A semi-blind watermarking method and system based on a SWT-based sound-map watermarking synchronization embedding algorithm

Abstract

The application belongs to the technical field of image copyright confirmation and information security guarantee, and particularly relates to a semi-blind watermarking method and system based on a sound-image watermark synchronous embedding algorithm of SWT, which comprises the following steps: based on a phase truncation algorithm and a complex function algorithm, sound-image watermarks are synchronously embedded into a host image to be embedded with watermarks to obtain a watermark embedding image; the sound-image watermarks embedded in the watermark embedding image are synchronously extracted to obtain the embedded sound-image watermarks, thereby realizing the synchronous extraction of the sound-image watermarks. The application realizes the functions of sound-image watermark synchronous embedding and extraction, and guarantees the invisibility and robustness of the watermark system.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of image copyright confirmation and information security guarantee, and particularly relates to a semi-blind watermarking method and system based on a SWT sound-image watermark synchronous embedding algorithm. BACKGROUND

[0002] As a core technical means to guarantee image information security, image watermarking technology has become a research focus of the academic and industrial circles in recent years and has been widely and deeply explored in the field of image information security. Image watermarking technology embeds specific information in a target carrier image in a hidden manner, so that the embedded information is imperceptible at the visual level and does not affect the normal use of the carrier image, thereby protecting the information security of the image. Based on the visual perceptibility of watermark information, image watermarking technology can be divided into two categories: visible watermarking method and invisible watermarking method. Among them, visible watermarking refers to a watermark form that can be directly perceived by the human eye, and its application scenarios mainly cover the fields of images and videos. Visible watermarking technology adds semi-transparent text, symbols or images and other information on documents, pictures and videos to form a recognizable ownership identifier, thereby preventing the carrier information from being spread on the network and being used by others. The main feature of visible watermarking is that the watermark in the carrier has visual recognition but does not significantly interfere with the perception of the theme content, and it usually requires strong anti-removal capability to ensure that the original quality of the carrier is not damaged, thereby realizing the copyright ownership protection of the carrier. However, with the rapid development of digital image processing technology, attackers can use various software tools and image processing algorithms to implement traceless removal operation on visible watermarking, which poses a serious security challenge to the practical application of visible watermarking technology.

[0003] Corresponding to visible watermarking is invisible watermarking technology with better security, which is characterized in that the watermark information cannot be directly perceived by the human eye. Invisible watermarking technology has a wider application range and covers the copyright protection field of various digital carriers such as text, image, audio and video. The core technical feature of invisible watermarking is that the watermark information is imperceptible at the visual level, and the copyright owner can embed the characteristic information representing his identity into the target carrier as a watermark; when a copyright dispute occurs, the carrier copyright ownership can be confirmed by extracting and verifying the watermark information. Therefore, invisible digital watermarking technology has important practical application value in the field of product anti-counterfeiting and traceability.

[0004] In most of the existing digital watermark embedding schemes, the embedding operation of watermark information is mainly realized by using additive watermark mechanism, that is, the watermark information is embedded into the host image in an additive manner. In the watermark extraction process, most of the existing watermark methods based on singular value decomposition (SVD) need the participation of the original watermark image or the original host image to complete the extraction process. This technical feature leads to the problems of false positive and vulnerability to image processing attacks in the actual application of such watermark systems, which significantly affects the accuracy and reliability of watermark verification.

[0005] Therefore, the research and design of watermark systems have become a research hotspot and core direction in the field of information security. Constructing a watermark embedding algorithm with high security and high effectiveness, solving the false alarm problem and the weak anti-image processing attack ability of the existing watermark system, has important theoretical value and positive practical significance for promoting the technical development of the image copyright protection field. SUMMARY

[0006] To solve the problems existing in the prior art, the present application provides a semi-blind watermark method and system based on a SWT-based acoustic-image watermark synchronous embedding algorithm, which aims to effectively avoid false positive detection of watermarks and has excellent anti-image processing attack performance on the basis of realizing semi-blind extraction of watermarks.

[0007] To achieve the above-mentioned purpose, the present application provides the following solutions:

[0008] A semi-blind watermark method based on a SWT-based acoustic-image watermark synchronous embedding algorithm, the method comprising:

[0009] synchronously embedding acoustic-image watermarks into the host image to be embedded with watermarks based on a phase truncation algorithm and a complex function algorithm to obtain a watermark embedded image;

[0010] synchronously extracting acoustic-image watermarks from the watermark embedded image to obtain embedded acoustic-image watermarks, thereby realizing synchronous extraction of acoustic-image watermarks.

[0011] Preferably, the method for synchronously embedding acoustic-image watermarks into the host image to be embedded with watermarks based on a phase truncation algorithm and a complex function algorithm to obtain a watermark embedded image comprises:

[0012] Step 1: decomposing the host image to be embedded with watermarks into sub-blocks of different frequencies through one-dimensional Haar synchronous compression wavelet transform to obtain low-frequency coefficients, high-frequency coefficients in the horizontal direction, high-frequency coefficients in the vertical direction and high-frequency coefficients in the diagonal direction of the host image;

[0013] Step 2: performing SVD on the high-frequency coefficients in the horizontal direction of the host image to obtain a singular value matrix of the host image;

[0014] Third step: synchronously embedding the original sound-image watermark into the singular value matrix of the host image through the phase truncation algorithm and the complex function algorithm to obtain the matrix after synchronously embedding the watermark, and obtaining the first side information based on the amplitude truncation algorithm;

[0015] Fourth step: performing singular value decomposition on the matrix after synchronously embedding the watermark SVD obtaining the singular value matrix of the matrix after synchronously embedding the watermark and the second side information, and then obtaining the high-frequency coefficient containing the watermark information through inverse singular value decomposition;

[0016] Fifth step: obtaining the watermark embedding image through inverse synchronous compression wavelet transform based on the high-frequency coefficient containing the watermark information.

[0017] Preferably, the method of synchronously embedding the original sound-image watermark into the singular value matrix of the host image through the phase truncation algorithm and the complex function algorithm to obtain the matrix after synchronously embedding the watermark, and obtaining the first side information based on the amplitude truncation algorithm comprises:

[0018] ;

[0019] ;

[0020] wherein, i represents an imaginary number, w v represents a two-dimensional audio watermark matrix reshaped from the original one-dimensional audio watermark, w I represents the original image watermark, w is the original sound-image watermark, PT (·) represents a phase truncation operation, w p represents a composite watermark obtained after performing the phase truncation operation on the original sound-image watermark;

[0021] ;

[0022] wherein, AT (·) represents an amplitude truncation operation, w a represents the first side information obtained after performing the amplitude truncation operation on the original sound-image watermark;

[0023] ;

[0024] wherein, is the watermark embedding strength, is the singular value matrix of the host image, is the matrix after synchronously embedding the watermark.

[0025] Preferably, the watermark-embedded image is subjected to acoustic-mapping watermark synchronous extraction to obtain the embedded acoustic-mapping watermark, and the method for realizing synchronous extraction of the acoustic-mapping watermark comprises the following steps:

[0026] Step 1: decomposing the host image to be embedded with the watermark into sub-blocks of different frequencies through one-dimensional Haar synchronous compression wavelet transform to obtain low-frequency coefficients, high-frequency coefficients in the horizontal direction, high-frequency coefficients in the vertical direction and high-frequency coefficients in the diagonal direction of the watermark-embedded image;

[0027] Step 2: performing singular value decomposition on the high-frequency coefficients in the horizontal direction of the watermark-embedded image to obtain a singular value matrix containing watermark information;

[0028] Step 3: performing inverse singular value decomposition on the singular value matrix containing watermark information based on the second side information, then taking the imaginary part, and combining the first side information to obtain the embedded acoustic-mapping watermark, thereby realizing synchronous extraction of the acoustic-mapping watermark.

[0029] The application further provides a semi-blind watermark system based on the acoustic-mapping watermark synchronous embedding algorithm of SWT, which is used to realize the method described above, and comprises an acoustic-mapping watermark synchronous embedding module and an acoustic-mapping watermark synchronous extraction module.

[0030] The acoustic-mapping watermark synchronous embedding module is used to synchronously embed acoustic-mapping watermark into the host image to be embedded with the watermark based on the phase truncation algorithm and the complex function algorithm to obtain a watermark-embedded image.

[0031] The acoustic-mapping watermark synchronous extraction module is used to synchronously extract acoustic-mapping watermark from the watermark-embedded image to obtain the embedded watermark, thereby realizing semi-blind extraction of the acoustic-mapping watermark.

[0032] Preferably, the acoustic-mapping watermark synchronous embedding module comprises a first SWT transformation unit, a first SVD decomposition unit, an acoustic-mapping watermark synchronous embedding unit, a second SVD decomposition unit and a second SWT transformation unit.

[0033] The first SWT transformation unit is used to decompose the host image to be embedded with the watermark into sub-blocks of different frequencies through one-dimensional Haar synchronous compression wavelet transform to obtain low-frequency coefficients, high-frequency coefficients in the horizontal direction, high-frequency coefficients in the vertical direction and high-frequency coefficients in the diagonal direction of the host image;

[0034] The first SVD decomposition unit is used to perform singular value decomposition on the high-frequency coefficients in the horizontal direction of the host image to obtain a singular value matrix of the host image; SVD

[0035] ​The sound-image watermark synchronous embedding unit is configured to embed the original sound-image watermark into a singular value matrix of a host image by using a phase truncation algorithm and a complex function algorithm to obtain a matrix after the watermark is synchronously embedded, and obtain first side information based on an amplitude truncation algorithm;

[0036] The second SVD decomposition unit is configured to perform singular value decomposition on the matrix after the watermark is synchronously embedded to obtain a singular value matrix of the matrix after the watermark is synchronously embedded and second side information. SVD The singular value matrix of the matrix after the watermark is synchronously embedded and the second side information are obtained, and then inverse singular value decomposition is performed to obtain high-frequency coefficients containing watermark information.

[0037] The second SWT transformation unit is configured to obtain a watermark embedding image by performing inverse synchronous compression wavelet transformation based on the high-frequency coefficients containing the watermark information.

[0038] Preferably, the method for the sound-image watermark synchronous embedding unit to embed the original sound-image watermark into the singular value matrix of the host image by using the phase truncation algorithm and the complex function algorithm to obtain the matrix after the watermark is synchronously embedded, and obtain the first side information based on the amplitude truncation algorithm includes:

[0039] ;

[0040] ;

[0041] wherein, i represents an imaginary number, w v represents a two-dimensional audio watermark matrix reshaped from an original one-dimensional audio watermark, w I represents an original image watermark, w is an original sound-image watermark, PT (·) represents a phase truncation operation, w p represents a composite watermark obtained after the phase truncation operation is performed on the original sound-image watermark;

[0042] ;

[0043] wherein, AT (·) represents an amplitude truncation operation, w a represents first side information obtained after the amplitude truncation operation is performed on the original sound-image watermark;

[0044] ;

[0045] wherein, is a watermark embedding strength, is a singular value matrix of a host image, is a matrix after the watermark is synchronously embedded.

[0046] Preferably, the sound-image watermark synchronous extraction module comprises a third SWT transformation unit, a third SVD decomposition unit and a sound-image watermark synchronous extraction unit.

[0047] The third SWT transformation unit is configured to decompose the image into sub-blocks of different frequencies through one-dimensional Haar The synchronous compression wavelet transformation embeds the watermark into the image to obtain low-frequency coefficients, high-frequency coefficients in the horizontal direction, high-frequency coefficients in the vertical direction and high-frequency coefficients in the diagonal direction of the watermark-embedded image.

[0048] The third SVD decomposition unit is configured to perform singular value decomposition on the high-frequency coefficients in the horizontal direction of the watermark-embedded image to obtain a singular value matrix containing watermark information.

[0049] The sound-image watermark synchronous extraction unit is configured to perform inverse singular value decomposition on the singular value matrix containing watermark information based on the second side information, then perform a virtual part operation, and combine the first side information to obtain the embedded sound-image watermark, thereby realizing synchronous extraction of the sound-image watermark.

[0050] Compared with the prior art, the sound-image watermark synchronous embedding algorithm has the following advantages:

[0051] Compared with the additive watermark embedding algorithm in the prior art, the sound-image watermark synchronous embedding algorithm is mainly used for semi-blind extraction of the watermark without the original watermark information.

[0052] (1) Unlike the conventional scheme in which the watermark image is implanted into the host image by using the additive embedding method in the prior art, the sound-image watermark synchronous embedding algorithm is realized based on the phase truncation algorithm and the complex function embedding algorithm.

[0053] (2) Compared with the widely used non-blind watermark system based on SVD, the sound-image watermark synchronous embedding algorithm can realize semi-blind extraction of the watermark.

[0054] (3) The method can avoid false positive detection problems.

[0055] (4) The sound-image watermark synchronous embedding algorithm selects the watermark embedding position of the host image by using the SWT according to the human eye visual characteristics, improves the watermark embedding capacity and ensures the invisibility of the watermark.

[0056] (5) The complex function embedding algorithm can be applied to any additive watermark embedding method. BRIEF DESCRIPTION OF DRAWINGS

[0057] In order to more clearly illustrate the technical solutions of the present application, the following briefly introduces the drawings needed in the embodiments. Obviously, the drawings described below only show some of the embodiments of the present application, and for those skilled in the art, other drawings can be obtained from these drawings without creative labor.

[0058] Figure 1 The figure is a schematic diagram of the principle of semi-blind sound-image watermark synchronous embedding in the embodiment of the present application.

[0059] Figure 2 The figure is a schematic diagram of the principle of semi-blind sound-image watermark synchronous extraction in the embodiment of the present application.

[0060] Figure 3 The figure is the original sound-image watermark, the original host image, the host image after embedding the watermark, and the sound-image watermark extracted without any attack in the embodiment of the present application; wherein (a-1) is a two-dimensional audio watermark, (a-2) is an original image watermark, (b) is an original host image, (c) is a host image after embedding the watermark, (d-1) is a two-dimensional audio watermark extracted without any attack, and (d-2) is an image watermark extracted without any attack.

[0061] Figure 4 The figure is the sound-image watermark extracted from the host image after embedding the watermark, which is attacked by 0.05 Gaussian noise, 6.25% cutting attack, and 25% cutting attack in the embodiment of the present application; wherein (a-1) and (a-2) are the extraction results when attacked by 0.05 Gaussian noise, (b-1) and (b-2) are the extraction results when attacked by 6.25% cutting attack, and (c-1) and (c-2) are the extraction results when attacked by 25% cutting attack.

[0062] Figure 5 The figure is the sound-image watermark extracted from the host image after embedding the watermark, which is attacked by screenshot attack, JPEG compression attack, and moire attack in the embodiment of the present application; wherein (a-1) and (a-2) are the extraction results when attacked by screenshot attack, (b-1) and (b-2) are the extraction results when attacked by JPEG compression attack with a strength of 50, and (c-1) and (c-2) are the extraction results when attacked by moire attack. DETAILED DESCRIPTION

[0063] The technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only some of the embodiments of the present application, and not all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the present application.

[0064] In order to make the above objectives, characteristics and advantages of the present application more obvious and easy to understand, the present application will be further described in detail below with reference to the drawings and specific embodiments.

[0065] In view of the technical problems of false positive error detection defects, non-blind design limitations and weak resistance to camouflage attacks existing in the widely used singular value decomposition (SVD) based additive watermark embedding system, the present application designs a new method of fusing sound-image watermark and host based on complex function, which can effectively avoid the false positive error detection phenomenon of sound-image watermark and has strong resistance to image processing attacks while realizing the semi-blind extraction function of sound-image watermark. Therefore, the present application constructs a semi-blind watermark method and system based on the sound-image watermark synchronous embedding algorithm of SWT, and the technical scheme adopted is that in the watermark system, a semi-blind watermark scheme based on the sound-image watermark synchronous embedding algorithm of SWT is designed and realized. According to the human eye visual characteristics, the SWT is adopted for the first time to select the watermark embedding position of the host image, which improves the watermark embedding capacity while ensuring the invisibility of the watermark; with the aid of the characteristics of the complex function, the original sound-image watermark is synchronously embedded into the host image through the phase truncation algorithm and the complex function algorithm. In this way, in the extraction process, only the imaginary part operation is needed for the host image after embedding the watermark, and the watermark information can be recovered. The whole extraction process avoids the participation of the original watermark, and realizes the semi-blind extraction of the watermark.

[0066] The related theorem of the complex function embedding algorithm adopted by the present application is as follows: for a square matrix , the matrix can be obtained through the complex function embedding algorithm.

[0067] ;

[0068] Through derivation, it can be obtained that:

[0069] ;

[0070] Wherein, represents an imaginary number, is a constant greater than 0, represents the real part operation, represents the imaginary part operation. Therefore, without the participation of the matrix , the input matrix can be directly recovered by the imaginary part operation of the matrix .

[0071] The semi-blind watermark method of the sound-image watermark synchronous embedding algorithm based on SWT proposed by the present application uses the above characteristics of the complex function embedding algorithm, and takes the host image and the watermark image as the matrix and Thus, only the imaginary part operation is needed in the extraction process, and the watermark information can be recovered. The whole extraction process avoids the participation of the original watermark and host image, and blind extraction of the watermark can be realized.

[0072] Embodiment one

[0073] The application provides a semi-blind watermark method based on a SWT-based sound-image watermark synchronous embedding algorithm, which comprises the following steps:

[0074] The phase truncation algorithm and the complex function algorithm are used to synchronously embed sound-image watermarks into a host image to be embedded with a watermark, so as to obtain a watermark embedding image.

[0075] The sound-image watermark is synchronously extracted from the watermark embedding image, and the amplitude truncation algorithm is combined to obtain the embedded sound-image watermark, so that the synchronous extraction of the sound-image watermark is realized.

[0076] As shown in Figure 1 , the method for synchronously embedding sound-image watermarks into a host image to be embedded with a watermark based on the phase truncation algorithm and the complex function algorithm comprises the following steps:

[0077] First stage: SWT

[0078] First step: decompose the host image to be embedded with a watermark into subblocks of different frequencies through one-dimensional synchronous compression wavelet transform (CWT) Haar , so as to obtain low-frequency coefficients SWT , high-frequency coefficients in a horizontal direction IH , high-frequency coefficients in a vertical direction CA , and high-frequency coefficients in a diagonal direction CH : CV CD

[0079] ;

[0080] wherein haar represents a wavelet transform base. Haar Based on the wavelet transform, the synchronous compression method is used to extract the wavelet ridge line, so that the time-frequency expression effect of the harmonic signal is clearer. SWT Second step: perform

[0081] on the high-frequency coefficients in the horizontal direction CH of the host image to obtain SVD , so as to obtain a singular value matrix of the host image:

[0082] ;

[0083] wherein U and​​​V both represent orthogonal matrix of host image.

[0084] Second stage: synchronous embedding of sound-image watermark

[0085] Third step: synchronous embedding of original sound-image watermark into singular value matrix of host image by phase truncation algorithm and complex function algorithm wherein matrix is obtained after synchronous embedding of watermark, and

[0086] ;

[0087] ;

[0088] wherein, i represents imaginary number, w v represents two-dimensional audio watermark matrix reshaped from original one-dimensional audio watermark voice ; w I represents original image watermark, w is original sound-image watermark, PT (·) represents phase truncation operation, w p represents the resultant watermark after phase truncation operation on original sound-image watermark.

[0089] ;

[0090] wherein, AT (·) represents amplitude truncation operation, w a represents first side information after amplitude truncation operation on original sound-image watermark.

[0091] ;

[0092] wherein, is watermark embedding strength and is set to 0.01 in this embodiment, is singular value matrix of host image, is matrix after synchronous embedding of watermark;

[0093] Third stage: ISWT

[0094] Fourth step: performing on matrix SVD after synchronous embedding of watermark to obtain wherein S 1 is singular value matrix of matrix after synchronous embedding of watermark. save as the second side information, wherein U 1 and V 1 represent orthogonal matrices of the matrix . Then, by inverse singular value decomposition, the high frequency coefficients containing the watermark information are obtained ISVD :

[0095] ;

[0096] Fifth step: based on the high frequency coefficients containing the watermark information, by inverse synchronous compression wavelet transform, the watermark embedding image is obtained ISWT :

[0097] .

[0098] As shown in Figure 2 , the sound-image watermark synchronous extraction is carried out on the watermark embedding image to obtain the embedded sound-image watermark, and the method for realizing the synchronous extraction of the sound-image watermark comprises:

[0099] First stage: SWT

[0100] First step: the watermark embedding image is decomposed into sub-blocks of different frequencies by one-dimensional Haar synchronous compression wavelet transform, to obtain the low frequency coefficients CAT , the high frequency coefficients in the horizontal direction CHT , the high frequency coefficients in the vertical direction CVT and the high frequency coefficients in the diagonal direction CDT of the watermark embedding image:

[0101] ;

[0102] Second step: singular value decomposition is carried out on the high frequency coefficients in the horizontal direction CHT of the watermark embedding image to obtain , wherein U 2 and V 2 represent orthogonal matrices of the high frequency coefficients CHT , and the singular value matrix containing the watermark information is obtained :

[0103] ;

[0104] Second stage: sound-image watermark synchronous extraction

[0105] Third step: based on the second side information (1, U 1, V 1), the singular value matrix containing the watermark information ​​Perform inverse singular value decomposition, and then obtain the embedded synthetic watermark by taking the imaginary part. :

[0106] ;

[0107] in, T This represents the matrix transpose operation. This represents the operation of taking the imaginary part.

[0108] Step 4: Create the watermark The first side information obtained by the amplitude truncation algorithm By combining the imaginary and real parts, the embedded audio watermark is obtained. With image watermark :

[0109] ;

[0110] ;

[0111] ;

[0112] in, For the initial watermark extraction, This represents the real part extraction operation. Finally, the two-dimensional audio watermark is... By reshaping it into one dimension, the embedded audio-visual watermark can be obtained. Voice .

[0113] Figure 3 (a-1) in the image represents a two-dimensional audio watermark. Figure 3 (a-2) in the image is the original image watermark. Figure 3 (b) in the image is the original host image. Figure 3 As shown in (c), the host image after embedding the watermark is indistinguishable from the original host image to the naked eye. The two-dimensional audio watermark and image watermark extracted without any attack are as follows: Figure 3 (d-1) and Figure 3As shown in (d-2), it can be seen that when the host image is not attacked, the obtained acoustic-image watermark is indistinguishable from the input acoustic-image watermark. Furthermore, to quantitatively evaluate the extraction results, this invention calculated the correlation coefficient, signal-to-noise ratio (SNR), and power-to-noise ratio (PSNR) between the original and extracted acoustic-image watermarks. The calculations showed that the correlation coefficient between the watermarked host image and the original host image was 1, and the PSNR was 84.0612 dB. These results indicate that even when simultaneously embedding an acoustic-image watermark of the same size as the host image, the watermarked host image maintains good invisibility. In addition, the correlation coefficient between the extracted image watermark and the original image watermark was 0.9982, and the SNR between the extracted audio watermark and the original audio watermark was 39.8618 dB, indicating that there is almost no difference between the extracted and original acoustic-image watermarks. Therefore, the semi-blind watermarking method based on SWT-based synchronous embedding algorithm proposed in this invention can be effectively used for the synchronous extraction of acoustic-image watermarks.

[0114] The attacks employed by this invention on the host image after embedding a watermark include: extraction results under Gaussian noise attack with an intensity of 0.05, cropping attacks with intensities of 6.25% and 25%, screenshot attacks, JPEG compression attacks with an intensity of 50, and simulated moiré pattern attacks. Figure 4 , Figure 5 For the host image after embedding the watermark ( Figure 3 In (c) of the image processing diagram, the extracted audio-image watermarks after being subjected to different image processing attacks show that the extracted watermarks still contain the main information of the original watermark. Notably, even when the host image with the embedded audio-image watermark is subjected to a screenshot attack, the main information of the extracted audio-image watermark can still be identified. To quantitatively evaluate the extraction results, this invention calculated the correlation coefficient and signal-to-noise ratio (SNR) values ​​between the original and extracted audio-image watermarks. The calculated SNR values ​​between the extracted audio watermark and the original audio watermark were 20.65 dB, 21.69 dB, 18.09 dB, 31.49 dB, 22.66 dB, and 20.27 dB, respectively. The correlation coefficient values ​​between the extracted image watermark and the original image watermark were 0.9558, 0.9631, 0.7943, 0.9901, 0.9718, and 0.9273, respectively. The above results demonstrate that the semi-blind watermarking method based on the SWT-based audio-image watermarking synchronous embedding algorithm proposed in this invention has a certain degree of robustness against various image processing attacks, especially screenshot attacks.

[0115] In conclusion, the method directly embeds the watermark to avoid the false positive problem of the watermark; the watermark extraction process of the semi-blind watermark method based on the SWT-based sound-image watermark synchronous embedding algorithm does not need the participation of the original watermark, so that the semi-blind extraction of the sound-image watermark is realized; and the application can resist image processing attacks such as cutting, screen capture and simulated moire.

[0116] Embodiment two

[0117] Based on the same inventive concept, the application further provides a semi-blind watermark system based on an SWT-based sound-image watermark synchronous embedding algorithm, which is used to realize the method in the foregoing embodiments, and the system comprises a sound-image watermark synchronous embedding module and a sound-image watermark synchronous extraction module.

[0118] The sound-image watermark synchronous embedding module is used to synchronously embed a sound-image watermark into a host image to be embedded with the watermark based on a phase truncation algorithm and a complex function algorithm, so as to obtain a watermark embedded image.

[0119] The sound-image watermark synchronous extraction module is used to synchronously extract the sound-image watermark from the watermark embedded image, so as to obtain the embedded sound-image watermark and realize the synchronous extraction of the sound-image watermark.

[0120] Further, the sound-image watermark synchronous embedding module comprises a first SWT transformation unit, a first SVD decomposition unit, a sound-image watermark synchronous embedding unit, a second SVD decomposition unit and a second SWT transformation unit.

[0121] The first SWT transformation unit is used to decompose the host image to be embedded with the watermark into subblocks of different frequencies through one-dimensional synchronous compression wavelet transformation, so as to obtain low-frequency coefficients, high-frequency coefficients in a horizontal direction, high-frequency coefficients in a vertical direction and high-frequency coefficients in a diagonal direction of the host image. Haar

[0122] The first SVD decomposition unit is used to perform singular value decomposition on the high-frequency coefficients in the horizontal direction of the host image, so as to obtain a singular value matrix of the host image. SVD

[0123] The sound-image watermark synchronous embedding unit is used to synchronously embed the original sound-image watermark into the singular value matrix of the host image through the phase truncation algorithm and the complex function algorithm, so as to obtain a matrix after synchronous embedding of the watermark, and simultaneously obtain first side information based on the amplitude truncation algorithm.

[0124] The second SVD decomposition unit is used to perform singular value decomposition on the matrix after synchronous embedding of the watermark, so as to obtain a singular value matrix of the matrix after synchronous embedding of the watermark and second side information, and then obtain high-frequency coefficients containing watermark information through inverse singular value decomposition. SVD

[0125] ​​​The second SWT transformation unit is configured to obtain the watermark embedding image by inverse synchronous compression wavelet transformation based on the high frequency coefficient containing the watermark information.

[0126] Further, the sound-image watermark synchronous embedding unit synchronously embeds the original sound-image watermark into the singular value matrix of the host image by a phase truncation algorithm and a complex function algorithm to obtain a matrix after synchronous watermark embedding, and a method for obtaining the first side information based on an amplitude truncation algorithm comprises:

[0127] ;

[0128] ;

[0129] wherein, i represents an imaginary number, w v represents a two-dimensional audio watermark matrix reshaped from the original one-dimensional audio watermark, w I represents an original image watermark, w is an original sound-image watermark, PT (·) represents a phase truncation operation, w p represents a composite watermark obtained after the phase truncation operation on the original sound-image watermark;

[0130] ;

[0131] wherein, AT (·) represents an amplitude truncation operation, w a represents the first side information obtained after the amplitude truncation operation on the original sound-image watermark;

[0132] ;

[0133] wherein, is a watermark embedding strength, is a singular value matrix of a host image, is a matrix after synchronous watermark embedding.

[0134] Further, the sound-image watermark synchronous extraction module comprises a third SWT transformation unit, a third SVD decomposition unit, and a sound-image watermark synchronous extraction unit.

[0135] The third SWT transformation unit is configured to decompose the watermark embedding image into sub-blocks of different frequencies by one-dimensional Haar synchronous compression wavelet transformation to obtain low frequency coefficients, high frequency coefficients in the horizontal direction, high frequency coefficients in the vertical direction, and high frequency coefficients in the diagonal direction of the watermark embedding image;

[0136] a third SVD decomposition unit configured to perform singular value decomposition on the high frequency coefficients in the horizontal direction of the watermark embedding image to obtain a singular value matrix containing watermark information;

[0137] a sound-map watermark synchronous extraction unit configured to perform inverse singular value decomposition on the singular value matrix containing watermark information based on the second side information, then obtain the embedded sound-map watermark by taking the imaginary part operation, and combine the first side information to realize synchronous extraction of the sound-map watermark.

[0138] The above-described embodiments are merely intended to describe the preferred modes of the present application, and are not intended to limit the scope of the present application. Various modifications and improvements to the technical solutions of the present application made by those of ordinary skill in the art without departing from the design spirit of the present application shall fall within the protection scope of the present application as defined by the claims.

Claims

1. A semi-blind watermarking method based on the SWT-based audio-image watermarking synchronous embedding algorithm, characterized in that, The method includes: The acoustic-image watermark is simultaneously embedded into the host image to be watermarked based on the phase truncation algorithm and the complex function algorithm to obtain the watermark-embedded image. The watermark-embedded image is extracted synchronously with the sound-image watermark to obtain the embedded sound-image watermark, thus achieving synchronous extraction of the sound-image watermark. Methods for simultaneously embedding acoustic-image watermarks into the host image to be watermarked, based on phase truncation and complex function algorithms, to obtain watermark-embedded images include: Step 1: Through one-dimensional Haar Synchronous compressed wavelet transform decomposes the host image to be watermarked into sub-blocks of different frequencies, obtaining the low-frequency coefficients, high-frequency coefficients in the horizontal direction, high-frequency coefficients in the vertical direction, and high-frequency coefficients in the diagonal direction of the host image. Step 2: Perform high-frequency coefficient analysis on the horizontal direction of the host image. SVD , to obtain the singular value matrix of the host image; Step 3: The original sound-image watermark is synchronously embedded into the singular value matrix of the host image using a phase truncation algorithm and a complex function algorithm to obtain the matrix after synchronous watermark embedding. At the same time, the first side information is obtained based on the amplitude truncation algorithm. Step 4: Perform the following steps on the matrix after synchronously embedding the watermark: SVD The singular value matrix and second side information of the matrix after synchronous watermark embedding are obtained, and then the high-frequency coefficients containing watermark information are obtained through inverse singular value decomposition. Step 5: Based on the high-frequency coefficients containing watermark information, obtain the watermark-embedded image through inverse synchronous compressed wavelet transform; The original acoustic-image watermark is synchronously embedded into the singular value matrix of the host image using a phase truncation algorithm and a complex function algorithm to obtain the matrix after synchronous watermark embedding. Simultaneously, the method for obtaining the first side information based on the amplitude truncation algorithm includes: ； ； in, i represents an imaginary number, w v This represents a two-dimensional audio watermark matrix reshaped from the original one-dimensional audio watermark. w I Represents the original image watermark. w For the original sound-image watermark, PT (·) represents a phase truncation operation. w p This represents the composite watermark obtained after performing a phase truncation operation on the original acoustic-image watermark. ； in, AT (·) represents amplitude truncation. w a This represents the first side information obtained after performing an amplitude truncation operation on the original sound-image watermark. ； in, For watermark embedding strength, The singular value matrix of the host image. This is the matrix after the watermark is embedded synchronously.

2. The method according to claim 1, characterized in that, The method for synchronously extracting the sound-image watermark from the watermark-embedded image to obtain the embedded sound-image watermark includes: Step 1: Through one-dimensional Haar Synchronous compressed wavelet transform decomposes the watermark-embedded image into sub-blocks of different frequencies, obtaining the low-frequency coefficients, high-frequency coefficients in the horizontal direction, high-frequency coefficients in the vertical direction, and high-frequency coefficients in the diagonal direction of the watermark-embedded image. Step 2: Perform singular value decomposition on the high-frequency coefficients in the horizontal direction of the watermark embedded in the image to obtain a singular value matrix containing watermark information. Step 3: Based on the second side information, perform inverse singular value decomposition on the singular value matrix containing watermark information, and then obtain the embedded sound-image watermark by taking the imaginary part and combining it with the first side information, so as to realize the synchronous extraction of sound-image watermark.

3. A semi-blind watermarking system based on an SWT-based audio-image watermarking synchronous embedding algorithm, the system being used to implement the method described in any one of claims 1-2, characterized in that, The system includes: a synchronous embedding module for sound-image watermarking and a synchronous extraction module for sound-image watermarking; The acoustic-image watermarking synchronous embedding module is used to synchronously embed the acoustic-image watermark into the host image to be embedded with the watermark based on the phase truncation algorithm and the complex function algorithm, so as to obtain the watermark-embedded image. The audio-image watermark synchronous extraction module is used to extract the audio-image watermark synchronously from the watermark embedded in the image, thereby obtaining the embedded audio-image watermark and realizing the synchronous extraction of the audio-image watermark.

4. The system according to claim 3, characterized in that, The audio-image watermarking synchronous embedding module includes: a first SWT transformation unit, a first SVD decomposition unit, an audio-image watermarking synchronous embedding unit, a second SVD decomposition unit, and a second SWT transformation unit. The first SWT transformation unit is used to transform a one-dimensional... Haar Synchronous compressed wavelet transform decomposes the host image to be watermarked into sub-blocks of different frequencies, obtaining the low-frequency coefficients, high-frequency coefficients in the horizontal direction, high-frequency coefficients in the vertical direction, and high-frequency coefficients in the diagonal direction of the host image. The first SVD decomposition unit is used to decompose the high-frequency coefficients in the horizontal direction of the host image. SVD , to obtain the singular value matrix of the host image; The acoustic-image watermarking synchronous embedding unit is used to synchronously embed the original acoustic-image watermark into the singular value matrix of the host image through a phase truncation algorithm and a complex function algorithm to obtain the matrix after synchronous watermarking, and at the same time obtain the first side information based on the amplitude truncation algorithm. The second SVD decomposition unit is used to perform the same process on the matrix after synchronous watermark embedding. SVD The singular value matrix and second side information of the matrix after synchronous watermark embedding are obtained, and then the high-frequency coefficients containing watermark information are obtained through inverse singular value decomposition. The second SWT transform unit is used to obtain the watermarked image by inverse synchronous compressed wavelet transform based on the high-frequency coefficients containing watermark information.

5. The system according to claim 4, characterized in that, The acoustic-image watermarking synchronous embedding unit synchronously embeds the original acoustic-image watermark into the singular value matrix of the host image using a phase truncation algorithm and a complex function algorithm, obtaining a matrix after synchronous watermark embedding. Simultaneously, the method for obtaining the first side information based on the amplitude truncation algorithm includes: ； ； in, i represents an imaginary number, w v This represents a two-dimensional audio watermark matrix reshaped from the original one-dimensional audio watermark. w I Represents the original image watermark. w For the original sound-image watermark, PT (·) represents a phase truncation operation. w p This represents the composite watermark obtained after performing a phase truncation operation on the original acoustic-image watermark. ； in, AT (·) represents amplitude truncation. w a This represents the first side information obtained after performing an amplitude truncation operation on the original sound-image watermark. ； in, For watermark embedding strength, The singular value matrix of the host image. This is the matrix after the watermark is embedded synchronously.

6. The system according to claim 5, characterized in that, The synchronous extraction module for sound-image watermarking includes: a third SWT transformation unit, a third SVD decomposition unit, and a synchronous extraction module for sound-image watermarking. The third SWT transformation unit is used to transform one dimension... Haar Synchronous compressed wavelet transform decomposes the watermark-embedded image into sub-blocks of different frequencies, obtaining the low-frequency coefficients, high-frequency coefficients in the horizontal direction, high-frequency coefficients in the vertical direction, and high-frequency coefficients in the diagonal direction of the watermark-embedded image. The third SVD decomposition unit is used to perform singular value decomposition on the high-frequency coefficients in the horizontal direction of the watermark embedded image to obtain a singular value matrix containing watermark information. The sound-image watermark synchronous extraction unit is used to perform inverse singular value decomposition on the singular value matrix containing watermark information based on the second side information, and then obtain the embedded sound-image watermark by taking the imaginary part operation and combining it with the first side information, so as to realize the synchronous extraction of sound-image watermark.

Images