and detecting;
the digital watermark embedding part comprises the following specific steps:
firstly, selecting an original 8-bit gray image, and recording the image as F, wherein the size of the F is I multiplied by J; then, performing one-level discrete wavelet decomposition on the F to obtain three detail subgraphs and an approximation subgraph of the F, and marking the approximation subgraph of the F as FA with the size of FAThen, the FA is divided intoThe sub-blocks with the size of 8 multiplied by 8 and not overlapped with each other are marked as FABk(ii) a Dividing each sub-block in FA into 4 non-overlapping regions of 4 × 4 size, and dividing FABkThe upper left corner area in (1) is taken as FABk1 st area in (b) is noted asWill FABkThe upper right corner area in (1) is taken as FABkThe 2 nd area in (b) is noted asWill FABkThe lower left corner area of middle is taken as FABkThe 3 rd region of (1) is noted asWill FABkThe lower right corner area of middle is taken as FABkThe 4 th region of (1) is noted asWherein I represents the horizontal resolution of F, J represents the vertical resolution of F, and the symbolFor rounding down the arithmetic sign, k is a positive integer,
step (r 2) of performing discrete cosine transform on each region in each sub-block in the FA to obtain a discrete cosine transform coefficient matrix of each region in each sub-block in the FA, and performing discrete cosine transform on each region in each sub-block in the FAIs expressed as a matrix of discrete cosine transform coefficientsWill be provided withIs expressed as a matrix of discrete cosine transform coefficientsWill be provided withIs expressed as a matrix of discrete cosine transform coefficientsWill be provided withIs expressed as a matrix of discrete cosine transform coefficientsWherein the content of the first and second substances, andthe dimensions of (A) are all 4 x 4;
step 3, creating a characteristic watermark W, which comprises the following specific steps: calculating the mean value of the discrete cosine transform direct current coefficient of each sub-block in the FA, and converting FAB into FABkThe mean value of the discrete cosine transform direct current coefficient is recorded as FADk,Then, according to the mean value of the discrete cosine transform direct current coefficients of all the sub-blocks in the FA, a characteristic watermark W is created and obtained, and the k bit of the W is marked as Wk,Wherein W has a length of To representRow 1 and column 1 elements in (1),is composed ofThe direct current coefficient of the discrete cosine transform of (1),to representRow 1 and column 1 elements in (1),is composed ofD.c. coefficient of discrete cosine transform of (delta)1Is a set first threshold;
step 4, self-embedding a characteristic watermark W, which comprises the following specific processes: self-embedding one bit of the characteristic watermark W in each sub-block in the FA, for FABkTaking out WkIf W iskIs equal to 0 andthen orderThen orderRealization of WkIs self-embedding, thenWhen it is established, the pairAndby further adjustment, it is instantly orderedAnd order
If W iskIs equal to 0 andthen pairAndrealize W without processingkIs self-embedding, thenWhen it is established, the pairAndby further adjustment, it is instantly orderedAnd order
If Wk1 andthen orderThen orderRealization of WkIs self-embedding, thenWhen it is established, the pairAndby further adjustment, it is instantly orderedAnd order
If W isk1 andthen pairAndrealize W without processingkIs self-embedding, thenWhen it is established, the pairAndby further adjustment, it is instantly orderedAnd order
Wherein, the first and the second end of the pipe are connected with each other,to representRow 2 and column 2 elements in (1),to representLine 2 and column 2 elements in (1), G1 and G2 are all introduced intermediate variables, abs () is an absolute value function, δ2In order to set the second threshold value, wherein, the symbol is an assignment symbol;
firstly, carrying out inverse discrete cosine transform on each region on the basis of the first step 4 by a first step 5; then, recombining to obtain each sub-block embedded with the digital watermark, and further recombining to obtain an approximation subgraph embedded with the digital watermark; then, performing one-level inverse discrete wavelet transform on the approximation subgraph embedded with the digital watermark and the three detail subgraphs of F to obtain a watermark image embedded with the digital watermark;
the digital watermark extracting and detecting part comprises the following specific steps:
step two _1, the watermark image embedded with the digital watermark is an 8-bit gray image which is marked as TF, and the size of TF is I multiplied by J; then, carrying out one-level discrete wavelet decomposition on the TF to obtain three detail subgraphs and an approximation subgraph of the TF, and marking the approximation subgraph of the TF as TFA with the size of TFAFollowed by cleavage of TFA intoThe k-th sub-block in TFA is denoted as TFABk(ii) a Dividing each sub-block of TFA into 4 non-overlapping regions of 4 × 4 size, and dividing TFAB into two regionskThe upper left corner region in (1) is taken as TFABk1 st area in (b) is noted asWill TFABkThe upper right corner region in (1) is used as TFABkThe 2 nd area in (b) is noted asWill TFABkThe lower left corner region in (1) is taken as TFABkThe 3 rd area in (1) is markedWill TFABkThe lower right corner region in (1) is used as TFABkThe 4 th region of (1) is noted asWherein, the symbolIs a sign of a rounding-down operation, k is a positive integer,
step 2, performing discrete cosine transform on each region in each sub-block in the TFA to obtain a discrete cosine transform coefficient matrix of each region in each sub-block in the TFA, and performing discrete cosine transform on each region in each sub-block in the TFAIs expressed as a matrix of discrete cosine transform coefficientsWill be provided withIs expressed as a matrix of discrete cosine transform coefficientsWill be provided withIs expressed as a matrix of discrete cosine transform coefficientsWill be provided withIs expressed as a matrix of discrete cosine transform coefficientsWherein the content of the first and second substances, andthe dimensions of (A) are all 4 x 4;
step two-3, blindly extracting the characteristic watermark TW, which comprises the following specific processes: calculating the mean value of the discrete cosine transform direct current coefficient of each sub-block in TFA, and converting TFAB into TFABkThe mean value of the discrete cosine transform direct current coefficient is recorded as TFADk,Then, according to the mean value of the discrete cosine transform direct current coefficients of all the sub-blocks in the TFA, a characteristic watermark TW is obtained through blind extraction, and the k-th bit of the TW is recorded as TWk,Wherein the length of TW is To representRow 1 and column 1 elements in (1),is composed ofThe direct current coefficient of the discrete cosine transform of (1),to representRow 1 and column 1 elements in (1),is composed ofD.c. coefficient of discrete cosine transform of (delta)1Is a set first threshold;
step 4, blind extraction of authentication watermark TW*The specific process is as follows: blindly extracting one bit in each sub-block in TFA to obtain the authentication watermark TW*For TFABkExtracting to obtain TW*K-th bit TW in* k,Wherein TW*Has a length of To representRow 2 and column 2 elements in (1),to representRow 2, column 2 elements in (1);
step 5, according to TW and TW*Copyright protection is carried out, and the specific process is as follows: computing TW and TW*Normalized correlation coefficient between them, denoted as ρ (TW, TW)*) Using ρ (TW, TW)*) To perform copyright protection;
according to TW and TW*And (3) carrying out tampering positioning, wherein the specific process is as follows: according to each bit in TW and TW*Determines whether each sub-block in TFA has been tampered with, for TFABkIf TWkAnd TW* kEqual, then TFAB is determinedkHas not been tampered with; if TWkAnd TW* kIf not, TFAB is determinedkAfter being tampered, the full-blind tampering positioning is realized.
In the step (II-5),wherein the content of the first and second substances,represents the average of all elements in the TW,represents TW*Average value of all elements in (1).
Compared with the prior art, the invention has the advantages that:
1) the method can extract and detect the digital watermark only by the watermark image at the digital watermark extraction and detection end, does not need any related information of the original image or the original digital watermark at the digital watermark embedding end, not only saves the transmission cost and the storage cost required when the original image or the original digital watermark is transmitted by the digital watermark embedding end, but also avoids the passive attack or the explanation attack commonly existing on the Internet, and completely realizes the blind detection function.
2) The method can realize the double functions of copyright protection and tampering positioning only by embedding one characteristic watermark at the embedding end of the digital watermark, thereby avoiding the interference among multiple watermarks; and the creation and the embedding of the characteristic watermark are respectively carried out in different areas of each sub-block in the image, thereby avoiding the influence of the characteristic watermark embedding process on the self extraction.
3) The method can simultaneously realize the double functions of copyright protection and tampering positioning, and has wider practicability.
4) On the premise of ensuring the quality of the watermark image, the method has ideal robustness for conventional image processing and good sensitivity for malicious attacks.
Drawings
FIG. 1 is an original Lena grayscale image with a resolution of 512 × 512;
FIG. 2 is a watermark Lena gray scale image embedded with a watermark obtained by processing the original Lena gray scale image shown in FIG. 1 by using the method of the present invention;
fig. 3 is a watermark Lena gray level image obtained by performing histogram equalization processing on the watermark Lena gray level image shown in fig. 2;
fig. 4 is a JPEG lossy compression processing (the compression quality factor is selected to be 5%) is performed on the watermark Lena gray scale image shown in fig. 2, and the watermark Lena gray scale image after JPEG lossy compression is obtained;
fig. 5 is a watermark Lena gray scale image obtained by performing noise superposition processing (the superposed noise is gaussian distribution noise with a mean value of 0 and a variance of 0.001) on the watermark Lena gray scale image shown in fig. 2, and superposing gaussian noise;
fig. 6 is a watermark Lena gray scale image after median filtering, which is obtained by performing median filtering processing on the watermark Lena gray scale image shown in fig. 2 (the window size of the median filter is selected to be [5 × 5 ]);
FIG. 7 is a gray scale babon image of dimensions 128 × 128;
fig. 8 is a watermark Lena gray scale image obtained after replacing the (109:236,149:276) area of the watermark Lena gray scale image shown in fig. 2 with the baboon gray scale image shown in fig. 7;
fig. 9 is a detection result of tampering positioning on the watermark Lena grayscale image shown in fig. 8 by using the method of the present invention;
fig. 10 is a watermark Lena gray scale image obtained after the replacement operation, in which pixel points in the (129: 256) region of the watermark Lena gray scale image shown in fig. 2 are replaced with pixel points in another region (229: 356) of the self image;
fig. 11 is a detection result of tampering positioning on the watermark Lena grayscale image shown in fig. 10 by using the method of the present invention;
FIG. 12 is a watermark Lena gray scale image obtained after replacing (229:356,379:506) area of the original Lena gray scale image shown in FIG. 1 by (329:456,229:356) area of the watermark Lena gray scale image shown in FIG. 2;
fig. 13 is a detection result of tampering positioning on the watermark Lena grayscale image shown in fig. 12 by using the method of the present invention;
fig. 14 is a watermark Lena gray scale image obtained after the region cutting operation by cutting off image blocks of the watermark Lena gray scale image shown in fig. 2, which are located in the (214:267,189:227) region;
fig. 15 is a detection result of tampering positioning on the watermark Lena grayscale image shown in fig. 14 by using the method of the present invention;
fig. 16 is a watermark Lena gray scale image obtained after a pixel value increase operation is performed on an image block in which the watermark Lena gray scale image shown in fig. 2 is located in an area (329:456,129: 256);
fig. 17 is a detection result of tampering positioning of the watermark Lena grayscale image shown in fig. 16 by using the method of the present invention;
fig. 18 is a block diagram of an overall implementation of the method of the present invention.
Detailed Description
The invention is described in further detail below with reference to the accompanying examples.
The total implementation block diagram of the totally blind digital watermarking method with copyright protection and tampering positioning functions provided by the invention is shown in fig. 18, and the main processing procedures are as follows:
at a digital watermark embedding end, performing one-level discrete wavelet decomposition on an original 8-bit gray level image, dividing an obtained approximation subgraph into a plurality of non-overlapping subblocks with the size of 8 multiplied by 8, and dividing each subblock in the approximation subgraph into 4 non-overlapping regions with the size of 4 multiplied by 4; then, performing discrete cosine transform on each region in each sub-block in the approximation subgraph; then, according to discrete cosine transform direct current coefficients of the upper left corner region and the upper right corner region of all sub-blocks in the approximation subgraph, a characteristic watermark is created; according to the discrete cosine transform alternating current coefficient at the same position of the left lower corner region and the right lower corner region of each sub-block in the approximation subgraph and each bit in the created characteristic watermark, self-embedding one bit in the characteristic watermark in each sub-block in the approximation subgraph; finally, obtaining a watermark image embedded with the digital watermark through inverse discrete cosine transform and first-level inverse discrete wavelet transform;
at a digital watermark extraction and detection end, performing one-level discrete wavelet decomposition on a watermark image embedded with a digital watermark, dividing an obtained approximation subgraph into a plurality of non-overlapping subblocks with the size of 8 multiplied by 8, and dividing each subblock in the approximation subgraph into 4 non-overlapping regions with the size of 4 multiplied by 4; then, performing discrete cosine transform on each region in each sub-block in the approximation subgraph; secondly, blind extraction is carried out according to discrete cosine transform direct current coefficients of upper left corner regions and upper right corner regions of all sub-blocks in the approximation subgraph to obtain a characteristic watermark; blind extraction is carried out according to discrete cosine transform alternating current coefficients at the same positions of a lower left corner region and a lower right corner region of each sub-block in the approximation subgraph to obtain an authentication watermark; and finally, performing copyright protection by calculating a normalized correlation coefficient between the characteristic watermark obtained by blind extraction and the authentication watermark obtained by blind extraction, and realizing totally blind tampering and positioning by comparing the characteristic watermark obtained by blind extraction and the authentication watermark obtained by blind extraction bit by bit.
The totally blind digital watermarking method with copyright protection and tampering positioning functions comprises two parts, namely digital watermarking embedding, digital watermarking extracting and detecting;
the digital watermark embedding part comprises the following specific steps:
firstly, selecting an original 8-bit gray image, and recording the image as F, wherein the size of the F is I multiplied by J; then, carrying out one-level discrete wavelet decomposition on the F to obtain three detail subgraphs and an approximation subgraph of the F, and marking the approximation subgraph of the F as FA with the size of FAThen dividing FA into blocks in raster scan orderThe sub-blocks with the size of 8 multiplied by 8 and not overlapped with each other are marked as FABk(ii) a Dividing each sub-block in FA into 4 non-overlapping regions of 4 × 4 size, and dividing FABkThe upper left corner area in (1) is taken as FABk1 st area in (b) is noted asWill FABkThe upper right corner area in middle is taken as FABkThe 2 nd area in (b) is noted asWill FABkThe lower left corner area of middle is taken as FABkThe 3 rd region of (1) is noted asWill FABkThe lower right corner area of middle is taken as FABkThe 4 th region of (1) is noted asWherein I represents the horizontal resolution of F, J represents the vertical resolution of F, and the symbolIs a sign of a rounding-down operation, k is a positive integer,
step (r 2) of performing discrete cosine transform on each region in each sub-block in the FA to obtain a discrete cosine transform coefficient matrix of each region in each sub-block in the FA, and performing discrete cosine transform on each region in each sub-block in the FAIs expressed as a matrix of discrete cosine transform coefficientsWill be provided withIs expressed as a matrix of discrete cosine transform coefficientsWill be provided withIs expressed as a matrix of discrete cosine transform coefficientsWill be provided withIs expressed as a matrix of discrete cosine transform coefficientsWherein the content of the first and second substances, andare 4 x 4.
Step 3, creating a characteristic watermark W, which comprises the following specific steps: calculating the mean value of the discrete cosine transform direct current coefficient of each sub-block in the FA, and converting FAB into FABkThe mean value of the discrete cosine transform direct current coefficient is recorded as FADk,Then, according to the mean value of the discrete cosine transform direct current coefficients of all the sub-blocks in the FA, a characteristic watermark W is created and obtained, and the k bit of the W is marked as Wk,Wherein W has a length of To representRow 1 and column 1 elements in (1),is composed ofThe direct current coefficient of the discrete cosine transform of (1),to representRow 1 and column 1 elements in (1),is composed ofD.c. coefficient of discrete cosine transform of (delta)1To set the first threshold, in this embodiment, take δ1=1000。
Step 4, self-embedding a characteristic watermark W, which comprises the following specific processes: self-embedding one bit of the characteristic watermark W in each sub-block in the FA, for FABkTaking out WkIf W iskIs equal to 0 andthen orderThen orderI.e. exchangeAndvalue of (a), realizing WkIn order to improve the robustness of the embedded digital watermarkWhen it is established, the pairAndby further adjustment, it is instantly orderedAnd make an order
If W iskIs equal to 0 andthen pairAndrealize W without processingkThen in order to improve the robustness of the embedded digital watermarkWhen it is established, the pairAndby further adjustment, it is instantly orderedAnd order
If W isk1 andthen orderThen orderI.e. exchangeAnd withValue of (a), realizing WkThen in order to improve the robustness of the embedded digital watermarkWhen it is established, the pairAndby further adjustment, it is instantly orderedAnd make an order
If Wk1 andthen pairAndrealize W without processingkThen in order to improve the robustness of the embedded digital watermarkWhen it is established, the pairAndby further adjustment, it is instantly orderedAnd order
Wherein the content of the first and second substances,to representRow 2 and column 2 elements in (1),to representLine 2 and column 2 elements in (1), G1 and G2 are all introduced intermediate variables, abs () is the absolute value function, δ2To set the second threshold, in this embodiment, take δ2=20, Wherein, the symbol is assigned.
Firstly, carrying out inverse discrete cosine transform on each region on the basis of the first step 4 by a first step 5; then, recombining to obtain each sub-block embedded with the digital watermark, and further recombining to obtain an approximation subgraph embedded with the digital watermark; and then, performing one-level inverse discrete wavelet transform on the approximation subgraph embedded with the digital watermark and the three detail subgraphs of the F to obtain a watermark image embedded with the digital watermark.
The digital watermark extracting and detecting part comprises the following specific steps:
step two _1, the watermark image embedded with the digital watermark is an 8-bit gray image which is marked as TF, and the size of TF is I multiplied by J; then, carrying out one-level discrete wavelet decomposition on the TF to obtain three detail subgraphs and one approximation subgraph of the TF, and approximating the TFIs shown as TFA in the figure, the size of TFA isFollowed by a raster scan order of the TFA segmentation intoThe k-th sub-block in TFA is denoted as TFABk(ii) a Dividing each sub-block of TFA into 4 non-overlapping regions of 4 × 4 size, and dividing TFAB into two regionskThe upper left corner region in (1) is taken as TFABk1 st area in (b) is noted asWill TFABkThe upper right corner region in (1) is used as TFABkThe 2 nd area in (b) is noted asWill TFABkThe lower left corner region in (1) is used as TFABkThe 3 rd region of (1) is noted asWill TFABkThe lower right corner region in (1) is used as TFABkThe 4 th area in (1) is marked asWherein the symbolsIs a sign of a rounding-down operation, k is a positive integer,
step 2, performing discrete cosine transform on each region in each sub-block in the TFA to obtain a discrete cosine transform coefficient matrix of each region in each sub-block in the TFA, and performing discrete cosine transform on each region in each sub-block in the TFAIs expressed as a matrix of discrete cosine transform coefficientsWill be provided withIs expressed as a matrix of discrete cosine transform coefficientsWill be provided withIs expressed as a matrix of discrete cosine transform coefficientsWill be provided withIs expressed as a matrix of discrete cosine transform coefficientsWherein the content of the first and second substances, andare all 4 x 4.
Step two-3, blindly extracting the characteristic watermark TW, which comprises the following specific processes: calculating the mean value of the discrete cosine transform direct current coefficient of each sub-block in TFA, and converting TFAB into TFABkThe mean value of the discrete cosine transform direct current coefficient is recorded as TFADk,Then, according to the mean value of the discrete cosine transform direct current coefficients of all the sub-blocks in the TFA, a characteristic watermark TW is obtained through blind extraction, and the kth of the TW is usedThe bit is denoted TWk,Wherein the length of TW is To representRow 1 and column 1 elements in (1),is composed ofThe direct current coefficient of the discrete cosine transform of (1),to representRow 1 and column 1 elements in (1),is composed ofD.c. coefficient of discrete cosine transform of (delta)1To set the first threshold, in this embodiment, take δ1=1000。
Step 4, blind extraction of authentication watermark TW*The specific process is as follows: blindly extracting one bit in each sub-block in TFA to obtain the authentication watermark TW*For TFABkExtracting to obtain TW*K-th bit TW in* k,Wherein TW*Has a length of RepresentRow 2 and column 2 elements in (1),to representRow 2, column 2 elements in (1).
Step 5, according to TW and TW*Copyright protection is carried out, and the specific process is as follows: calculating TW and TW*Normalized correlation coefficient between them, denoted as ρ (TW, TW)*),Wherein, the first and the second end of the pipe are connected with each other,represents the average of all elements in the TW,represents TW*Average of all elements in (a); using ρ (TW, TW)*) For copyright protection.
According to TW and TW*And (3) carrying out tampering positioning, wherein the specific process is as follows: according to each bit in TW and TW*Determines whether each sub-block in TFA has been tampered with, for TFABkIf TWkAnd TW* kEqual, then TFAB is determinedkHas not been tampered with; if TWkAnd TW* kIs not equal, then TFAB is determinedkAfter being tampered, the full-blind tampering positioning is realized.
In order to better illustrate the inventionThe feasibility of the method applied to digital image copyright protection and tampering positioning is realized by taking an original Lena gray image with the resolution of 512 x 512 shown in figure 1 as a test image to perform experimental simulation, wherein the experimental simulation is performed on a Matlab7.5 platform. Wherein the first threshold value delta1A value of 1000, a second threshold value delta2The value is 20. Fig. 2 shows the watermark Lena gray scale image embedded with the watermark obtained by processing the original Lena gray scale image shown in fig. 1 by using the method of the present invention, and it can be seen from fig. 2 that the watermark Lena gray scale image embedded with the watermark has good objective and main quality, the peak signal-to-noise ratio reaches 38.23dB, and the watermark is hardly changed visually, so that the requirement of imperceptibility of the watermark is satisfied. The normalized correlation coefficient between the characteristic watermark extracted by the method and the authentication watermark is 1, which shows that the embedded characteristic watermark can be completely and nondestructively extracted when the Lena gray level image embedded with the watermark is not damaged by any treatment or attack, so the method can be used for copyright protection of digital image works.
In this case, the image quality of the digital image after embedding the watermark and the objective evaluation of the image quality after image processing both adopt peak signal to noise ratio (PSNR),wherein I is 512, J is 512, I is not less than 1 and not more than I, J is not less than 1 and not more than J, F (I, J) represents the pixel value of the pixel with the coordinate position (I, J) in the original gray image F, TF (I, J) represents the pixel value of the pixel with the coordinate position (I, J) in the gray image TF after the watermark is embedded or after the watermark image is processed and tampered, F (I, J) represents the pixel value of the pixel with the coordinate position (I, J) in the gray image TF after the watermark is embedded or after the watermark image is processed and tamperedmaxRepresents the maximum pixel value in the original grayscale image F, and I × J represents the resolution of the original grayscale image F.
The feasibility of the method for digital image copyright protection and tampering positioning is verified by carrying out various processing and tampering operations on the Lena gray-scale image embedded with the watermark.
1) Histogram equalization
Histogram equalization processing is performed on the watermark Lena gray level image shown in fig. 2, so as to obtain a watermark Lena gray level image after histogram equalization processing, as shown in fig. 3. After histogram equalization processing, the pixel value distribution of pixel points in the watermark Lena gray level image is obviously changed, and the peak signal-to-noise ratio (PSNR) is only 19.04 dB. However, the normalized correlation coefficient between the feature watermark extracted by the method and the authentication watermark reaches 0.89, which fully shows that the embedded feature watermark can be well extracted, and shows the robustness of the method.
2) JPEG lossy compression
The watermark Lena gray level image shown in fig. 2 is subjected to JPEG lossy compression processing, and the compression quality factor is selected to be 5%, so that the watermark Lena gray level image subjected to JPEG lossy compression is obtained, as shown in fig. 4. As can be seen from fig. 4, the watermark Lena gray image after JPEG lossy compression exhibits a certain blocking effect, the visual quality is degraded, and the peak signal-to-noise ratio (PSNR) is reduced to 33.62 dB. However, the normalized correlation coefficient between the feature watermark extracted by the method and the authentication watermark reaches 0.83, which fully shows that the embedded feature watermark can be well extracted, and shows the robustness of the method.
3) Superimposed gaussian (Gauss) noise
The watermark Lena grayscale image shown in fig. 2 is subjected to noise superposition processing, and the superposed noise is gaussian distribution noise with a mean value of 0 and a variance of 0.001, so that a watermark Lena grayscale image after gaussian noise superposition is obtained, as shown in fig. 5. As can be seen from fig. 5, the visual quality of the Lena grayscale image of the watermark after being superimposed with gaussian noise is severely degraded, and the peak signal-to-noise ratio (PSNR) is only 29.41 dB. However, the normalized correlation coefficient between the feature watermark extracted by the method and the authentication watermark has 0.96, which shows that the embedded feature watermark is hardly influenced.
4) Median filtering
The watermark Lena grayscale image shown in fig. 2 is subjected to median filtering, and the window size of the median filter is selected to be [5 × 5], so as to obtain the watermark Lena grayscale image after median filtering, as shown in fig. 6. As can be seen from fig. 6, the detail information of the median-filtered watermark Lena gray scale image is blurred, and the peak signal-to-noise ratio (PSNR) drops to 32.89 dB. However, the normalized correlation coefficient between the feature watermark extracted by the method and the authentication watermark is still 0.86, which indicates that the embedded feature watermark can be extracted well and used for copyright protection of the digital image works.
In order to further explain the versatility of the method of the present invention, a malicious tampering detection experiment is performed on the watermark Lena grayscale image shown in fig. 2, and the detected tampered sub-blocks are displayed with black marks. The malicious tampering of the image mainly comprises operations of replacement, cutting, pixel value increase and the like.
Area replacement operation 1: the (109:236,149:276) area of the watermark Lena grayscale image shown in fig. 2 is replaced by the baboon grayscale image with the size of 128 × 128 shown in fig. 7, and the watermark Lena grayscale image obtained after the replacement operation is shown in fig. 8. The tampering positioning detection result obtained by the method of the invention is shown in fig. 9, and the tampered image area can be well positioned.
Area replacement operation 2: the watermark Lena gray scale image obtained after the replacement operation is shown in fig. 10, in which the pixel points in the area (129: 256) of the watermark Lena gray scale image shown in fig. 2 are replaced with the pixel points in the other area (229: 356) of the self image. The tampering positioning detection result obtained by the method of the invention is shown in fig. 11, and the tampered image area can be well positioned.
Area replacement operation 3: the region (329:456,229:356) of the watermark Lena gray scale image shown in fig. 2 is replaced by the region (229:356,379:506) of the original Lena gray scale image shown in fig. 1, and the watermark Lena gray scale image obtained after the replacement operation is shown in fig. 12. The falsification positioning detection result obtained by the method of the invention is shown in fig. 13, and the falsified image area can be accurately positioned.
And (3) area cutting operation: image blocks of the watermark Lena gray scale image shown in fig. 2, which are located in the (214:267,189:227) region, are cut, that is, the pixel value of each pixel point in the region is set to 0, and the watermark Lena gray scale image after the region cutting operation is obtained is shown in fig. 14. The falsification positioning detection result obtained by the method of the present invention is shown in fig. 15, and the falsified image area can be accurately detected and positioned.
Pixel value increase operation: the image block of the watermark Lena grayscale image shown in fig. 2, which is located in the (329:456,129:256) area, is subjected to a pixel value increasing operation, for example, the pixel value of each pixel point in the area is added to 50, so as to obtain the watermark Lena grayscale image after the pixel value increasing operation, as shown in fig. 16. The tampering positioning detection result obtained by the method of the invention is shown in fig. 17, and the tampered image area can be well positioned.
Claims (2)
1. A totally blind digital watermarking method with copyright protection and tampering positioning functions is characterized by comprising the following main processing procedures:
at a digital watermark embedding end, performing one-level discrete wavelet decomposition on an original 8-bit gray level image, dividing an obtained approximation subgraph into a plurality of non-overlapping subblocks with the size of 8 multiplied by 8, and dividing each subblock in the approximation subgraph into 4 non-overlapping regions with the size of 4 multiplied by 4; then, performing discrete cosine transform on each region in each sub-block in the approximation subgraph; then, according to discrete cosine transform direct current coefficients of the upper left corner region and the upper right corner region of all sub-blocks in the approximation subgraph, a characteristic watermark is created; according to the discrete cosine transform alternating current coefficients at the same positions of the left lower corner region and the right lower corner region of each sub-block in the approximation subgraph and each bit in the created characteristic watermark, self-embedding one bit in the characteristic watermark in the left lower corner region and the right lower corner region of each sub-block in the approximation subgraph; finally, obtaining a watermark image embedded with the digital watermark through inverse discrete cosine transform and first-level inverse discrete wavelet transform;
at a digital watermark extraction and detection end, performing one-level discrete wavelet decomposition on a watermark image embedded with a digital watermark, dividing an obtained approximation subgraph into a plurality of non-overlapping subblocks with the size of 8 multiplied by 8, and dividing each subblock in the approximation subgraph into 4 non-overlapping regions with the size of 4 multiplied by 4; then, performing discrete cosine transform on each region in each sub-block in the approximation subgraph; secondly, blind extraction is carried out according to discrete cosine transform direct current coefficients of upper left corner regions and upper right corner regions of all sub-blocks in the approximation subgraph to obtain a characteristic watermark; blind extraction is carried out according to discrete cosine transform alternating current coefficients at the same positions of a lower left corner region and a lower right corner region of each sub-block in the approximation subgraph to obtain an authentication watermark; finally, performing copyright protection by calculating a normalized correlation coefficient between the characteristic watermark obtained by blind extraction and the authentication watermark obtained by blind extraction, and realizing totally blind tampering and positioning by comparing the characteristic watermark obtained by blind extraction and the authentication watermark obtained by blind extraction bit by bit;
the totally blind digital watermarking method comprises two parts, namely digital watermark embedding, digital watermark extracting and detecting;
the digital watermark embedding part comprises the following specific steps:
firstly, selecting an original 8-bit gray image, and recording the image as F, wherein the size of the F is I multiplied by J; then, carrying out one-level discrete wavelet decomposition on the F to obtain three detail subgraphs and an approximation subgraph of the F, and marking the approximation subgraph of the F as FA with the size of FAThen, the FA is divided intoThe sub-blocks with the size of 8 multiplied by 8 and not overlapped with each other are marked as FABk(ii) a Dividing each sub-block in FA into 4 non-overlapping regions of 4 × 4 size, and dividing FABkThe upper left corner area in (1) is taken as FABk1 st area in (1) is notedWill FABkThe upper right corner area in (1) is taken as FABkThe 2 nd area in (b) is noted asWill FABkThe lower left corner area of middle is taken as FABkThe 3 rd region of (1) is noted asWill FABkThe lower right corner area of middle is taken as FABkThe 4 th region of (1) is noted asWherein I represents the horizontal resolution of F, J represents the vertical resolution of F, and the symbolIs a sign of a rounding-down operation, k is a positive integer,
step (r 2) of performing discrete cosine transform on each region in each sub-block in the FA to obtain a discrete cosine transform coefficient matrix of each region in each sub-block in the FA, and performing discrete cosine transform on each region in each sub-block in the FAIs expressed as a matrix of discrete cosine transform coefficientsWill be provided withIs expressed as a matrix of discrete cosine transform coefficientsWill be provided withIs expressed as a matrix of discrete cosine transform coefficientsWill be provided withIs expressed as a matrix of discrete cosine transform coefficientsWherein the content of the first and second substances, andthe dimensions of (A) are all 4 x 4;
step 3, creating a characteristic watermark W, which comprises the following specific steps: calculating the mean value of the discrete cosine transform direct current coefficient of each sub-block in the FA, and converting FAB into FABkThe mean value of the discrete cosine transform direct current coefficient is recorded as FADk,Then, according to the mean value of the discrete cosine transform direct current coefficients of all the sub-blocks in the FA, a characteristic watermark W is created and obtained, and the k bit of the W is marked as Wk,Wherein W has a length of To representRow 1 and column 1 elements in (a),is composed ofThe direct current coefficient of the discrete cosine transform of (1),to representRow 1 and column 1 elements in (1),is composed ofD.c. coefficient of discrete cosine transform of (delta)1Is a set first threshold;
step 4, self-embedding a characteristic watermark W, which comprises the following specific processes: self-embedding one bit of the characteristic watermark W in each sub-block in the FA, for FABkTaking out WkIf W iskIs equal to 0 andthen orderThen orderRealization of WkIs self-embedding, thenWhen it is established, the pairAndmake further adjustmentInstant commandAnd order
If WkIs equal to 0 andthen pairAndrealize W without processingkIs self-embedding, thenWhen it is established, the pairAndby further adjustment, it is instantly orderedAnd order
If W isk1 andthen orderThen orderRealization of WkIs self-embedding, thenWhen it is established, the pairAndby further adjustment, it is instantly orderedAnd order
If W isk1 andthen pairAndrealize W without processingkIs self-embedding, thenWhen it is established, the pairAndby further adjustment, it is instantly orderedAnd order
Wherein the content of the first and second substances,to representRow 2 and column 2 elements in (1),to representLine 2 and column 2 elements in (1), G1 and G2 are all introduced intermediate variables, abs () is the absolute value function, δ2In order to set the second threshold value, wherein, the symbol is an assignment symbol;
firstly, carrying out inverse discrete cosine transform on each region on the basis of the first step 4 by a first step 5; then each subblock embedded with the digital watermark is obtained through recombination, and then an approximation subgraph embedded with the digital watermark is obtained through recombination; then, performing one-level inverse discrete wavelet transform on the approximation subgraph embedded with the digital watermark and the three detail subgraphs of F to obtain a watermark image embedded with the digital watermark;
the digital watermark extracting and detecting part comprises the following specific steps:
step two _1, the watermark image embedded with the digital watermark is an 8-bit gray image which is marked as TF, and the size of TF is I multiplied by J; then, one-level discrete wavelet decomposition is carried out on the TF to obtain three details of the TFThe approximation subgraph of the TF is marked as TFA, and the size of the TFA isFollowed by cleavage of TFA intoThe k-th sub-block in TFA is denoted as TFABk(ii) a Dividing each sub-block of TFA into 4 non-overlapping regions of 4 × 4 size, and dividing TFAB into two regionskThe upper left corner region in (1) is taken as TFABk1 st area in (b) is noted asWill TFABkThe upper right corner region in (1) is used as TFABkThe 2 nd area in (b) is noted asWill TFABkThe lower left corner region in (1) is taken as TFABkThe 3 rd region of (1) is noted asWill TFABkThe lower right corner region in (1) is used as TFABkThe 4 th region of (1) is noted asWherein, the symbolFor rounding down the arithmetic sign, k is a positive integer,
step 2, performing discrete cosine transform on each region in each sub-block in TFA to obtain a discrete cosine transform coefficient of each region in each sub-block in TFAMatrix ofIs expressed as a matrix of discrete cosine transform coefficientsWill be provided withIs expressed as a matrix of discrete cosine transform coefficientsWill be provided withIs expressed as a matrix of discrete cosine transform coefficientsWill be provided withIs expressed as a matrix of discrete cosine transform coefficientsWherein the content of the first and second substances, andthe dimensions of (A) are all 4 x 4;
step two-3, blindly extracting the characteristic watermark TW, which comprises the following specific processes: calculating the mean value of the discrete cosine transform direct current coefficient of each sub-block in TFA, and converting TFAB into TFABkThe mean value of the discrete cosine transform direct current coefficient is recorded as TFADk,Then, according to the mean value of the discrete cosine transform direct current coefficients of all the sub-blocks in the TFA, a characteristic watermark TW is obtained through blind extraction, and the k-th bit of the TW is recorded as TWk,Wherein the length of TW is To representRow 1 and column 1 elements in (1),is composed ofThe direct current coefficient of the discrete cosine transform of (1),to representRow 1 and column 1 elements in (1),is composed ofD.c. coefficient of discrete cosine transform of (delta)1Is a set first threshold;
step 4, blind extraction of authentication watermark TW*The specific process is as follows: blindly extracting one bit in each sub-block in TFA to obtain the authentication watermark TW*For TFABkExtracting to obtain TW*K-th bit TW in* k,Wherein TW*Has a length of To representRow 2 and column 2 elements in (1),to representRow 2, column 2 elements in (1);
step 5, according to TW and TW*Copyright protection is carried out, and the specific process is as follows: calculating TW and TW*Normalized correlation coefficient between, noted as ρ (TW, TW)*) Using ρ (TW, TW)*) To perform copyright protection;
according to TW and TW*And (3) carrying out tampering positioning, wherein the specific process is as follows: according to each bit in TW and TW*Determines whether each sub-block in TFA has been tampered with, for TFABkIf TWkAnd TW* kEqual, then TFAB is determinedkHas not been tampered with; if TWkAnd TW* kIs not equal, then TFAB is determinedkAfter being tampered, the full-blind tampering positioning is realized.
2. A copyrighted according to claim 1The totally blind digital watermarking method for protecting and tampering with the positioning function is characterized in that in the step II-5,wherein the content of the first and second substances,represents the average of all elements in the TW,represents TW*Average value of all elements in (1).
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810299604.9A CN108648130B (en) | 2018-04-04 | 2018-04-04 | Totally-blind digital watermarking method with copyright protection and tampering positioning functions |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810299604.9A CN108648130B (en) | 2018-04-04 | 2018-04-04 | Totally-blind digital watermarking method with copyright protection and tampering positioning functions |
Publications (2)
Publication Number | Publication Date |
---|---|
CN108648130A CN108648130A (en) | 2018-10-12 |
CN108648130B true CN108648130B (en) | 2022-05-27 |
Family
ID=63745452
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201810299604.9A Active CN108648130B (en) | 2018-04-04 | 2018-04-04 | Totally-blind digital watermarking method with copyright protection and tampering positioning functions |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN108648130B (en) |
Families Citing this family (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109977878B (en) * | 2019-03-28 | 2021-01-22 | 华南理工大学 | Vehicle detection method based on heavily weighted Anchor |
CN110580405B (en) * | 2019-08-16 | 2022-07-05 | 湖北工业大学 | Picture copyright protection system and method based on social network |
CN110555792B (en) * | 2019-08-16 | 2022-05-17 | 广东外语外贸大学南国商学院 | Image tampering blind detection method based on normalized histogram comprehensive feature vector |
Family Cites Families (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101699508B (en) * | 2009-09-03 | 2012-01-11 | 中兴通讯股份有限公司 | Image digital watermark embedding and extracting method and system |
CN102184520B (en) * | 2011-05-26 | 2012-09-12 | 浙江工商大学 | Complete blind detection robust quantization watermarking method |
CN102208097B (en) * | 2011-05-26 | 2013-06-19 | 浙江工商大学 | Network image copyright real-time distinguishing method |
CN102223561B (en) * | 2011-07-01 | 2013-07-31 | 宁波大学 | Blind watermark embedding and extracting method of stereoscopic video image |
CN102682418B (en) * | 2012-05-11 | 2014-06-25 | 宁波大学 | Method for embedding and extracting multiple zero watermarks of digital image |
-
2018
- 2018-04-04 CN CN201810299604.9A patent/CN108648130B/en active Active
Also Published As
Publication number | Publication date |
---|---|
CN108648130A (en) | 2018-10-12 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Preda | Semi-fragile watermarking for image authentication with sensitive tamper localization in the wavelet domain | |
CN108416727B (en) | Color image full-blind robust digital watermarking method based on self-embedding technology | |
EP1953752B1 (en) | Embedding and detecting hidden information | |
Keshavarzian et al. | ROI based robust and secure image watermarking using DWT and Arnold map | |
US7792377B2 (en) | Method of image authentication and restoration | |
Zhang et al. | A data hiding scheme based on multidirectional line encoding and integer wavelet transform | |
Qin et al. | Adaptive self-recovery for tampered images based on VQ indexing and inpainting | |
Amornraksa et al. | Enhanced images watermarking based on amplitude modulation | |
CN108648130B (en) | Totally-blind digital watermarking method with copyright protection and tampering positioning functions | |
CN102024249A (en) | Digital image watermarking method based on visual perception characteristics | |
Bolourian Haghighi et al. | An effective semi-fragile watermarking method for image authentication based on lifting wavelet transform and feed-forward neural network | |
Kaur et al. | A review on digital watermarking using LSB | |
Rhayma et al. | Semi-fragile self-recovery watermarking scheme based on data representation through combination | |
Baaziz | Adaptive watermarking schemes based on a redundant contourlet transform | |
Tiwari et al. | Digital Image Watermarking using fractional Fourier transform with different attacks | |
CN117522666A (en) | Method and device for embedding and extracting invisible digital watermark in image | |
Rawat et al. | DCT and SVD based new watermarking scheme | |
Hana et al. | A robust multiple watermarking scheme based on the DWT | |
Wang et al. | Review on digital image watermarking based on singular value decomposition | |
Haitao et al. | Low luminance smooth blocks based watermarking scheme in DCT domain | |
Zope-Chaudhari et al. | Robust copyright protection of raster images using wavelet based digital watermarking | |
Zaker et al. | Improving security of pixel value differencing steganographic method | |
Jiao et al. | Framelet image watermarking considering dynamic visual masking | |
Chukwuchekwa et al. | Improved DCT based Digital Image Watermarking Technique using Particle Swarm Optimization Algorithm | |
Patil et al. | Digital video watermarking using Dwt and Pca |
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 |