A spatial domain color digital image blind watermarking method integrating LU decomposition

Abstract

The present invention combines the advantages of fast running speed of spatial domain digital watermarking algorithm and high robustness of frequency domain digital watermarking algorithm, and discloses a spatial domain color digital image blind watermarking method integrating LU decomposition. According to the correlation of the elements of the lower triangular matrix obtained after LU decomposition, the present invention obtains the high correlation elements of the lower triangular matrix obtained after LU decomposition of the image block in the spatial domain, and uses these elements to complete the embedding and blind extraction of the digital watermark in the spatial domain without the need for real LU decomposition. The invention can embed the color image digital watermark into the color host image, and not only has good watermark concealment and strong robustness, but also has good real-time performance, which solves the problem of slow running speed of large-capacity color image digital watermark, and is suitable for occasions of fast and efficient digital media copyright protection.

Description

Technical Field

[0001] The invention belongs to the technical field of information security and relates to rapid copyright protection of large-capacity color digital images. Background Art

[0002] With the rapid development of network technology, the transmission volume of digital multimedia information such as color digital images has exploded, and at the same time, illegal acts such as piracy and infringement have emerged in an endless stream. Therefore, the relevant copyright protection issues have gradually attracted widespread attention from scholars at home and abroad. For this reason, on the one hand, the copyright protection logo is required to be beautiful, practical and have a high amount of information. On the other hand, with the widespread popularization of mobile terminal devices and the gradual upgrading of hardware configuration, people are increasingly pursuing faster and more efficient work efficiency, and the single frequency domain watermark algorithm with a long running time is difficult to meet people's application needs, so it is necessary to further improve its running speed. In addition, according to the different working domains of the host image, the digital watermark algorithm also includes a spatial domain digital watermark algorithm with a simple algorithm and fast operation, but it has the disadvantage of weak robustness. Therefore, how to fully combine the advantages of high real-time performance of the spatial domain digital watermark algorithm and strong robustness of the frequency domain digital watermark algorithm, and design a large-capacity digital watermark algorithm with high invisibility, strong robustness and high real-time performance has become one of the problems to be solved. Summary of the invention

[0003] The purpose of the present invention is to provide a spatial domain color digital image blind watermarking method integrating LU decomposition, which is characterized by being implemented through a specific watermark embedding process and an extraction process. The watermark embedding process is described as follows:

[0004] Step 1: First, set a pixel size to N × N Color watermark image W Divide into 3 layered watermark images in the order of red, green and blue primary colors W i ; Then, using the key-based Ka i The two-dimensional composite chaotic map is used to map each layered watermark image. W i Finally, each decimal pixel value in the scrambled layered watermark image is represented by an 8-bit binary number and connected in sequence to form a length of 8 N 2 The layered watermark bit sequence SW i ,in i =1, 2, 3 represent the red, green, and blue layers respectively;

[0005] Step 2: Set a pixel size to M × MOriginal color host image H Divide into 3 layered watermark images in the order of red, green and blue primary colors H i , and divide it into pixel sizes m × m Non-overlapping image blocks; according to the length of the layered watermark bit sequence 8 N 2 , using key-based Kb i The pseudo-random sequence generated by the Matlab system built-in function randperm(.) is used in the three layered watermark images. H i All image blocks to be embedded with watermarks are randomly selected from i =1, 2, 3 represent the red, green, and blue layers respectively;

[0006] Step 3: Select an image block to be embedded with a watermark A , using formula (1) to directly calculate the lower triangular matrix obtained after LU decomposition in the spatial domain L In the k The element in row and column 1 L k,1 ;

[0007] L k,1 = A k,1 / A 1,1 (1)

[0008] in, A k,1 yes A In the k The pixel value of row 1 column, k ∈{ p , q}, 1≤ p , q ≤ m ,and p≠q , m is the pixel size of the image block;

[0009] Step 4: Sequence the watermark bits from the layer SW i Take out a bit of watermark information to be embedded w , according to the embedded watermark information and formula (2), the lower triangular matrix L Change the values ​​of the corresponding positions of the elements in the first column to obtain a new lower triangular matrix L * ;

[0010] (2)

[0011] in, yes L * In the k The element in row and column 1, L k,1 yes L In the k The element in row and column 1, k ∈{ p , q}, 1≤ p , q ≤ m ,and p≠q , m is the pixel size of the image block, T i It is i The quantization step size of the image channels, i =1, 2, 3 represent the red, green, and blue layers respectively;

[0012] Step 5: Using formula (3), the change of high correlation elements Δ k Distributed to the image blocks to be embedded with watermarks A The watermarked image block is obtained on the relevant pixels of A * ;

[0013] (3)

[0014] in, , yes L * In the k The element in row and column 1, L k,1 yes L In the k The element in row and column 1, yes A * In the k Line j The pixel value of the column, A k,j yes A In the k Line j The pixel value of the column, k ∈{ p , q}, 1≤ p , q ≤ m ,and p≠q , 1≤ j ≤m , m is the pixel size of the image block;

[0015] Step 6: Use watermarked image blocks A * Replace the host image H The image block without watermark at the corresponding position in A , completing the process of embedding one bit of watermark information into an image block;

[0016] Step 7: Repeat steps 3 to 6 of this process until all watermark information is embedded. Finally, reconstruct the three layered watermarked images. H i * Get color watermarked image H * ,in i =1, 2, 3 represent the red, green, and blue layers respectively;

[0017] The watermark extraction process is described as follows:

[0018] Step 1: Convert the color watermarked image H * Divide into 3 layers of watermarked images in the order of red, green and blue primary colors H i * , and divide it into pixel sizes m × m non-overlapping image patches, where i =1, 2, 3 represent the red, green, and blue layers respectively;

[0019] Step 2: Using the key-based Kb i The pseudo-random sequence generated by the built-in function randperm(.) of the Matlab system selects all the image blocks to be extracted watermarks, where i =1, 2, 3 represent the red, green, and blue layers respectively;

[0020] Step 3: Select an image block to extract the watermark A * , using formula (4) to directly calculate the lower triangular matrix obtained after LU decomposition in the spatial domain L * In the k The element in row and column 1 ;

[0021] (4)

[0022] in, yes A * In the k The pixel value of row 1 column, k ∈{ p , q}, 1≤ p , q ≤ m ,and p≠q , m is the pixel size of the image block;

[0023] Step 4: Using formula (5), extract the watermark from the image block A * Extract the watermark information contained in w * ;

[0024] (5)

[0025] in, yes L * In the k The element in row and column 1, k ∈{ p , q}, 1≤ p , q ≤ m ,and p≠q , m is the pixel size of the image block;

[0026] Step 5: Repeat steps 3 to 4 until all binary watermark bits are extracted, and then obtain the extracted layered binary watermark sequence SW i * , and then convert each 8-bit binary information into a decimal pixel value as a group, where i =1, 2, 3 represent the red, green, and blue layers respectively;

[0027] Step 6: Perform key-based calculation on the converted decimal pixel values ​​of each layer Ka i Inverse two-dimensional composite chaotic mapping and obtain layered watermark image W i * ,in i =1, 2, 3 represent the red, green, and blue layers respectively;

[0028] Step 7: Combine the obtained layered watermark images W i * Form the final extracted watermark imageW * ,in i =1, 2,3 represent the red, green and blue layers respectively.

[0029] This method uses the lower triangular matrix after LU decomposition L A fast spatial domain calculation method for medium and high correlation elements and the distribution law of their element changes in spatial domain pixels are proposed. Variable quantization step size is used in the spatial domain to complete the embedding and blind extraction of color digital watermarks. This method not only has good watermark invisibility, but also has strong watermark algorithm robustness and high algorithm real-time performance. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] Figure 1 (a) Figure 1 (b) are two original color host images.

[0031] Figure 2 Is an original color watermark image.

[0032] Figure 3 (a) Figure 3 (b) is to Figure 2 The watermarks shown are embedded into the host image in sequence Figure 1 (a) Figure 1 The structural similarity SSIM values ​​of the watermarked images obtained after (b) are 0.9672 and 0.9732, respectively, and their peak signal-to-noise ratio PSNR values ​​are 40.0112dB and 41.1987dB, respectively.

[0033] Figure 4 (a) Figure 4 (b) is from Figure 3 (a) Figure 3 The normalized correlation coefficient NC values ​​of the watermarks extracted in (b) are 1.0000 and 1.0000 respectively.

[0034] Figure 5 (a) Figure 5 (b) Figure 5 (c) Figure 5 (d) Figure 5 (e) Figure 5 (f) is to Figure 3 The watermark extracted from the watermarked image shown in (a) after being attacked with JPEG 2000 (3: 1), salt and pepper noise (1%), cropping (25%), scaling (400%), Gaussian low-pass filtering (3×3), and histogram equalization, has the normalized cross-correlation coefficient NC values ​​of 0.9971, 0.9918, 0.9642, 0.9681, 0.9705, and 0.9847, respectively.

[0035] Figure 6 (a) Figure 6 (b) Figure 6 (c) Figure 6 (d) Figure 6 (e) Figure 6 (f) is to Figure 3 The watermark extracted from the watermarked image shown in (b) after being attacked with JPEG 2000 (3: 1), salt and pepper noise (1%), cropping (25%), scaling (400%), Gaussian low-pass filtering (3×3), and histogram equalization, has the normalized cross-correlation coefficient NC values ​​of 0.9971, 0.9903, 0.9642, 0.9615, 0.9681, and 0.9937, respectively. DETAILED DESCRIPTION

[0036] The purpose of the present invention is to provide a spatial domain color digital image blind watermarking method integrating LU decomposition, which is characterized by being implemented through a specific watermark embedding process and an extraction process. The watermark embedding process is described as follows:

[0037] Step 1: First, take a color watermark image with a pixel size of 32×32 W Divide into 3 layered watermark images in the order of red, green and blue primary colors W i ; Then, using the key-based Ka i The two-dimensional composite chaotic map is used to map each layered watermark image. W i Finally, each decimal pixel value in the scrambled layered watermark image is represented by an 8-bit binary number (for example, 155 can be converted into the binary number 10011011), and is sequentially connected to form a length of 8×32 2 =8192 layered watermark bit sequence SW i ,in i =1, 2, 3 represent the red, green, and blue layers respectively;

[0038] Step 2: Convert an original color host image with a pixel size of 512×512 H Divide into 3 layered watermark images in the order of red, green and blue primary colors H i , and divide it into non-overlapping image blocks with a pixel size of 4×4; according to the length of the layered watermark bit sequence 8×32 2 =8192, using key-based Kb i The pseudo-random sequence generated by the Matlab system built-in function randperm(.) is used in the three layered watermark images. Hi All image blocks to be embedded with watermarks are randomly selected from i =1, 2, 3 represent the red, green, and blue layers respectively;

[0039] Step 3: Select an image block to be embedded with a watermark A , using formula (1) to directly calculate the lower triangular matrix obtained after LU decomposition in the spatial domain L In the k The element in row and column 1 L k,1 ;

[0040] L k,1 = A k,1 / A 1,1 (1)

[0041] in, A k,1 yes A In the k The pixel value of row 1 column, k ∈{ p , q}, 1≤ p , q ≤ m ,and p≠q , m is the pixel size of the image block; here, let the selected image block , m =4, p =2, q =3, A 1,1 =209, A 2,1 =205, A 3,1 =204, then we get L 2,1 =0.9809, L 3,1 =0.9761;

[0042] Step 4: Sequence the watermark bits from the layer SW i Take out a bit of watermark information to be embedded w , according to the embedded watermark information and formula (2), the lower triangular matrix L Change the values ​​of the corresponding positions of the elements in the first column to obtain a new lower triangular matrix L * ;

[0043] (2)

[0044] in, yes L * In the k The element in row and column 1, L k,1 yes L In the k The element in row and column 1, k ∈{ p , q}, 1≤ p , q ≤ m ,and p≠q , m is the pixel size of the image block, T i It is i The quantization step size of the image channels, i =1, 2, 3 represent the red, green, and blue layers respectively; here, let i =1, w ='1', T 1 =0.0246, p =2, q =3, L 2,1 - L 3,1 =0.0048<2× T 1 =0.0493, then =1.0031, =0.9539;

[0045] Step 5: Using formula (3), the change of high correlation elements Δ k Distributed to the image blocks to be embedded with watermarks A The watermarked image block is obtained on the relevant pixels of A * ;

[0046] (3)

[0047] in, , yes L * In the k The element in row and column 1, L k,1 yes L In the k The element in row and column 1, yes A * In thek Line j The pixel value of the column, A k,j yes A In the k Line j The pixel value of the column, k ∈{ p , q}, 1≤ p , q ≤ m ,and p≠q , 1≤ j ≤ m , m is the pixel size of the image block; here, m =4, p =2, q =3,Δ 2 =0.0222, Δ 3 =-0.0222, get the watermarked image block ;

[0048] Step 6: Use watermarked image blocks A * Replace the host image H The image block without watermark embedded in the corresponding position A , completing the process of embedding one bit of watermark information into an image block;

[0049] Step 7: Repeat steps 3 to 6 of this process until all watermark information is embedded. Finally, reconstruct the three layered watermarked images. H i * Get color watermarked image H * ,in i =1, 2, 3 represent the red, green, and blue layers respectively;

[0050] The watermark extraction process is described as follows:

[0051] Step 1: Convert the color watermarked image H * Divide into 3 layers of watermarked images in the order of red, green and blue primary colors H i * , and divide it into pixel sizes m × m non-overlapping image patches, where i =1, 2, 3 represent the red, green, and blue layers respectively;

[0052] Step 2: Using the key-based Kb i The pseudo-random sequence generated by the built-in function randperm(.) of the Matlab system selects all the image blocks to be extracted watermarks, where i =1, 2, 3 represent the red, green, and blue layers respectively;

[0053] Step 3: Select an image block to extract the watermark A * , using formula (4) to directly calculate the lower triangular matrix obtained after LU decomposition in the spatial domain L * In the k The element in row and column 1 ;

[0054] (4)

[0055] in, yes A * In the k The pixel value of row 1 column, k ∈{ p , q}, 1≤ p , q ≤ m ,and p≠q , m is the pixel size of the image block; here, let the selected image block , m =4, p =2, q =3, =209, =210, =199, then =1.0048, =0.9522;

[0056] Step 4: Using formula (5), extract the watermark from the image block A * Extract the watermark information contained in w * ;

[0057] (5)

[0058] in, yes L * In the k The element in row and column 1, k ∈{ p , q}, 1≤p , q ≤ m ,and p≠q , m is the pixel size of the image block; here, p =2, q =3, =1.0048> =0.9522, we get w * ='1';

[0059] Step 5: Repeat steps 3 to 4 until all binary watermark bits are extracted, and then obtain the extracted layered binary watermark sequence SW i * , and then convert each 8-bit binary information into a decimal pixel value as a group, where i =1, 2, 3 represent the red, green, and blue layers respectively;

[0060] Step 6: Perform key-based calculation on the converted decimal pixel values ​​of each layer Ka i Inverse two-dimensional composite chaotic mapping and obtain layered watermark image W i * ,in i =1, 2, 3 represent the red, green, and blue layers respectively;

[0061] Step 7: Combine the obtained layered watermark images W i * Form the final extracted watermark image W * ,in i =1, 2,3 represent the red, green and blue layers respectively.

[0062] This method uses the lower triangular matrix after LU decomposition L A fast spatial domain calculation method for medium and high correlation elements and the distribution law of their element changes in spatial domain pixels are proposed. Variable quantization step size is used in the spatial domain to complete the embedding and blind extraction of color digital watermarks. This method not only has good watermark invisibility, but also has strong watermark algorithm robustness and high algorithm real-time performance.

[0063] Verification of the effectiveness of the present invention

[0064] In order to prove the effectiveness of the present invention, Figure 1 (a) Figure 1 The two 24-bit standard images with a pixel size of 512×512 shown in (b) are used as host images and are respectively Figure 2 A 24-bit color image with a pixel size of 32×32 is shown as a digital watermark for verification.

[0065] Figure 3 (a) Figure 3 (b) is to Figure 2 The watermarks shown are embedded into the host image in sequence Figure 1 (a) Figure 1 (b) The structural similarity SSIM values ​​of the watermarked images obtained after are 0.9672 and 0.9732, respectively, and the peak signal-to-noise ratio PSNR values ​​are 40.0112dB and 41.1987dB, respectively; Figure 4 (a) Figure 4 (b) is from Figure 3 (a) Figure 3 The normalized correlation coefficient NC values ​​of the watermarks extracted in (b) are 1.0000 and 1.0000 respectively; Figure 5 (a) Figure 5 (b) Figure 5 (c) Figure 5 (d) Figure 5 (e) Figure 5 (f) is to Figure 3 The watermarks extracted from the watermarked image shown in (a) are subjected to JPEG 2000 (3: 1), salt and pepper noise (1%), cropping (25%), scaling (400%), Gaussian low-pass filtering (3×3), and histogram equalization attacks. The normalized cross-correlation coefficient NC values ​​are 0.9971, 0.9918, 0.9642, 0.9681, 0.9705, and 0.9847 respectively. Figure 6 (a) Figure 6 (b) Figure 6 (c) Figure 6 (d) Figure 6 (e) Figure 6 (f) is to Figure 3 The watermark extracted from the watermarked image shown in (b) after being attacked with JPEG 2000 (3: 1), salt and pepper noise (1%), cropping (25%), scaling (400%), Gaussian low-pass filtering (3×3), and histogram equalization, has the normalized cross-correlation coefficient NC values ​​of 0.9971, 0.9903, 0.9642, 0.9615, 0.9681, and 0.9937, respectively.

[0066] The algorithm has been run nearly 10,000 times on the platform 2.00GHZ CPU, 16.00GB RAM, Win 10, MATLAB (R2017a). The average embedding time of the digital watermark is 0.08883 seconds, the average extraction time is 0.05588 seconds, and the total time is 0.14471 seconds.

[0067] In summary, the embedded color image digital watermark has good invisibility, which meets the invisibility requirement of the watermark algorithm; at the same time, the color image digital watermarks extracted from various attacked images have good identifiability and high NC values, indicating that the method has strong robustness; in addition, the average total running time of the algorithm is less than 1 second, which meets the needs of rapid copyright protection of multimedia big data.

Claims

1. A spatial domain color digital image blind watermarking method integrating LU decomposition, Features This is achieved through a specific watermark embedding process and extraction process. The watermark embedding process is described as follows: Step 1: First, set a pixel size to N × N Color watermark image W Divide into 3 layered watermark images in the order of red, green and blue primary colors W i ; Then, using the key-based Ka i The two-dimensional composite chaotic map is used to map each layered watermark image. W i Finally, each decimal pixel value in the scrambled layered watermark image is represented by an 8-bit binary number and connected in sequence to form a length of 8 N 2 The layered watermark bit sequence SW i ,in i =1, 2, 3 represent the red, green, and blue layers respectively; Step 2: Set a pixel size to M × M Original color host image H Divide into 3 layered watermark images in the order of red, green and blue primary colors H i , and divide it into pixel sizes m × m Non-overlapping image blocks; According to the layered watermark bit sequence length 8 N 2 , using key-based Kb i The pseudo-random sequence generated by the Matlab system built-in function randperm(.) is used in the three layered watermark images. H i All image blocks to be embedded with watermarks are randomly selected from i =1, 2, 3 represent the red, green, and blue layers respectively; Step 3: Select an image block to be embedded with a watermark A , using formula (1) to directly calculate the lower triangular matrix obtained after LU decomposition in the spatial domain L In the k The element in row and column 1 L k,1 ; L k,1 = A k,1 / A 1,1 (1) in, A k,1 yes A In the k The pixel value of row 1 column, k ∈{ p , q }, 1≤ p , q ≤ m ,and p≠q , m is the pixel size of the image block; Step 4: Sequence the watermark bits from the layer SW i Take out a bit of watermark information to be embedded w , according to the embedded watermark information and formula (2), the lower triangular matrix L Change the values ​​of the corresponding positions of the elements in the first column to obtain a new lower triangular matrix L * ; (2) in, yes L * In the k The element in row and column 1, L k,1 yes L In the k The element in row and column 1, k ∈{ p , q }, 1≤ p , q ≤ m ,and p≠q , m is the pixel size of the image block, T i It is i The quantization step size of the image channels, i =1, 2, 3 represent the red, green, and blue layers respectively; Step 5: Using formula (3), the change of high correlation elements Δ k Distributed to the image blocks to be embedded with watermarks A The watermarked image block is obtained on the relevant pixels of A * ; (3) in, , yes L * In the k The element in row and column 1, L k,1 yes L In the k The element in row and column 1, yes A * In the k Line j The pixel value of the column, A k,j yes A In the k Line j The pixel value of the column, k ∈{ p , q }, 1≤ p , q ≤ m ,and p≠q , 1≤ j ≤ m , m is the pixel size of the image block; Step 6: Use watermarked image blocks A * Replace the host image H The image block without watermark at the corresponding position in A , completing the process of embedding one bit of watermark information into an image block; Step 7: Repeat steps 3 to 6 of this process until all watermark information is embedded. Finally, reconstruct the three layered watermarked images. H i * Get color watermarked image H * ,in i =1, 2, 3 represent the red, green, and blue layers respectively; The watermark extraction process is described as follows: Step 1: Convert the color watermarked image H * Divide into 3 layers of watermarked images in the order of red, green and blue primary colors H i * , and divide it into pixel sizes m × m non-overlapping image patches, where i =1, 2, 3 represent the red, green, and blue layers respectively; Step 2: Using the key-based Kb i The pseudo-random sequence generated by the built-in function randperm(.) of the Matlab system selects all the image blocks to be extracted watermarks, where i =1, 2, 3 represent the red, green, and blue layers respectively; Step 3: Select an image block to extract the watermark A * , using formula (4) to directly calculate the lower triangular matrix obtained after LU decomposition in the spatial domain L * In the k The element in row and column 1 ; (4) in, yes A * In the k The pixel value of row 1 column, k ∈{ p , q }, 1≤ p , q ≤ m ,and p≠q , m is the pixel size of the image block; Step 4: Using formula (5), extract the watermark from the image block A * Extract the watermark information contained in w * ; (5) in, yes L * In the k The element in row and column 1, k ∈{ p , q }, 1≤ p , q ≤ m ,and p≠q , m is the pixel size of the image block; Step 5: Repeat steps 3 to 4 until all binary watermark bits are extracted, and then obtain the extracted layered binary watermark sequence SW i * , and then convert each 8-bit binary information into a decimal pixel value as a group, where i =1, 2, 3 represent the red, green, and blue layers respectively; Step 6: Perform key-based calculation on the converted decimal pixel values ​​of each layer Ka i Inverse two-dimensional composite chaotic mapping and obtain layered watermark image W i * ,in i =1, 2, 3 represent the red, green, and blue layers respectively; Step 7: Combine the obtained layered watermark images W i * Form the final extracted watermark image W * ,in i =1, 2, 3 represent the red, green and blue layers respectively.

Images