Image watermark embedding method, extraction method, device, equipment and storage medium
The watermark information is embedded through K-layer fractional-order wavelet transformation and fractional-order Fourier transformation, and the hybrid optical bistable model is used to mess and reverse the mess, which solves the problem of watermark robustness and insufficient information capacity in the existing technology, and realizes the Higher security and robustness.
Patent Information
- Application Number
- CN202410279949.3
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-03-12
- Publication Date
- 2024-11-19
- Estimated Expiration
- 2044-03-12
AI Technical Summary
The existing digital watermarking technology has shortcomings in terms of robustness and information capacity, and cannot effectively resist attacks such as filtering and compression, and the security performance of a single time frequency domain method needs to be improved.
Using K-layer fractional-order wavelet transformation and fractional-order Fourier transformation, the watermark information is embedded in the complex coefficient of the image, and the phase information is retained through fractional-order wavelet inverse transformation, increasing the capacity and security of the watermark information, and using a hybrid optical bistable model Stop and reverse the chaos to extract the watermark.
It improves the security and robustness of watermarks, enhances the capacity of watermark information, and can effectively extract watermarks when the amplitude information of the carrier image is small, improving the ability to protect digital copyrights.
Smart Images

Figure CN118014815B_ABST
Abstract
Description
Technical Field
[0001] The present invention belongs to the technical field of image processing, and in particular relates to an image watermark embedding method, an extraction method, a device, a equipment and a storage medium. Background Art
[0002] With the rapid development of digital communications, the information security and copyright protection issues of digital information transmission have arisen. Digital watermarking technology is an effective means to solve this problem. By identifying the identification information embedded in the digital carrier, it can determine whether the information carrier has been modified, thereby achieving the purpose of copyright protection and integrity authentication of digital images.
[0003] According to the application requirements of digital watermarking, which should be imperceptible, robust and secure, early research on digital watermarking technology was based on spatial domain algorithms, which ensured the invisibility of the watermark. However, the amount of embedded information was small, and the ability to resist attacks such as filtering and compression was weak, which could not meet the robustness requirements of digital watermarking, and its robustness needed to be further improved.
[0004] The transform domain method of digital watermarking technology embeds digital watermark information by modifying the transform domain coefficients of the original carrier. The transform domain digital watermarking algorithm mainly focuses on a series of methods such as DFT, DCT, DWT and hybrid transform domain methods. The watermark information is embedded by changing some coefficient values of some frequency bands after the host image is transformed. The algorithm reflects the idea of embedding the watermark information into the higher energy part of the image to ensure the invisibility of the watermark. However, due to the imbalance of high and low frequency coefficients and the imperceptibility of the watermark, the above method cannot embed too much information. In addition, the above method belongs to a single time-frequency domain method. Since the decomposition coefficients are relatively fixed, the security performance needs to be improved. Summary of the invention
[0005] In order to overcome the above-mentioned shortcomings of the prior art, the present invention provides an image watermark embedding method, comprising the following steps:
[0006] Perform K-layer p-order fractional wavelet transform on the original image I to obtain 3K+1 subbands;
[0007] Performing an s-order fractional Fourier transform on the K-th layer subband of the transformed image to obtain complex coefficients FLL, FHL, FLH and FHH after subband transformation;
[0008] Remove the correlation between the pixels of the watermark image, and perform a two-dimensional fractional transform on the removed pixels to obtain the complex coefficients FW;
[0009] Embed the real and imaginary parts of the complex coefficient FW into FLL, FHL, FLH and FHH to obtain watermarked coefficients FLL′, FHL′, FLH′ and FHH′;
[0010] Performing s-order fractional Fourier inverse transform on FLL′, FHL′, FLH′ and FHH′ to obtain the first watermark carrier image;
[0011] Perform K-layer p-order fractional wavelet inverse transform on the 4 subband coefficients of the Kth layer and the remaining subband coefficients of the first watermark carrier image to obtain the reconstructed image complex coefficients I W , retaining the complex coefficient phase information angle(I W ) and I W Find the modulus and get the second watermark carrier image I αW ;
[0012] The second watermark carrier image I αW The difference between the original image I and the watermark mode ΔW loaded into the second watermark carrier image is extracted, the amplitude of ΔW is limited to obtain the limited watermark ΔW′, and ΔW′ is re-embedded into the original image I to obtain the final watermark carrier image I △W .
[0013] Preferably, removing the correlation between pixels of the watermark image and performing a two-dimensional fractional order transform on the removed pixels to obtain complex coefficients FW comprises the following steps:
[0014] The Arnold transform algorithm is used to transform the original watermark image W for n times to obtain the scrambled watermark image W′.
[0015] Perform a q-order fractional Fourier transform on the scrambled image W′ to obtain the transformed complex coefficients FW.
[0016] Preferably, the value range of the s order is s<10 -3 .
[0017] Preferably, the value range of the q order is q<10 -3 .
[0018] Preferably, the original image I is an image of N×N pixels, and the original watermark image W is an image of M×M pixels, where M=N / 2 k .
[0019] Preferably, the present invention also provides an image watermark extraction method, comprising the following steps:
[0020] In the watermark carrier image I △W Add phase information to get the complex coefficient I′ ΔW ;
[0021] For complex coefficient I′ ΔW Perform K-layer p-order fractional wavelet transform to obtain 3K+1 subbands of fractional wavelet transform;
[0022] For the K-th layer subband coefficient {HH Δ3 ,HL Δ3 ,LH Δ3 ,HH Δ3} Perform a q-order two-dimensional fractional Fourier transform to obtain the complex coefficients {FLL Δ3 , FHL Δ,3 , FLH Δ3 , FHH Δ3};
[0023] For complex coefficients {FLL Δ3 , FHL Δ3 , FLH Δ3 , FHH Δ3} performs difference operation with {FLL3, FLH3, FHL3, FHH3} to extract the real part FW1′ and imaginary part FW2′ of {FLLΔ3, FHLΔ3, FLHΔ3, FHHΔ3};
[0024] Perform q-order fractional Fourier inverse transform on the real part FW1′ and the imaginary part FW2′, and perform modulus operation on the transformed complex coefficients to extract the watermark image;
[0025] The watermark information is extracted by transforming the initial position according to the hybrid optical bistable model and performing inverse scrambling on the watermark information.
[0026] The present invention also provides an image watermark embedding device, comprising:
[0027] The subband acquisition module is used to perform K-layer p-order fractional wavelet transform on the original image I to obtain 3K+1 subbands;
[0028] A fractional Fourier transform module is used to perform an s-order fractional Fourier transform on the K-th layer subband of the transformed image to obtain complex coefficients FLL, FHL, FLH and FHH after subband transformation;
[0029] A watermark image preprocessing module is used to preprocess the watermark image to obtain a complex coefficient FW;
[0030] A complex coefficient embedding module, used for embedding the real part and the imaginary part of the complex coefficient FW into FLL, FHL, FLH and FHH to obtain watermarked coefficients FLL′, FHL′, FLH′ and FHH′;
[0031] A fractional Fourier inverse transform module is used to perform an s-order fractional Fourier inverse transform on FLL′, FHL′, FLH′ and FHH′ to obtain a first watermark carrier image;
[0032] The image reconstruction module is used to perform K-layer p-order fractional wavelet inverse transformation on the four subband coefficients of the Kth layer and the subband coefficients of the remaining layers of the first watermark carrier image to obtain the reconstructed image complex coefficients I W , retaining the complex coefficient phase information angle(I W ) and I W Find the modulus and get the second watermark carrier image I aW ;
[0033] The limiting module is used to convert the second watermark carrier image I aW The difference between the original image I and the watermark mode ΔW loaded into the second watermark carrier image is extracted, the amplitude of ΔW is limited to obtain the limited watermark ΔW′, and ΔW′ is re-embedded into the original image I to obtain the final watermark carrier image I △W .
[0034] The present invention also provides an image watermark extraction device, comprising:
[0035] The first complex coefficient acquisition module is used to obtain the watermark carrier image I △W Add phase information to get the complex coefficient I′ ΔW ;
[0036] Fractional wavelet transform module, used for complex coefficients I′ ΔW Perform K-layer p-order fractional wavelet transform to obtain 3K+1 subbands of fractional wavelet transform;
[0037] The second complex coefficient acquisition module is used to obtain the K-th layer subband coefficient {HH Δ3 ,HL Δ3 ,LH Δ3 ,HH Δ3} Perform a q-order two-dimensional fractional Fourier transform to obtain the complex coefficients {FLL Δ3 , FHL Δ3 , FLH Δ3 , FHH Δ3};
[0038] Difference operation module, used for complex coefficients {FLL Δ3 ,FHL Δ3 ,FLH Δ3 ,FHH Δ3} and {FLL3, FLH3, FHL3, FHH3} to extract {FLL Δ3 , FHL Δ3 , FLH Δ3 , FHH Δ3}real part FW1′ and imaginary part FW2′;
[0039] The watermark image acquisition module is used to perform a q-order fractional Fourier inverse transform on the real part FW1′ and the imaginary part FW2′, and perform a modulus operation on the transformed complex coefficients to extract the watermark image;
[0040] The watermark extraction module is used to transform the initial position according to the hybrid optical bistable model, perform inverse scrambling on the watermark information, and extract the watermark.
[0041] The present invention also provides a computer device, comprising a memory and a processor; the memory stores a computer program, and the processor is used to run the computer program in the memory to execute the steps of the image watermark embedding method and the steps of the image watermark extraction method.
[0042] The present invention also provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, wherein the computer program is suitable for being loaded by a processor to execute the steps of the image watermark embedding method and the steps of the image watermark extraction method.
[0043] The image watermark embedding method, extraction method, device, equipment and storage medium provided by the present invention have the following beneficial effects:
[0044] The present invention embeds the complex coefficient FW of the watermark image into the carrier image that has undergone fractional wavelet transform and fractional Fourier transform, and can perform appropriate encoding or deformation according to the characteristics of the transform coefficients, or selectively embed by blocks, thereby increasing the complexity of the watermark algorithm; by utilizing the phase information retained by the fractional wavelet inverse transform after the image is embedded in the watermark, the capacity of the watermark information can be increased when the amplitude information in the carrier image is small; in addition, the transformations of the watermark image and the carrier image of the present invention are both complex transformations, and the transformation results provide coefficient phase information. The phase spectrum of the image well retains the information of the edge and overall structure of the image, and the security and robustness of the watermark are improved by utilizing the reconstructed coefficient phase information. BRIEF DESCRIPTION OF THE DRAWINGS
[0045] In order to more clearly illustrate the embodiment of the present invention and its design scheme, the following briefly introduces the drawings required for this embodiment. The drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.
[0046] Figure 1 is a flow chart of an image watermark embedding method according to an embodiment of the present invention;
[0047] Figure 2 is the original image;
[0048] Figure 3 For watermarking and scrambling watermarked images;
[0049] Figure 4 Extract the watermark front image for embedding after scrambling;
[0050] Figure 5 Extract watermark three-dimensional stereogram for embedding after scrambling;
[0051] Figure 6 A three-dimensional image with a clipped watermark embedded after scrambling;
[0052] Figure 7 For embedding infinite watermark image, PSNR = 31.2552dB;
[0053] Figure 8 For embedding limited clip watermark image, PSNR = 51.0316dB;
[0054] Fig. 9 For normal watermark extraction (p = 0.0000000054874331), NC = 0.9717;
[0055] Fig.10 It is the phase image of the carrier after reconstructing after embedding the watermark;
[0056] Fig.11 The watermark image is extracted for the key error (p=0.0000000054874332);
[0057] Fig.12 To cut the attack image;
[0058] Fig.13 is the noise attack image;
[0059] Fig.14 To filter the attack image;
[0060] Fig.15 To histogram equalize the attack image;
[0061] Fig.16 To compress the attack image;
[0062] Fig.17 Brighten images for images;
[0063] Fig.18 Darken the image for the image;
[0064] Fig.19 Add contrast to the image;
[0065] Fig. 20 To reduce contrast images;
[0066] Fig.21 For the rotation attack image. DETAILED DESCRIPTION
[0067] In order to enable those skilled in the art to better understand the technical solution of the present invention and implement it, the present invention is described in detail below in conjunction with the accompanying drawings and specific embodiments. The following embodiments are only used to more clearly illustrate the technical solution of the present invention, and cannot be used to limit the scope of protection of the present invention.
[0068] In the description of the present invention, it is to be understood that the terms “center”, “longitudinal”, “lateral”, “length”, “width”, “thickness”, “up”, “down”, “front”, “back”, “left”, “right”, “vertical”, “horizontal”, “top”, “bottom”, “inside”, “outside”, “axial”, “radial”, “circumferential”, etc., indicating orientations or positional relationships based on the orientations or positional relationships shown in the accompanying drawings, are only for the convenience of describing the technical solutions of the present invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as a limitation on the present invention.
[0069] In addition, the terms "first", "second", etc. are used for descriptive purposes only and are not to be understood as indicating or implying relative importance. In the description of the present invention, it should be noted that, unless otherwise clearly specified or limited, the terms "connected" and "connected" should be understood in a broad sense. For example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection or an indirect connection through an intermediate medium. For ordinary technicians in this field, the specific meanings of the above terms in the present invention can be understood according to the specific circumstances. In the description of the present invention, unless otherwise specified, "plurality" means two or more, which will not be described in detail here.
[0070] Example
[0071] The present invention provides an image watermark embedding method, specifically as follows Figure 1 As shown, the following steps are included:
[0072] Step 1: Perform K-layer p-order fractional wavelet transform on the original image I to obtain 3K+1 sub-bands; select any two sub-bands from the K-th layer of sub-bands to perform s-order fractional Fourier transform to obtain the complex coefficients FLL and FHL after sub-band transformation.
[0073] The present invention specifically performs a three-layer two-dimensional fractional wavelet transform of order p1=p2=0.000000004992489657 on an original image "Vase.tif" of size 512×512, and selects four sub-band complex coefficients {LL3, HL3, LH3, HH3} of the third layer;
[0074] The four subbands are subjected to fractional Fourier transform with the order of q1=q2=0.000005105.
[0075]
[0076]
[0077] The two-subband fractional-order Fourier frequency domain coefficients {FLL3, FLH3, FHL3, FHH3} are obtained.
[0078] Step 2: Remove the correlation between the pixels of the watermark image, and perform a two-dimensional fractional order transform on the removed pixels to obtain the complex coefficient FW, specifically:.
[0079] The hybrid optical bistable model is used to make the watermark W generate a chaotic sequence, and the watermark pixel point (x, y) is transformed to a new coordinate point (x′, y′) to obtain the scrambled watermarks W′ and W″;
[0080] Among them, the initial values of the watermark pixel are x0=1.038, x1=0.5087, and the hybrid optical bistable model is as follows
[0081] x n+1 =A sin 2(x n -B) (A=4,B=2.5)
[0082] Perform two-dimensional fractional Fourier transform of order q1=q2=0.000005105 on the scrambled watermarks W′ and W″, respectively, and convert the watermark coefficients into fractional frequency domain coefficients FW1 and FW2.
[0083]
[0084] Step 3: Embed the real and imaginary parts of the complex coefficient FW into FLL, FHL, FLH and FHH to obtain watermarked coefficients FLL′, FHL′, FLH′ and FHH′.
[0085] The real part and imaginary part FW1, FW2 of FW are embedded in the sub-band transform coefficients, and the scaling coefficients a1, a2, a3, and a4 may not be unique and may be selected as needed, generally (0.2, -1) or (-1, -0.2). In this embodiment, the specific values of the scaling coefficients are a1 = 0.5, a2 = -0.8, a3 = 0.5, and a4 = -0.8.
[0086] FLL3′=real(FLL3)+imag(FLL3)*j+0.5*real(FW1);
[0087] FLH3′=real(FLH3)+imag(FLH3)*j-0.8*imag(FW1)*j;
[0088] FHL3′=real(FHL3)+imag(FHL3)*j+0.5*real(FW2);
[0089] FHH3′=real(FHH3)+imag(FHH3)*j-0.8*imag(FW2)*j;
[0090] The real and imaginary parts of FW1 and FW2 are embedded in different sub-bands and different scaling coefficients are selected to increase the security of watermark information and reduce the false alarm rate of watermark extraction.
[0091] Step 4: Perform s-order fractional Fourier inverse transform on FLL′, FHL′, FLH′ and FHH′ to obtain the first watermark carrier image.
[0092] Perform two-dimensional fractional Fourier inverse transform of order q1=q2=0.000005105 on FLL′3, FLH′3, FHL′3, and FHH′3, convert the fractional domain coefficients into fractional wavelet domain, and obtain the Kth layer subband coefficients [LL′3, LH′3, HL′3, HH′3] of embedded watermark information
[0093]
[0094]
[0095] Step 5: Perform K-layer p-order fractional wavelet inverse transform on the 4 subband coefficients of the Kth layer of the first watermark carrier image and the subband coefficients of the remaining layers to obtain the reconstructed image complex coefficients I W , retain the complex coefficient phase information angle(I′) and W Find the modulus and get the second watermark carrier image I aW .
[0096] Combine [LL′3, LH′3, HL′3, HH′3] with the other two layers of decomposition coefficients to perform p1=p2=0.000000004300089657 order fractional wavelet inverse transform to obtain the watermarked reconstructed complex coefficient I W , where the real and imaginary parts constitute the reconstruction coefficient I W Phase information.
[0097] Retain the phase information corresponding to each pixel of the watermark carrier image:
[0098] A=imag(I W (x,y)) / real(IW (x,y)
[0099] Among them, (x, y) is the coordinate position of the pixel point of the watermark carrier image.
[0100] Step 6: Place the second watermark carrier image I aW Difference is calculated from the original image I, the mode ΔW of the watermark loaded into the second watermark carrier image is extracted, the amplitude of ΔW is limited, and the limited watermark ΔW′ is obtained. ΔW′ is re-embedded into the original image I to obtain the second watermark carrier image I △W .
[0101] Will I aW Subtract from the original image I,
[0102] ΔW=I aW -I
[0103] Extract the modality ΔW of the embedded carrier image watermark. If a certain pixel point is I aW When (x, y)>I(x, y), ΔW is “+”, indicating that the watermark of the pixel is superimposed on the pixel of the source carrier image; if a pixel is I aW When (x, y) < I(x, y), ΔW is "-", indicating that the watermark at the pixel point offsets the pixel value of the source carrier image. In this embodiment, the range of ΔW of the watermark embedded in the carrier image is obtained to be between [-10, +10], and ΔW is further limited to the range of [-1, 1] to obtain ΔW', then ΔW' < ΔW.
[0104] The limited watermark ΔW′ is embedded into the original image I again to obtain the final watermarked image I △W .
[0105] The present invention also provides an image watermark extraction method, comprising the following steps:
[0106] (1) According to the pixel point (x, y) position in the final watermark carrier image I △W Add phase information to get the complex coefficient I′ ΔW ;
[0107] I′ ΔW (x,y)=I ΔW (x,y).*(cos(angle(I ΔW (x,y)))+j*sin(angle(I ΔW (x,y))))
[0108] right I′ ΔW Perform K-layer p-order fractional wavelet transform to obtain 3K+1 subbands of fractional wavelet transform;
[0109] (3) For the K-th layer subband coefficient {HH Δ3 ,HL Δ3 ,LH Δ3 ,HH Δ3} Perform a two-dimensional fractional Fourier transform of order q1=q2=0.000005105 to obtain the complex coefficients {FLL Δ3 , FHL Δ,3 , FLH Δ3 , FHH Δ3},
[0110]
[0111]
[0112] (4) The complex coefficients {FLL Δ3 , FHL Δ3 , FLH Δ3 , FHH Δ3} and {FLL3, FLH3, FHL3, FHH3} are subtracted and divided by the scaling factors a1, a2, a3, a4 to extract {FLL Δ3 , FHL Δ3 , FLH Δ3 , FHH Δ3}Real part FW1′ and imaginary part FW2′.
[0113] FW1′=real(FW1′)+imag(FW1′)*j;
[0114] FW′2=real(FW2′)+imag(FW2′)*j
[0115]
[0116] imag(FW1′)=(real(FLH3′)-real(FLH3)+(imag(FLH3′)-imag(FLH3))*j) / (-0.8*j)
[0117]
[0118] imag(FW′2)=(real(FHH′3)-real(FLH3)+(imag(FHH′3)-imag(FHH3))*j) / (-0.8*j)
[0119] (5) Perform two-dimensional fractional Fourier inverse transform of order q1=q2=0.000005105 on FW1′ and FW′2 respectively.
[0120]
[0121]
[0122] Modulo the transformed complex coefficients WW1 and WW2, and get the watermark information W T1 =abs(WW1),W T2 = abs(WW2);
[0123] (6) According to the hybrid optical bistable model, the initial position is transformed to the watermark information W T1 and W T2 Perform inverse scrambling and extract the watermark W TW1 and W TW2 .
[0124] In the present invention, the watermark W TW1 and W TW2 The value is bit-anded to get the third watermark information W TW3 ,This step mainly deals with shear attacks.
[0125] Peak signal-to-noise ratio (PSNR) is used to objectively judge the difference between the source carrier image and the image embedded with the watermark.
[0126]
[0127] Among them, I(i,j), I w (i,j) represents the original and watermarked images, (i,j) represents the coordinates of the original carrier image and the watermarked image, and N1 and N2 represent the rows and columns of the image. The higher the PSNR value, the better the watermark is imperceptible and the better the transparency. When the PSNR value is greater than 30dB, the human eye cannot perceive the difference between the original image and the watermarked image.
[0128] Normalized cross correlation (NC) is used to judge the similarity between the extracted watermark and the original watermark.
[0129]
[0130] Among them, W0(i,j) is the original watermark coefficient value, W1(i,j) is the extracted watermark coefficient value, and M1 and M2 represent the rows and columns of the image. Represents the mean of the original watermark and the extracted watermark image respectively. The larger the NC value, the better the robustness of the watermark algorithm. The value of NC is between 0 and 1. The larger the NC value, the higher the similarity between the original watermark and the extracted watermark.
[0131] The present invention also provides an image watermark embedding device, comprising:
[0132] The subband acquisition module is used to perform K-layer p-order fractional wavelet transform on the original image I to obtain 3K+1 subbands;
[0133] A fractional Fourier transform module is used to perform an s-order fractional Fourier transform on the K-th layer subband of the transformed image to obtain complex coefficients FLL, FHL, FLH and FHH after subband transformation;
[0134] A watermark image preprocessing module is used to preprocess the watermark image to obtain a complex coefficient FW;
[0135] A complex coefficient embedding module is used for embedding the real part and the imaginary part of the complex coefficient FW into FLL, FHL, FLH and FHH to obtain watermarked coefficients FLL′, FHL′, FLH′ and FHH′;
[0136] A fractional Fourier inverse transform module is used to perform an s-order fractional Fourier inverse transform on FLL′, FHL′, FLH′ and FHH′ to obtain a first watermark carrier image;
[0137] The image reconstruction module is used to perform K-layer p-order fractional wavelet inverse transform on the 4 subband coefficients of the Kth layer of the first watermark carrier image and the subband coefficients of the remaining layers to obtain the reconstructed image complex coefficients I W , retaining the complex coefficient phase information angle(I W ) and I W Find the modulus and get the second watermark carrier image I αW ;
[0138] The limiting module is used to convert the second watermark carrier image I αW The difference between the original image I and the watermark mode ΔW loaded into the second watermark carrier image is extracted, the amplitude of ΔW is limited to obtain the limited watermark ΔW′, and ΔW′ is re-embedded into the original image I to obtain the final watermark carrier image I △W .
[0139] The present invention also provides an image watermark extraction device, comprising:
[0140] The first complex coefficient acquisition module is used to obtain the final watermark carrier image I △W Add phase information to get the complex coefficient I′ ΔW ;
[0141] Fractional wavelet transform module, used for complex coefficients I′ ΔW Perform K-layer p-order fractional wavelet transform to obtain 3K+1 subbands of fractional wavelet transform;
[0142] The second complex coefficient acquisition module is used to obtain the K-th layer subband coefficient {HH Δ3 ,HLΔ3 ,LH Δ3 ,HH Δ3} Perform a q-order two-dimensional fractional Fourier transform to obtain the complex coefficients {FLL Δ3 , FHL Δ3 , FLH Δ3 , FHH Δ3};
[0143] Difference operation module, used for complex coefficients {FLL Δ3 ,FHL Δ3 ,FLH Δ3 ,FHH Δ3} and {FLL3, FLH3, FHL3, FHH3} to extract {FLL Δ3 , FHL Δ3 , FLH Δ3 , FHH Δ3}real part FW1′ and imaginary part FW2′;
[0144] The watermark image acquisition module is used to perform a q-order fractional Fourier inverse transform on the real part FW1′ and the imaginary part FW2′, and perform a modulus operation on the transformed complex coefficients to extract the watermark image;
[0145] The watermark extraction module is used to transform the initial position according to the hybrid optical bistable model, perform inverse scrambling on the watermark information, and extract the watermark.
[0146] The present invention also provides a computer device, comprising a memory and a processor; the memory stores a computer program, and the processor is used to run the computer program in the memory to execute the steps of the image watermark embedding method and the steps of the image watermark extraction method.
[0147] The present invention also provides a computer-readable storage medium storing a computer program, wherein the computer program is suitable for being loaded by a processor to execute the steps of the image watermark embedding method and the steps of the image watermark extraction method.
[0148] Example 2
[0149] Watermark embedding and extraction simulation experiment
[0150] Figure 2 The standard image of size 512×512 shown as "Vase.tif" is used as the carrier image, and the following Figure 3The two images of size 64×64 are used as digital watermark and scrambled watermark images for experimental verification to prove the effectiveness of the present invention. In the experiment, the fractional wavelet transform order p1=p2=0.000000004992489657 and the Fourier transform order q1=q2=0.000005105 are used. The infinite watermark is obtained by the difference between the standard image and the watermark image (see Figure 4 ), when embedding an unlimited watermark (see Figure 5 ) and limited amplitude (see Figure 6 ) The peak signal-to-noise ratio (PSNR) values before and after watermarking are 31.2552dB (see Figure 7 ) and 51.0316dB (see Figure 8 ), indicating that the watermark is significantly less visible after being significantly limited and then embedded in the carrier image. (See Fig. 9 The extracted watermark normalized cross-correlation coefficient NC value W TW1 , W TW2 and W TW3 They are 0.9725, 0.9723, and 0.9717 respectively. This indicates that by compensating the phase of the carrier image ( Fig.10 ), even when the embedded watermark amplitude is greatly reduced, the impact on the integrity of the extracted watermark information is very small.
[0151] The present invention also carries out security analysis of the digital watermark extraction method.
[0152] In the watermark extraction step, the minimum change in the watermark key that can cause the extraction error corresponds to the security of the method. The smaller the minimum change, the higher the security of the method. The variables involved in the security performance of this method are: (1) the image initial values x0 and x1 are scrambled by the hybrid optical bistable model; (2) when the carrier image is subjected to fractional wavelet transform, the change in the key order p causes the watermark extraction error (such as Fig.11 shown).
[0153] (1) Change the initial values x0 and x1 of the scrambled image in the hybrid optical bistable model.
[0154] key1=1.367738; key2=0.5085987
[0155] (2) Changes in the key order p
[0156] p1=p2=0.0000000054874331; the extracted key is p1=p2=0.0000000054874332,△p= + 10 -16 , Fig.11The extracted watermark image is a garbled image, indicating that at low p values, the watermark extraction process is extremely sensitive to changes in p values. The smaller the p value, the larger the key space range.
[0157] (3) In the extraction stage, the watermark fractional Fourier domain complex coefficients are different from the embedded subband, resulting in erroneous watermarks.
[0158] The present invention also analyzes the robustness of the digital watermark embedding method. Figure 12-21 ,
[0159] (1) Shear attack
[0160] To verify the performance of the watermarking method under attacks at different shearing ratios, the carrier image was sheared by 1 / 4 and 1 / 2 respectively.
[0161] (2) Noise attack
[0162] (3) Filter Attack
[0163] (4) Histogram Equalization Attack
[0164] (5) Compression attack imwrite(XW1,'atkJPG.jpeg','jpeg','Quality',10)
[0165] (6) Image brightening XW1 = Imadjust(XW1,[0.10.8],
[01] )
[0166] (7) Image darkening XW1 = imadjust(XW1,[],[0.1,0.8])
[0167] (8) Increase the image contrast XW1 = imadjust(XW1, [0.1, 0.9], [])
[0168] (9) Reduce contrast XW1 = imadjust(XW1,[],[0.1,0.9])
[0169] (10) Spin attack
[0170] Fig.12 It shows that under large-area shearing attacks, by adding phase information to the carrier image, the loaded double watermark (NC1, NC2) information is extracted and (NC) is obtained through bit operation, which improves the watermark extraction effect. Figure 13-Figure 21 It shows that under various strong attacks that change the pixel values of the carrier image, by adding phase information to the carrier image, the loaded double watermark (NC1, NC2) information is extracted, and the watermark information (NC) after the two watermarks are bit-operated has a good overall effect.
[0171] The generalized transform domain digital watermarking method based on phase characteristics of the present invention has the characteristics of simple algorithm, fast and effective, high security, strong robustness and low false alarm rate. By using the phase information retained by the fractional order wavelet inverse transform after the image is embedded with the watermark, the watermark can be well extracted even when the amplitude information in the carrier image is small, thus increasing the capacity of the watermark information and having certain application value in digital copyright protection.
[0172] The fractional spectrum geometric features of the invented fractional Fourier transform reflect all the characteristics of the signal gradually changing from the time domain to the frequency domain as the transform order p1 changes. As the value of the order p2 (key) of FRWT changes, the energy concentration and coefficient sparsity of the image transform coefficients change. In addition, both transformations are complex transformations, and the transformation results provide coefficient phase information. The phase spectrum of the image well preserves the edge and overall structure information of the image. Therefore, variable watermark amplitude information can be embedded in the host image, and the reconstructed coefficient phase information can be fully utilized when extracting the watermark to improve the security and robustness of the watermark.
[0173] The above embodiments are only preferred specific implementation methods of the present invention, and the protection scope of the present invention is not limited thereto. Any simple changes or equivalent replacements of the technical solutions that can be obviously obtained by any technician familiar with the field within the technical scope disclosed in the present invention belong to the protection scope of the present invention.
Claims
1. A method for embedding an image watermark, characterized in that: The steps include: Perform K-layer p-order fractional wavelet transform on the original image I to obtain 3K+1 subbands; where K=3, p=p1=p2=0.000000004992489657; Perform an s-order fractional Fourier transform on the K-th layer subband of the transformed image to obtain complex coefficients FLL, FHL, FLH and FHH after subband transformation; wherein the complex coefficients FLL, FHL, FLH and FHH are the frequency domain coefficients FLL3, FHL3, FLH3 and FHH3 obtained by performing an s-order fractional Fourier transform on the four subband complex coefficients LL3, HL3, LH3 and HH3 of the third layer after wavelet transformation, s=q1=q2=0.000005105; Remove the correlation between the pixels of the watermark image, and perform a two-dimensional fractional Fourier transform on the removed pixels to obtain the complex coefficient FW; specifically: use a hybrid optical bistable model to make the watermark image W generate a chaotic sequence, transform the watermark pixel point (x, y) to a new coordinate point (x′, y′), and obtain scrambled watermarks W′ and W″; perform a two-dimensional fractional Fourier transform of order q1=q2=0.000005105 on the scrambled watermarks W′ and W″, respectively, and convert the watermark coefficients into fractional frequency domain coefficients FW1 and FW2, where FW1 and FW2 are the real part and imaginary part of the complex coefficient FW respectively; Embed the real and imaginary parts of the complex coefficient FW into FLL, FHL, FLH and FHH to obtain watermarked coefficients FLL′, FHL′, FLH′ and FHH′; Performing an s-order fractional Fourier inverse transform on FLL′, FHL′, FLH′ and FHH′ to obtain a first watermark carrier image, wherein the specific representation of FLL′, FHL′, FLH′ and FHH′ is FLL′3, FLH′3, FHL′3, FHH′3; Perform K-layer p-order fractional wavelet inverse transform on the 4 subband coefficients of the Kth layer and the remaining subband coefficients of the first watermark carrier image to obtain the reconstructed image complex coefficients I W , retaining the complex coefficient phase information angle(I W ) and I W Find the modulus and get the second watermark carrier image I αW ; Among them, the 4 sub-band coefficients are LL3′, HL3′, LH3′ and HH3′, as follows: The second watermark carrier image I αW The difference between the original image I and the watermark mode ΔW loaded into the second watermark carrier image is extracted, the amplitude of ΔW is limited to obtain the limited watermark ΔW′, and ΔW′ is re-embedded into the original image I to obtain the final watermark carrier image I △W .
2. A method for extracting an image watermark, characterized in that: The following steps are involved: In the watermark carrier image I △W Add phase information to get the complex coefficient I′ ΔW ; where I′ ΔW (x,y)=I ΔW (x,y).*(cos(angle(I ΔW (x,y)))+j*sin(angle(I ΔW (x,y)))), (x, y) is the coordinate value of the pixel point; For complex coefficient I′ ΔW Perform K-layer p-order fractional wavelet transform to obtain 3K+1 subbands of fractional wavelet transform; wherein, K=3, p=p1=p2=0.000000004992489657; For the K-th layer subband coefficient {HH Δ3 ,HL Δ3 ,LH Δ3 ,HH Δ3 } Perform a q-order two-dimensional fractional Fourier transform to obtain the complex coefficients {FLL Δ3 , FHL Δ3 , FLH Δ3 , FHH Δ3 }; where q = q1 = q2 = 0.000005105; For complex coefficients {FLL Δ3 ,FHL Δ3 ,FLH Δ3 ,FHH Δ3 } and {FLL3, FLH3, FHL3, FHH3} to extract {FLL Δ3 , FHL Δ3 , FLH Δ3 , FHH Δ3 }real part FW1′ and imaginary part FW2′; Perform q-order fractional Fourier inverse transform on the real part FW1′ and the imaginary part FW2′, and perform modulus operation on the transformed complex coefficients to extract the watermark image; The watermark information is extracted by transforming the initial position according to the hybrid optical bistable model and performing inverse scrambling on the watermark information.
3. An image watermark embedding device, characterized in that: include: The subband acquisition module is used to perform K-layer p-order fractional wavelet transform on the original image I to obtain 3K+1 subbands; wherein, K=3, p=p1=p2=0.000000004992489657; The fractional Fourier transform module is used to perform an s-order fractional Fourier transform on the K-th layer subband of the transformed image to obtain complex coefficients FLL, FHL, FLH and FHH after subband transformation; wherein the complex coefficients FLL, FHL, FLH and FHH are the frequency domain coefficients FLL3, FHL3, FLH3 and FHH3 obtained by performing an s-order fractional Fourier transform on the four subband complex coefficients LL3, HL3, LH3 and HH3 of the third layer after wavelet transformation, and s=q1=q2=0.000005105; The watermark image preprocessing module is used to remove the correlation between the pixels of the watermark image and perform a two-dimensional fractional Fourier transform on the removed pixels to obtain the complex coefficient FW; specifically: a hybrid optical bistable model is used to generate a chaotic sequence in the watermark image W, and the watermark pixel point (x, y) is transformed to a new coordinate point (x′, y′) to obtain scrambled watermarks W′ and W″; a two-dimensional fractional Fourier transform of order q1=q2=0.000005105 is performed on the scrambled watermarks W′ and W″, respectively, to convert the watermark coefficients into fractional frequency domain coefficients FW1 and FW2, where FW1 and FW2 are the real part and imaginary part of the complex coefficient FW respectively; A complex coefficient embedding module is used for embedding the real part and the imaginary part of the complex coefficient FW into FLL, FHL, FLH and FHH to obtain watermarked coefficients FLL′, FHL′, FLH′ and FHH′; A fractional Fourier inverse transform module is used to perform an s-order fractional Fourier inverse transform on FLL′, FHL′, FLH′ and FHH′ to obtain a first watermark carrier image, wherein the specific representation of FLL′, FHL′, FLH′ and FHH′ is FLL′3, FLH′3, FHL′3, FHH′3; The image reconstruction module is used to perform K-layer p-order fractional wavelet inverse transformation on the four subband coefficients of the Kth layer and the subband coefficients of the remaining layers of the first watermark carrier image to obtain the reconstructed image complex coefficients I W , retaining the complex coefficient phase information angle(I W ) and I W Find the modulus and get the second watermark carrier image I αW ; Among them, the 4 sub-band coefficients are LL3′, HL3′, LH3′ and HH3′, as follows: The limiting module is used to convert the second watermark carrier image I αW The difference between the original image I and the watermark mode ΔW loaded into the second watermark carrier image is extracted, the amplitude of ΔW is limited to obtain the limited watermark ΔW′, and ΔW′ is re-embedded into the original image I to obtain the final watermark carrier image I △W .
4. An image watermark extraction device, characterized in that: include: The first complex coefficient acquisition module is used to obtain the watermark carrier image I △W Add phase information to get the complex coefficient I′ ΔW ; where I′ ΔW (x,y)=I ΔW (x,y).*(cos(angle(I ΔW (x,y)))+j*sin(angle(I ΔW (x,y)))), (x, y) is the coordinate value of the pixel point; Fractional wavelet transform module, used for complex coefficients I′ ΔW Perform K-layer p-order fractional wavelet transform to obtain 3K+1 subbands of fractional wavelet transform; wherein, K=3, p=p1=p2=0.000000004992489657; The second complex coefficient acquisition module is used to obtain the K-th layer subband coefficient {HH Δ3 ,HL Δ3 ,LH Δ3 ,HH Δ3 } Perform a q-order two-dimensional fractional Fourier transform to obtain the complex coefficients {FLL Δ3 , FHL Δ3 , FLH Δ3 , FHH Δ3 }; where q = q1 = q2 = 0.000005105; Difference operation module, used for complex coefficients {FLL Δ3 ,FHL Δ3 ,FLH Δ3 ,FHH Δ3 } and {FLL3, FLH3, FHL3, FHH3} to extract {FLL Δ3 , FHL Δ3 , FLH Δ3 , FHH Δ3 }real part FW1′ and imaginary part FW2′; The watermark image acquisition module is used to perform a q-order fractional Fourier inverse transform on the real part FW1′ and the imaginary part FW2′, and perform a modulus operation on the transformed complex coefficients to extract the watermark image; The watermark extraction module is used to transform the initial position according to the hybrid optical bistable model, perform inverse scrambling on the watermark information, and extract the watermark.
5. A computer device, characterized in that: It comprises a memory and a processor; the memory stores a computer program, and the processor is used to run the computer program in the memory to execute the steps of the image watermark embedding method according to claim 1 or the steps of the image watermark extraction method according to claim 2.
6. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and the computer program is suitable for being loaded by a processor to execute the steps of the image watermark embedding method according to claim 1 or the steps of the image watermark extraction method according to claim 2.
Citation Information
Patent Citations
Zero-watermark embedding and extracting methods and zero-watermark embedding and extracting devices based on DWT and principal component analysis (PCA) compression
CN106023056A
Remote-sensing image digital watermark embedding and extracting method based on quaternary wavelet
CN107292806A
Robust screen watermark shooting method based on wavelet domain cascade network and reverse recovery
CN117455749A
Enhanced cyber security system with digital watermarking using transformation approach
IN202241052810A
Electronic watermark embedding method, and encoder and decoder capable of utilizing the method
JP2004221715A
