CN111242827A - Robust color digital image watermarking method based on DT-CWT and SVD conversion - Google Patents

Robust color digital image watermarking method based on DT-CWT and SVD conversion Download PDF

Info

Publication number
CN111242827A
CN111242827A CN202010042854.1A CN202010042854A CN111242827A CN 111242827 A CN111242827 A CN 111242827A CN 202010042854 A CN202010042854 A CN 202010042854A CN 111242827 A CN111242827 A CN 111242827A
Authority
CN
China
Prior art keywords
watermark
level
sub
embedding
image
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
Application number
CN202010042854.1A
Other languages
Chinese (zh)
Other versions
CN111242827B (en
Inventor
宦文楠
钱振兴
张新鹏
李晟
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Fudan University
Original Assignee
Fudan University
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Fudan University filed Critical Fudan University
Priority to CN202010042854.1A priority Critical patent/CN111242827B/en
Publication of CN111242827A publication Critical patent/CN111242827A/en
Application granted granted Critical
Publication of CN111242827B publication Critical patent/CN111242827B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

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
    • G06T1/005—Robust watermarking, e.g. average attack or collusion attack resistant

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 belongs to the technical field of digital watermarking, and particularly relates to a robust color digital image watermarking method based on DT-CWT and SVD. In the invention, a watermark embedding party firstly carries out 4-level DT-CWT transformation on a color image U channel, carries out block processing on 6 high-pass sub-bands of a fourth level and a third level and carries out SVD transformation in each sub-block; then, selecting a first characteristic value in each sub-block diagonal matrix to embed a watermark; generating a perception signal by each sub-block to modulate the watermark; finally, calculating the embedding strength by using a drosophila optimization algorithm to optimize robustness; and the watermark detection party receives the image transmitted by the channel, extracts the embedded watermark according to the inverse embedding process, compares the extracted watermark with the watermark generated by the shared key and detects whether the watermark exists or not. The invention avoids the need of obtaining the watermark information and the original image size information in advance by a detector, and has better invisibility and stronger robustness under the condition of total blindness.

Description

Robust color digital image watermarking method based on DT-CWT and SVD conversion
Technical Field
The invention belongs to the technical field of digital watermarking, and particularly relates to a robust color digital image watermarking method based on DT-CWT and SVD conversion.
Background
Most of the current digital image watermarking technologies are controlled to modify all or some selected areas of the carrier image to embed the watermark, and meanwhile, the invisibility and the robustness of the image containing the watermark are ensured. But the robust digital image watermark should be continuously developed under the condition of different attacks on digital products. The copyright protection function should be effectively performed against various attacks in practical application. Only by continuously breaking through the limitation of the original digital image watermarking algorithm in actual attack, the method can take advantage in the countermeasure of information security.
One-bit watermarking technology, i.e. watermarking, is used as an authentication code without practical significance, wherein the role of the detector is to check the presence/absence of the watermark. In practice, copy detection, copyright protection, broadcast monitoring, etc. may be implemented using one-bit watermarking techniques. In copyright protection, a detector can detect the copyright only by holding the secret key, and certain convenience is provided for practical application.
The existing one-bit watermarking technology mainly comprises a spatial domain and a transform domain. Although the spatial domain approach is simple and easy to implement, and easy to apply and implement, the transform domain approach provides higher performance in terms of invisibility and robustness. It is of interest to choose which transform method and how to embed the watermark using the selected transform domain. As can be seen from the literature, most of the existing methods have great limitations in practical application.
Disclosure of Invention
The invention provides a robust color digital image watermarking method based on DT-CWT (double-number complex wavelet transform) and SVD (singular value decomposition) by utilizing the characteristics of approximate shift invariance, good direction selectivity and reconfigurability, limited redundancy and low computational complexity of DT-CWT (fractional wavelet transform), and aims to realize the copyright protection of digital images by utilizing a robust image watermarking algorithm.
The invention provides a robust color digital image watermarking method based on DT-CWT and SVD transformation, wherein a watermark embedder E and a watermark extractor D are arranged; the method comprises the following specific steps:
firstly, a watermark embedder E generates a watermark matrix W consisting of 1 and-1 according to the size of an image by using a Key Key shared by a detector;
then, extracting U channel of color image, making 4-level DT-CWT transform on U channel, and using 6 high-pass sub-band signals of fourth level
Figure BDA0002368363730000011
And a third level of 6 high-pass sub-band signals
Figure BDA0002368363730000012
Dividing the high-pass sub-band into non-overlapping blocks, wherein l is more than or equal to 1 and less than or equal to 6, and numbering the high-pass sub-bands of 6;
then, SVD transformation is carried out in each sub-block, and a first eigenvalue in a diagonal matrix of the sub-block is extracted
Figure BDA0002368363730000021
And
Figure BDA0002368363730000022
embedding watermarks, wherein 3 and 4 are series, l is a high-pass sub-band, and k is the number of blocks;
then generating a perceptual signal using each sub-block
Figure BDA0002368363730000023
And
Figure BDA0002368363730000024
as weights for embedding watermarks, the embedded watermark signals are modulated using a drosophila optimization algorithm [1]]Optimizing the PSNR, SSIM and normalized cross-correlation function (NCC) values of the watermarks under different attacks and the extracted watermark by using the watermark embedding factor α;
and finally, detecting whether the watermark exists or not by a watermark detector according to the extracted watermark information.
The method of the invention needs to have the following conditions:
(1) the watermark embedder E and the watermark detector D have the same Key.
The method comprises the following specific operation processes:
the watermark embedder E:
step 1: e is to generate a watermark signal embedded in a color image I of size mxnx3. Generating m-bit pseudo-random sequence w ═ w composed of 1 and-1 by using Key Key1,w2,w3,...,wm]M is M/25The watermark W consists of n rows of identical W, W ═ W1,w2,w3,...,wn]N is N/25。
Step 2: the watermark is embedded in the U channel of image I.
The method specifically comprises the following substeps:
step 2.1: the RGB image I is changed into a YCbCr image, and the calculation formula is as follows (2-1):
Figure BDA0002368363730000025
step 2.2: performing four-level DT-CWT transformation on the U channel with the size of MxN, and extracting the fourth-level 6 high-pass sub-bands and the third-level 6 high-pass sub-bands
Figure BDA0002368363730000026
Wherein, level is the progression, and the value is 3, 4, generates the perception signal:
Figure BDA0002368363730000027
Figure BDA0002368363730000028
in the formula (2-2), R is a coefficient for modulating the size of the sensing signal, and the embedding time value is fixed to be 8; and (2-3) is a low-pass filter.
Step 2.3: will be provided with
Figure BDA0002368363730000029
And
Figure BDA00023683637300000210
respectively block-wise processed, fourth stage
Figure BDA00023683637300000211
And
Figure BDA00023683637300000212
divided into non-overlapping sub-blocks of size 2 x 2, third level
Figure BDA00023683637300000214
And
Figure BDA00023683637300000213
into non-overlapping sub-blocks of size 4 x 4.
Step 2.4: the SVD decomposition is performed in each sub-block,
Figure BDA00023683637300000215
sub-block generation UUlevel,l,k、USlevel,l,k、UVlevel,l,kThe matrix is a matrix of a plurality of matrices,
Figure BDA00023683637300000216
sub-block generation MUlevel,l,k、MSlevel,l,k、MVlevel,l,kWhere k is the number of blocks.
Step 2.5: selection of diagonal matrix USlevel,l,kFirst characteristic value of
Figure BDA0002368363730000033
Embedding watermark, selecting diagonal matrix MSlevel,l,kFirst characteristic value of
Figure BDA0002368363730000034
Modulating the watermark, and embedding the watermark into the watermark by the formula (2-4):
Figure BDA0002368363730000031
wherein α is the watermark embedding strength, k is the number of blocks, and the value is M/25×N/25(ii) a The level is a series number and is 3, 4; l is the 6 high-pass subband numbers.
Step 2.6: UU Using Each sub-Blocklevel,l,k、UVlevel,l,kAnd a diagonal matrix US 'after embedding the watermark'level,l,kInverse SVD transform to generate watermarked
Figure BDA0002368363730000035
Step 2.7: and replacing the original high-pass sub-band with the fourth-level and third-level high-pass sub-bands after the watermark is embedded, and performing inverse DT-CWT transformation to generate a Uw channel after the watermark is embedded.
And step 3: and changing the Y, Uw and V back to the RGB channel to generate an image Iw embedded with the watermark.
And 4, step 4: and performing K attacks on the image Iw embedded with the watermark.
And 5: optimizing the embedding intensity factor by using a drosophila optimization algorithm [1], wherein the algorithm is shown in an appendix, and an optimization objective function is shown as (2-5):
Figure BDA0002368363730000032
wherein, αiFor the intensity factor embedded in each round; λ is a coefficient for adjusting the PSNR value, which is set to 38; omegajJ is more than or equal to 1 and less than or equal to 3; w is the watermark generated from the key, W*The watermarks extracted from the images under different attacks; k is the number of attacks.
And 6, re-embedding the watermark by using the optimal embedding factor α obtained in the step 5.
And the watermark detector is:
step 1: the RGB image I ' having a size of M ' × N ' × 3 is changed to a YCbCr image.
Step 2: watermark extraction is performed in the U channel.
The method specifically comprises the following substeps:
step 2.1: performing four-level DT-CWT transformation on the U channel with the size of M 'multiplied by N', and extracting the fourth-level 6 high-pass sub-bands and the third-level 6 high-pass sub-bands
Figure BDA0002368363730000043
Wherein, the level is a series number and the value is 3 and 4;
step 2.2: will be provided with
Figure BDA0002368363730000044
Block processing, fourth stage
Figure BDA0002368363730000045
Divided into non-overlapping sub-blocks of size 2 x 2, third level
Figure BDA0002368363730000046
Dividing into non-overlapping sub-blocks of size 4 x 4;
step 2.3: the SVD decomposition is performed in each sub-block,
Figure BDA0002368363730000047
sub-block generates UU'level,l,k、US’level,l,k、UV’level,l,kMatrix, select diagonal matrix US'level,l,kFirst characteristic value of
Figure BDA0002368363730000048
Extracting the watermark, wherein the extraction formula is as (3-1) and (3-2):
Figure BDA0002368363730000041
Figure BDA0002368363730000042
and step 3: and the watermark detector D performs watermark detection according to the extracted watermark information.
The method specifically comprises the following substeps:
step 3.1: generating m-bit pseudo-random sequence w ═ w composed of 1 and-1 by using Key shared with watermark embedder1,w2,w3,...,wm]M is M'/25The watermark W is composed of n rows of identical W, and W is ═ W1,w2,w3,...,wn]N is N'/25。
Step 3.2: in order to improve the accuracy of detecting the watermark under different attacks, the watermark is dynamically detected on different levels; obtaining the NCC value of the generated watermark W and the extracted watermark W', wherein when the NCC value is larger than a set threshold value, the image contains the watermark; and when the NCC value is smaller than the set threshold value, the image does not contain the watermark.
The method of the invention combines the advantages of DT-CWT and SVD to embed the watermark. The watermark embedding party generates a watermark by using a secret key according to the size of the image; 4-level DT-CWT transformation is carried out on a color image U channel, 6 high-pass sub-bands of the fourth level and the third level are subjected to blocking processing, and SVD transformation is carried out in each sub-block; then, selecting a first characteristic value in each sub-block diagonal matrix to embed a watermark; generating a perception signal by each sub-block to modulate the watermark; and finally, calculating the embedding strength by using a drosophila optimization algorithm, and optimizing robustness on the premise of ensuring that the peak signal-to-noise ratio (PSNR) and the Structural Similarity (SSIM) meet visual requirements. And the watermark detection party receives the image transmitted by the channel, extracts the embedded watermark according to the inverse embedding process, compares the extracted watermark with the watermark generated by the shared secret key and detects the presence or absence of the watermark.
The method provided by the invention breaks through the limitation of digital image watermarking, develops a new digital image method, and has the following beneficial effects:
1. the invention avoids the abnormal image perception caused by watermark embedding.
2. The invention breaks through the limitation of digital image watermark in practical use, and the watermark detector can detect the watermark only by possessing the secret key without knowing the original image size information or the original watermark size information. Watermark embedding is realized by using the advantages of the U channel of the image in invisibility and the DT-CWT and SVD in robustness, so that the image watermark is more perceptually vivid and has stronger robustness in different attacks.
Drawings
Fig. 1 is a flowchart of a robust color digital image watermarking method based on DT-CWT and SVD transforms according to this embodiment.
Fig. 2 shows a front lens and a rear lens of embedding a watermark. Wherein, (a) lena diagram of the watermark to be embedded, and (b) lena diagram after the watermark is embedded.
Fig. 3 is a cutting diagram of the watermark to be extracted.
Detailed Description
The invention will be further illustrated with reference to the following specific examples.
The watermark embedder:
step 1: e is to generate a watermark signal embedded in a color image lena of 512 × 512 × 3, where the lena diagram is denoted as I, as shown in fig. 2 (a). Generating an m-bit pseudo-random sequence w composed of 1 and-1 by using a Key Key 3, wherein the m-bit pseudo-random sequence w is [1,1, -1,1, -1,1, -1,1,1,1, -1, -1, -1,1, -1,1, -1]M is 16, and the watermark W is composed of n rows of identical W, W ═ W1,w2,w1,...,wn]And n is 16 in size.
Step 2: the watermark is embedded in the U channel of image I.
The method specifically comprises the following substeps:
step 2.1: according to the formula (2-1), changing the RGB image I into a YCbCr image;
step 2.2: the U channel with the size of 512 multiplied by 512 is subjected to four-level DT-CWT transformation, and the fourth-level 6 high-pass sub-bands and the third-level 6 high-pass sub-bands are extracted
Figure BDA0002368363730000051
Wherein, the level is a series number, the value is 3 and 4, and corresponding perception signals are generated according to formulas (2-2) and (2-3);
step 2.3: will be provided with
Figure BDA0002368363730000052
And
Figure BDA0002368363730000053
respectively block-wise processed, fourth stage
Figure BDA0002368363730000054
And
Figure BDA0002368363730000055
divided into non-overlapping sub-blocks of size 2 x 2, third level
Figure BDA0002368363730000057
And
Figure BDA0002368363730000056
into non-overlapping sub-blocks of size 4 x 4.
Step 2.4: the SVD decomposition is performed in each sub-block,
Figure BDA0002368363730000058
sub-block generation UUlevel,l,k、USlevel,l,k、UVlevel,l,kThe matrix is a matrix of a plurality of matrices,
Figure BDA0002368363730000059
sub-block generation MUlevel,l,k、MSlevel,l,k、MVlevel,l,kWhere k is the number of blocks.
Step 2.5: selection of diagonal matrix USlevel,l,kFirst characteristic value of
Figure BDA00023683637300000510
Embedding watermark, selecting diagonal matrix MSlevel,l,kFirst characteristic value of
Figure BDA00023683637300000511
The watermark is modulated and embedded according to equation (2-4) where α is initially set to 6 and the value of k is 256.
Step 2.6: UU Using Each sub-Blocklevel,l,k、UVlevel,l,kAnd a diagonal matrix US 'after embedding the watermark'level,l,kInverse SVD transform to generate watermarked
Figure BDA0002368363730000061
Step 2.7: and replacing the original high-pass sub-band with the fourth-level and third-level high-pass sub-bands after embedding the watermark, and performing inverse DT-CWT transformation to generate a Uw channel after embedding the watermark.
And step 3: changing Y, Uw and V back to RGB channels to generate an image Iw embedded with a watermark, and step 4: performing K kinds of attacks on the image Iw after the watermark is embedded, wherein the attack includes: rotation, zooming, clipping, rotation + zooming, rotation + clipping, zooming + clipping, JPEG compression + rotation + zooming + clipping and other geometric attacks and median filtering and other signal attacks,
and 5: the embedding intensity factor is optimized by utilizing a drosophila optimization Algorithm, the formula (2-5) is an optimization objective function, and the calculation process is shown as Algorithm 1.
In step 6, the watermark is re-embedded using the optimal embedding factor α, which is obtained in step 5, as 8, and the generated watermark-embedded image is as shown in fig. 2 (b).
And the watermark detector is:
the image after embedding the watermark is subjected to a cropping attack, and the size after cropping is 410 × 512 × 3, as shown in fig. 3.
Step 1: the RGB image I' having a size of 410 × 512 × 3 is changed into a YCbCr image.
Step 2: watermark extraction is performed in the U channel.
The method specifically comprises the following substeps:
step 2.1: the U channel with the size of 410 multiplied by 512 is subjected to four-level DT-CWT transformation, and the fourth-level 6 high-pass sub-bands and the third-level 6 high-pass sub-bands are extracted
Figure BDA0002368363730000062
Wherein, the level is a series number and the value is 3 and 4;
step 2.2: will be provided with
Figure BDA0002368363730000063
Block processing, fourth stage
Figure BDA0002368363730000064
Divided into non-overlapping sub-blocks of size 2 x 2, third level
Figure BDA0002368363730000065
Dividing into non-overlapping sub-blocks of size 4 x 4;
step 2.3: the SVD decomposition is performed in each sub-block,
Figure BDA0002368363730000066
sub-block generates UU'level,l,k、US’level,l,k、UV’level,l,kMatrix, select diagonal matrix US'level,l,kFirst characteristic value of
Figure BDA0002368363730000067
And extracting the watermark according to the formulas (3-1) and (3-2).
And step 3: and the watermark detector D performs watermark detection according to the extracted watermark information.
The method specifically comprises the following substeps:
step 3.1: an m-bit pseudo-random sequence w composed of 1 and-1 is generated by using a Key shared with a watermark embedder, i.e., [1,1, -1,1, -1,1,1,1, -1, -1, -1,1,1]M is 13, and the watermark W is composed of n rows of identical W, W ═ W1,w2,w3,...,wn]And n is 16 in size.
Step 3.2: in order to improve the accuracy of detecting the watermark under different attacks, the watermark is dynamically detected on different levels; if the maximum NCC value of the generated watermark W and the extracted watermark W' is 1 and the NCC value is greater than a predetermined threshold value of 0.65, the image contains the watermark.
It can be known from the foregoing description that the present invention breaks through the limitation of digital image watermarking in practical application, strives to develop a digital watermarking direction with more practical application value, and proposes a one-bit watermarking feasible scheme using DT-CWT and SVD transformation. Under the premise that digital products are widely spread, more and more copying and pirating products emerge, conditions for realizing copyright protection by using the one-bit watermarking technology are gradually mature, and the one-bit watermarking technology has wide development space and application prospect.
Experiments show that the method provided by the invention can realize copyright protection on the color image by using a watermark method of DT-CWT and SVD conversion.
The foregoing is merely a preferred embodiment of the invention and is not intended to limit the invention in any manner and in any manner. It should be noted that, for those skilled in the art, without departing from the method of the present invention, several modifications and additions can be made, and these modifications and additions should also be considered as the scope of protection of the present invention, and that several changes, modifications and variations which are equivalent to the above-disclosed technical contents can be made without departing from the spirit and scope of the present invention. Meanwhile, any changes, modifications and variations of the above-described embodiments, which are equivalent to those of the technical spirit of the present invention, are within the scope of the technical solution of the present invention.
Reference to the literature
[1]Pan W T.A new Fruit Fly Optimization Algorithm:Taking thefinancial distress model as an example[J].Knowledge-Based Systems,2012,26(none):69-74.
Appendix
Figure BDA0002368363730000081
Figure BDA0002368363730000091

Claims (3)

1. A robust color digital image watermarking method based on DT-CWT and SVD transformation is provided with a watermark embedder E and a watermark detector D; the watermark embedder E and the watermark detector D have the same Key Key; the method is characterized by comprising the following specific steps:
(1) the watermark embedder E generates a watermark matrix W consisting of 1 and-1 according to the size of the image by using a Key Key shared by the detector;
(2) extracting U channel of color image, making 4-stage DT-CWT conversion on U channel, and using 6 high-pass sub-band signals of fourth stage
Figure FDA0002368363720000011
And a third level of 6 high-pass sub-band signals
Figure FDA0002368363720000012
Dividing the high-pass sub-band into non-overlapping blocks, wherein l is more than or equal to 1 and less than or equal to 6, and numbering the high-pass sub-bands of 6;
(3) SVD conversion is carried out in each sub-block, and the diagonal moment of the sub-block is extractedFirst eigenvalue in the array
Figure FDA0002368363720000013
And
Figure FDA0002368363720000014
embedding watermarks, wherein 3 and 4 are series numbers, l is a high-pass sub-band number, and k is a block number;
(4) generating a perceptual signal with each sub-block
Figure FDA0002368363720000015
And
Figure FDA0002368363720000016
modulating the embedded watermark signal as the weight of the embedded watermark, and optimizing a watermark embedding factor α by utilizing a drosophila optimization algorithm to PSNR, SSIM and a normalized cross-correlation function value (NCC) of the watermark under different attacks and the extracted watermark;
(5) and detecting whether the watermark exists or not by the watermark detector according to the extracted watermark information.
2. The DT-CWT and SVD transform-based robust color digital image watermarking method of claim 1, wherein the specific operation flow of the watermark embedder E is as follows:
step 1: e, generating a watermark signal embedded in the color image I with the size of M multiplied by N multiplied by 3; generating m-bit pseudo-random sequence w ═ w composed of 1 and-1 by using Key Key1,w2,w3,...,wm]M is M/25The watermark W consists of n rows of identical W, W ═ W1,w2,w3,...,wn]N is N/25；
Step 2: embedding a watermark into a U channel of an image I, wherein the embedding comprises the following specific steps:
step 2.1: the RGB image I is changed into a YCbCr image, and the calculation formula is as follows (2-1):
Figure FDA0002368363720000017
step 2.2: performing four-level DT-CWT transformation on the U channel with the size of MxN, and extracting the fourth-level 6 high-pass sub-bands and the third-level 6 high-pass sub-bands
Figure FDA0002368363720000018
Wherein, level is a series, the value is 3, 4, and the perception signal is generated:
Figure FDA0002368363720000019
Figure FDA00023683637200000110
in the formula (2-2), R is a coefficient for modulating the size of the sensing signal, and the embedding time value is fixed to be 8; (2-3) a low-pass filter;
step 2.3: will be provided with
Figure FDA0002368363720000021
And
Figure FDA0002368363720000022
respectively block-wise processed, fourth stage
Figure FDA0002368363720000023
And
Figure FDA0002368363720000024
divided into non-overlapping sub-blocks of size 2 x 2, third level
Figure FDA0002368363720000025
And
Figure FDA0002368363720000026
dividing into non-overlapping sub-blocks of size 4 x 4;
step 2.4: the SVD decomposition is performed in each sub-block,
Figure FDA0002368363720000027
sub-block generation UUlevel,l,k、USlevel,l,k、UVlevel,l,kThe matrix is a matrix of a plurality of matrices,
Figure FDA0002368363720000028
sub-block generation MUlevel,l,k、MSlevel,l,k、MVlevel,l,kWherein k is the number of blocks;
step 2.5: selection of diagonal matrix USlevel,l,kFirst characteristic value of
Figure FDA0002368363720000029
Embedding watermark, selecting diagonal matrix MSlevel,l,kFirst characteristic value of
Figure FDA00023683637200000210
Modulating the watermark, and embedding the watermark into the watermark by the formula (2-4):
Figure FDA00023683637200000211
wherein α is the watermark embedding strength, k is the number of blocks, and the value is M/25×N/25(ii) a The level is a series number and is 3, 4; l is the number of 6 high-pass sub-bands;
step 2.6: UU Using Each sub-Blocklevel,l,k、UVlevel,l,kAnd a diagonal matrix US 'after embedding the watermark'level,l,kAfter inverse SVD conversion, generating embedded watermark
Figure FDA00023683637200000212
Step 2.7: replacing the original high-pass sub-band with the fourth-level and third-level high-pass sub-bands after embedding the watermark, and performing inverse DT-CWT (differential transformation-coarse wavelet transform) to generate a Uw channel after embedding the watermark;
and step 3: changing Y, Uw and V back to RGB channels to generate an image Iw embedded with a watermark;
and 4, step 4: carrying out K attacks on the image Iw embedded with the watermark;
and 5: optimizing the embedding intensity factor by utilizing a drosophila optimization algorithm, wherein an optimization objective function is as follows (2-5):
Figure FDA00023683637200000213
wherein, αiFor the intensity factor embedded in each round; λ is a coefficient for adjusting the PSNR value; omegajJ is more than or equal to 1 and less than or equal to 3; w is the watermark generated from the key, W*The watermarks extracted from the images under different attacks; k is the attack number;
and 6, re-embedding the watermark by using the optimal embedding factor α obtained in the step 5.
3. The DT-CWT and SVD transform-based robust color digital image watermarking method of claim 2, wherein the watermark detector D operates as follows:
step 1: changing an RGB image I ' with a size of M ' × N ' × 3 into a YCbCr image;
step 2: watermark extraction is carried out in the U channel, and the specific steps of the extraction are as follows:
step 2.1: performing four-level DT-CWT transformation on the U channel with the size of M 'multiplied by N', and extracting the fourth-level 6 high-pass sub-bands and the third-level 6 high-pass sub-bands
Figure FDA0002368363720000031
Wherein, the level is a series number and the value is 3 and 4;
step 2.2: a fourth stage
Figure FDA0002368363720000032
Dividing into non-overlapping sub-blocks of size 2 × 2, and dividing into third stage
Figure FDA0002368363720000033
Dividing into non-overlapping sub-blocks of size 4 x 4;
step (ii) of2.3: the SVD decomposition is performed in each sub-block,
Figure FDA0002368363720000034
sub-block generates UU'level,l,k、US’level,l,k、UV’level,l,kMatrix, select diagonal matrix US'level,l,kFirst characteristic value of
Figure FDA0002368363720000035
Extracting the watermark, wherein the extraction formula is as (3-1) and (3-2):
Figure FDA0002368363720000036
Figure FDA0002368363720000037
and step 3: the watermark detector D carries out watermark detection according to the extracted watermark information, and the detection specific steps are as follows:
step 3.1: generating m-bit pseudo-random sequence w ═ w composed of 1 and-1 by using Key shared with watermark embedder1,w2,...,wm]M is M'/25The watermark W consists of n rows of identical W, W ═ W1,w2,...,wn]N is N'/25；
Step 3.2: in order to improve the accuracy of detecting the watermark under different attacks, the watermark is dynamically detected on different levels; obtaining the NCC value of the generated watermark W and the extracted watermark W', wherein when the NCC value is larger than a set threshold value, the image contains the watermark; and when the NCC value is smaller than the set threshold value, the image does not contain the watermark.
CN202010042854.1A 2020-01-15 2020-01-15 Robust color digital image watermarking method based on DT-CWT and SVD conversion Active CN111242827B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202010042854.1A CN111242827B (en) 2020-01-15 2020-01-15 Robust color digital image watermarking method based on DT-CWT and SVD conversion

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202010042854.1A CN111242827B (en) 2020-01-15 2020-01-15 Robust color digital image watermarking method based on DT-CWT and SVD conversion

Publications (2)

Publication Number Publication Date
CN111242827A true CN111242827A (en) 2020-06-05
CN111242827B CN111242827B (en) 2022-12-20

Family

ID=70876149

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202010042854.1A Active CN111242827B (en) 2020-01-15 2020-01-15 Robust color digital image watermarking method based on DT-CWT and SVD conversion

Country Status (1)

Country Link
CN (1) CN111242827B (en)

Cited By (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112927700A (en) * 2021-02-06 2021-06-08 兰州理工大学 Blind audio watermark embedding and extracting method and system
CN115834979A (en) * 2022-11-14 2023-03-21 复旦大学 Blind robust video watermarking method based on DT CWT-SVD and three-dimensional sub-band side view
CN116318690A (en) * 2023-05-25 2023-06-23 山东大学 Safe outsourcing calculation method and system based on matrix singular value decomposition

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101908201A (en) * 2010-06-28 2010-12-08 宁波大学 Robust digital image adaptive zero-watermarking method
CN105931174A (en) * 2016-04-12 2016-09-07 上海理工大学 High-robustness digital watermarking method making color QR code embedded in color image
CN108564519A (en) * 2017-12-22 2018-09-21 重庆邮电大学 A kind of Watermarking for Color Images method based on different transform domain
CN109544434A (en) * 2018-10-16 2019-03-29 江苏信实云安全技术有限公司 CWT-QIM digital blind watermark algorithm for RAW format-pattern
CN110189243A (en) * 2019-05-13 2019-08-30 杭州电子科技大学上虞科学与工程研究院有限公司 A kind of color image robust watermarking method based on tensor singular value decomposition

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101908201A (en) * 2010-06-28 2010-12-08 宁波大学 Robust digital image adaptive zero-watermarking method
CN105931174A (en) * 2016-04-12 2016-09-07 上海理工大学 High-robustness digital watermarking method making color QR code embedded in color image
CN108564519A (en) * 2017-12-22 2018-09-21 重庆邮电大学 A kind of Watermarking for Color Images method based on different transform domain
CN109544434A (en) * 2018-10-16 2019-03-29 江苏信实云安全技术有限公司 CWT-QIM digital blind watermark algorithm for RAW format-pattern
CN110189243A (en) * 2019-05-13 2019-08-30 杭州电子科技大学上虞科学与工程研究院有限公司 A kind of color image robust watermarking method based on tensor singular value decomposition

Non-Patent Citations (5)

* Cited by examiner, † Cited by third party
Title
MD. ASIKUZZAMAN: "A blind and robust video watermarking scheme in the DT CWT and SVD domain", 《IEEE》 *
XIANGHONG TANG: "A Color Video Watermarking Algorithm based on DTCWT and Motion Estimation", 《IEEE》 *
陈亚: "基于小波变换域的数字水印算法研究", 《中国优秀硕士学位论文全文数据库信息科技辑》 *
高仕龙: "一种DT-CWT域内的自适应图像零水印算法", 《四川大学学报》 *
高晓玲等: "基于压缩感知和模糊理论的鲁棒数字水印算法", 《计算机工程与设计》 *

Cited By (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112927700A (en) * 2021-02-06 2021-06-08 兰州理工大学 Blind audio watermark embedding and extracting method and system
CN112927700B (en) * 2021-02-06 2024-03-19 兰州理工大学 Blind audio watermark embedding and extracting method and system
CN115834979A (en) * 2022-11-14 2023-03-21 复旦大学 Blind robust video watermarking method based on DT CWT-SVD and three-dimensional sub-band side view
CN116318690A (en) * 2023-05-25 2023-06-23 山东大学 Safe outsourcing calculation method and system based on matrix singular value decomposition
CN116318690B (en) * 2023-05-25 2023-08-15 山东大学 Safe outsourcing calculation method and system based on matrix singular value decomposition

Also Published As

Publication number Publication date
CN111242827B (en) 2022-12-20

Similar Documents

Publication Publication Date Title
Al-Otum Semi-fragile watermarking for grayscale image authentication and tamper detection based on an adjusted expanded-bit multiscale quantization-based technique
Zou et al. Steganalysis based on Markov model of thresholded prediction-error image
CN111242827B (en) Robust color digital image watermarking method based on DT-CWT and SVD conversion
Song et al. A robust region-adaptive dual image watermarking technique
Roy et al. A perception based color image adaptive watermarking scheme in YCbCr space
EP1336161A1 (en) Method and arrangement for embedding a watermark in an information signal
Arya et al. A secure non-blind block based digital image watermarking technique using DWT and DCT
Sakshi et al. Least Significant Bit Steganography for Text and Image hiding
CN101833745B (en) Method for detecting embedding and extracting of multiple binary embedded watermarks of digital image
CN105512999A (en) Dual-transform color image holographic watermarking method
Rakhmawati et al. Blind Robust and Self-Embedding Fragile Image Watermarking for Image Authentication and Copyright Protection with Recovery Capability.
CN109754358B (en) Image watermarking method and system based on significance detection and contourlet transformation
Kumar et al. WSOmark: An adaptive dual-purpose color image watermarking using white shark optimizer and Levenberg–Marquardt BPNN
Tiwari et al. Digital Image Watermarking using fractional Fourier transform with different attacks
Thomas et al. Contourlet and Gould transforms for hybrid image watermarking in RGB color images
Al-Otum et al. Adaptive color image watermarking based on a modified improved pixel-wise masking technique
Kekre et al. Robust Watermarking Technique using Hybrid Wavelet Transform Generated from Kekre Transform and Discrete Cosine Transform
Wajid et al. Robust and imperceptible image watermarking using full counter propagation neural networks
Wang et al. Review on digital image watermarking based on singular value decomposition
Yang et al. High‐Quality and Robust Reversible Data Hiding by Coefficient Shifting Algorithm
Chukwuchekwa et al. Improved DCT based Digital Image Watermarking Technique using Particle Swarm Optimization Algorithm
Tanveer et al. A technique to reconstruct wavelet-based watermark immune against salt & pepper noise
Jamali et al. Adaptive image watermarking using human perception based fuzzy inference system
Taherinia et al. A robust spread spectrum watermarking method using two levels DCT
Phadikar et al. QIM data hiding for tamper detection and correction in digital images using wavelet transform

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