CN114666453A - Separable ciphertext domain reversible data hiding method based on bit plane segmentation - Google Patents

Abstract

The invention discloses a separable ciphertext domain reversible data hiding method based on bit plane segmentation, which firstly provides a block classification scrambling algorithm and a traversal matrix encryption algorithm to complete the encryption of an original image; then, encrypting the watermark image by using a Hilbert curve; in a ciphertext domain, an all-bit plane scrambling technology is adopted, bit plane rotation is firstly carried out, then the block is divided into continuous blocks and discontinuous blocks, and finally reversible data hiding is realized according to a pixel prediction method. Meanwhile, the invention realizes the key separation, has the hidden key to extract the hidden watermark, has the encryption key to recover the original image, and has the double keys to extract the watermark image and recover the original image. The invention realizes 100% reversibility based on pixel prediction, can not only extract watermark images without errors, but also recover original images by 100%, and has high safety, large capacity, high quality and good robustness.

Description

Separable ciphertext domain reversible data hiding method based on bit plane segmentation

Technical Field

The invention relates to a reversible data hiding method for a cryptograph domain, in particular to a separable reversible data hiding method for the cryptograph domain based on bit plane segmentation.

Background

The information hiding technology is distinguished from a plurality of encryption technologies by the functions of integrity authentication, copyright protection and the like, and reversible data hiding is an information hiding technology which can recover an original image without damage and extract information without errors. However, the traditional reversible data hiding has the potential safety hazard of original image exposure, so that ciphertext domain reversible data obtained by combining an encryption technology and the reversible data hiding becomes a hot point of academic attention.

The ciphertext domain Reversible information Hiding (RDHEI) hides information while protecting Image content from being leaked, so that a user can decrypt the Image content, extract the hidden information and restore an original Image without damage according to the type of a secret key owned by the user and the requirement of the user, and the confidentiality and the usability of the Image are guaranteed. The original RDHEI algorithm was to make Room After the image was encrypted, i.e. ciphertext reversible information hiding based on the vrae (trading from After encryption) framework and symmetric encryption.

As early as 2008, Puech et al, by dividing an encrypted image, hide additional information in each block, and finally complete information extraction and image restoration by using a local standard deviation of image pixels, have not been paid attention by researchers; zhang utilizes stream cipher to encrypt original image, then divides the encrypted image into blocks with same size, hides additional information by turning LSB of partial pixel in the blocks, the algorithm can utilize natural image pixel correlation to realize better performance; zhou et al propose to use two classes of support vector machine classifiers to distinguish image regions in the decoding stage, thereby extracting information and restoring images. However, the above-mentioned classical RDHEI method has a common application limitation that the information extraction and image restoration processes of the algorithm are inseparable.

In order to solve the problem, Ge and the like uses a stream cipher to scramble and encrypt an image block, then adopts a histogram translation method to realize information hiding, and finally extracts information at a receiver and recovers the image without damage according to a scrambling key and an encryption key; zhang et al propose a high-fidelity thumbnail protection encryption scheme, which not only accurately implements encryption and decryption, but also has thumbnails closer to the original plaintext image and lower noise; wang et al rearranges and encrypts all pixel blocks, hides information by adaptively predicting the MSB (Most Significant Bit, MSB), and the algorithm makes full use of the intra-block pixel correlation, improving the embedding capacity.

With the continuous improvement of data security requirements at home and abroad, ciphertext domain reversible information hiding algorithms based on symmetric encryption and RRBE (reproducing from Before encryption) framework and public key/homomorphic encryption are generally concerned by students. Puteaus et al propose that the information hiding of the ciphertext domain does not need to consider the problem of image visual quality after encryption and hiding, and therefore the highest bit plane of the image is replaced to embed the additional information, and therefore a higher effective load is obtained. Chen and the like generate a ciphertext image by using a method of combining stream cipher and position scrambling, and embed additional information in a reserved space by using a multi-MSB compression method, so that the method can resist various attacks and effectively improve the capacity. Ke et al have designed a LSB information hiding method based on key exchange, which realizes the direct extraction of information from the encryption domain without the need of a private key.

However, the VRAE-based framework has its own limitations, for example, some may have an error rate in data extraction or image recovery, and cannot be completely reversible; some are inseparable in the information extraction and image restoration processes.

Disclosure of Invention

The invention provides a separable ciphertext domain reversible data hiding method based on bit plane segmentation, which aims to solve the technical problems in the prior art.

The technical solution of the invention is as follows: a separable ciphertext domain reversible data hiding method based on bit plane segmentation is characterized by comprising the following steps in sequence:

step 1: carrying out block-by-block classified scrambling encryption and traversal matrix encryption on original images

Step 1.1 divides an original image of size M × N into p × q sub-blocks of size M × N that do not overlap, where p is M/M and q is N/N;

step 1.2, dividing the sub-blocks into a brightness block and a darkness block according to a formula (1), and marking the brightness block with 1 and the darkness block with 0 to generate a classification chart B;

in the formula (1), num represents the number of pixels with the gray value being equal to or greater than 127 in the sub-block, i belongs to (1,2,3 … p), and j belongs to (1,2,3 … q);

step 1.3, Huffman compression coding is carried out on the classification chart B, and the coded data is used as a secret key K₁Storing;

step 1.4, traversing the classification graph B to rearrange, placing all the brightness blocks in front of the darkness blocks, rearranging the brightness blocks according to the original sequence, and rearranging the darkness blocks in the reverse sequence to obtain the image after the block classification scrambling;

step 1.5 Generation of a random number sequence L of length m n by equation (2)₁In the formula 3.5699456<μ≤4，0<x<1, the random number sequence L₁Disorder and no repetition, and the initial values of μ and x in formula (2)₀As a key K₂Storing;

x_k+1＝μx_k(1-x_k) (2)

step 1.6 sequence of random numbers L₁Sorting, the number sequence after sorting is marked as L₂According to L₁And L₂Generating a number pair according to the one-to-one correspondence relationship of the number pair, and constructing a traversal matrix according to the correspondence relationship of the number pair;

step 1.7, performing matrix traversal encryption for multiple times by taking the image subblocks after block classification scrambling as a unit to obtain a ciphertext image, wherein the matrix traversal encryption is to rearrange pixels in the subblocks according to a constructed traversal matrix, and take the number of times of the matrix traversal encryption as a secret key K₃Storing;

step 2: watermark image encryption

Dividing the watermark image into non-overlapping 8 x 8 sub-blocks, traversing all the sub-blocks in a snake shape in a horizontal direction first, and traversing all pixel points in each sub-block in a non-crossing manner according to a Hilbert curve to generate one-dimensional data;

and step 3: hiding reversible data in a ciphertext domain, which comprises the following steps:

step 3.1: carrying out all-position plane scrambling on the ciphertext image, namely combining the highest bits of 8 different pixels into a new 1 x 8 matrix, combining the seventh bits of 8 different pixels into a new 1 x 8 matrix, repeating the above steps, finally combining the lowest bits of 8 different pixels into a new 1 x 8 matrix, and recombining the 8 new 1 x 8 matrices into an 8 x 8 matrix to obtain the ciphertext image subjected to all-position plane scrambling;

step 3.2, respectively forming the front 4 elements and the rear 4 elements of each row with the same position in the new bit plane of the ciphertext image after the all-bit plane scrambling into 2 multiplied by 2 subblocks;

step 3.3, classifying all the 2 × 2 sub-blocks generated by scanning into continuous blocks and discontinuous blocks, wherein the continuous blocks are the 2 × 2 sub-blocks of which all pixels in the block are 1 or 0, and the discontinuous blocks are the 2 × 2 sub-blocks of which the pixels in the block are 0 and 1; using 1 to represent continuous block and 0 to represent discontinuous block, generating block label matrix as key K₄Storing;

step 3.4, rearranging all the 2 x 2 sub-blocks according to the sequence of the preceding non-continuous block and the succeeding non-continuous block of the continuous block to obtain 8 new bit plane matrixes rearranged based on the 2 x 2 sub-blocks;

and 3.5, for the continuous block in the new bit plane matrix, setting the position of the pos in the block as a reserved position, embedding 1-bit data into the rest three positions respectively to obtain the continuous block after data embedding, wherein the pos is calculated according to a formula (3), MOD (DEG) is a modulus function, INT (DEG) is a downward integer function, and K is a K₅Is a random number as a key K₅Storing;

pos＝MOD(INT(K₅ ³),4) (3)

step 3.6, for the non-continuous block in the new bit plane matrix, the position of the pos in the block is a predicted position, the value of the predicted position is a predicted value, and the pos is calculated according to the formula (3);

step 3.7 determines the value of the predicted position, i.e. the predicted value p, according to equation (4),

when the predicted value is equal toWhen the predicted values are the same, the non-continuous block is an embeddable block, 1-bit data is embedded in the predicted position, otherwise, the non-continuous block is an un-embeddable block, and the non-continuous block with the embedded data is obtained; marking the embeddable blocks with 1 and the non-embeddable blocks with 0, generating a sequence L with the same length as the number of blocks, and using L as a secret key K₆Storing;

step 3.8, all the one-dimensional data are embedded into the sub-blocks, and the recording block is embedded into the tag end position and used as a secret key K₇Storing, namely recombining 8bit planes subjected to one-dimensional data embedding into an image to obtain a ciphertext image containing a watermark;

step 4 image recovery and data extraction

Using an encryption key K₁-K₄And extracting a decrypted image from the ciphertext image with the watermark, which comprises the following specific steps:

step 4.1 Using Key K₄Obtaining a block label matrix, rearranging blocks to obtain a new bit plane;

step 4.2, the new bit plane is rotated by 90 degrees in an inverse manner to obtain an original bit plane;

step 4.3, each row of 8-bit binary system of the original bit plane is converted into a decimal gray value, and all the gray values are combined into a ciphertext image;

step 4.4 Using Key K₂Constructing a traversal matrix;

step 4.5 Using Key K₃Performing inverse matrix decryption on the ciphertext image to obtain a decrypted image matrix;

step 4.6 Using Key K₁Rearranging the brightness blocks and the darkness blocks of the obtained classification image to obtain a decrypted image;

using an encryption key K₅-K₇Extracting a watermark image from a ciphertext image containing the watermark, which comprises the following specific steps:

step 4.7 Using Key K₅Determining a data embedding position;

step 4.8 Using Key K₇Reading the end position of the block embedded label;

step 4.9, for the continuous block, sequentially reading all values except the determined position to obtain data embedded into the continuous block;

step 4.10 Using Key K₆Dividing the non-continuous blocks into embeddable blocks and non-embeddable blocks;

step 4.11, extracting the value of the determined data position in the embeddable block to obtain the data in the discontinuous block;

step 4.12, decrypting the data based on the Hilbert curve to obtain a watermark image;

step 4.13 Using Key K₅Determining the data embedding position and the reserved position by using the secret key K₇Reading the end position of the block embedded label;

step 4.14, for the continuous block, reading the value of the reserved position, and replacing all the values in the current block with the values of the reserved position;

step 4.15, for the non-continuous blocks, dividing the non-continuous blocks into embeddable blocks and non-embeddable blocks;

step 4.16, for the embeddable block, reading values of other three positions except the embedded position, if the number of 1 in the three values is more, setting the embedded position value to be 1, otherwise, setting the embedded position value to be 0, and obtaining a recovered ciphertext image;

the method adopts a double secret key to extract a watermark image from a ciphertext image containing the watermark and obtain an original image, and comprises the following specific steps:

step 4.17 Using Key K₅Determining a data embedding position;

step 4.18 Using Key K₇Reading the end position of the block embedded label;

step 4.19, for the continuous block, sequentially reading all values except the determined position to obtain data embedded into the continuous block;

step 4.20 Using Key K₆Dividing the non-continuous blocks into embeddable blocks and non-embeddable blocks;

step 4.21, extracting the value of the determined data position in the embeddable block to obtain the data in the discontinuous block;

step 4.22, decrypting the data based on the Hilbert curve to obtain a watermark image;

step 4.23 with the Key K₅Determining the data embedding position,Reserving location, using key K₇Reading the end position of the block embedded label;

step 4.24, for the continuous block, reading the value of the reserved position, and replacing all the values in the current block with the values of the reserved position;

step 4.25, for the non-continuous blocks, dividing the non-continuous blocks into embeddable blocks and non-embeddable blocks;

step 4.26, for the embeddable block, reading values of other three positions except the embedded position, if the number of 1 in the three values is more, setting the embedded position value to be 1, otherwise, setting the embedded position value to be 0, and obtaining a recovered ciphertext image;

step 4.27 Using Key K₄Obtaining a block label matrix, rearranging blocks of the recovered ciphertext image to obtain a new bit plane;

step 4.28, the new bit plane is rotated by 90 degrees in an inverse manner to obtain an original bit plane;

step 4.29, each row of 8-bit binary system of the original bit plane is converted into a decimal gray value, and all the gray values are combined into a ciphertext image;

step 4.30 Using Key K₂Constructing a traversal matrix;

step 4.31 Using Key K₃Carrying out inverse matrix decryption on the ciphertext image to obtain a decrypted image matrix;

step 4.32 Using Key K₁And rearranging the brightness block and the darkness block of the obtained classification image to obtain an original image.

Firstly, providing a block classification scrambling algorithm and a traversal matrix encryption algorithm to finish the encryption of an original image; then, encrypting the watermark image by using a Hilbert curve; in a ciphertext domain, an all-bit plane scrambling technology is adopted, bit plane rotation is firstly carried out, then the block is divided into a continuous block and a discontinuous block, and finally reversible data hiding is realized according to a pixel prediction method. Meanwhile, the invention realizes the separability of the secret key, has the hidden secret key to extract the hidden watermark, has the encryption secret key to recover the original image, and has double secret keys to extract the watermark image and recover the original image. The invention realizes 100% reversibility based on pixel prediction, can not only extract watermark images without errors, but also recover original images by 100%, and has high safety, large capacity, high quality and good robustness.

Drawings

FIG. 1 is a flow chart of an embodiment of the present invention.

Fig. 2 is an original image of an embodiment of the present invention.

FIG. 3 is a block-based sorting scrambling diagram according to an embodiment of the present invention.

Fig. 4 is a schematic diagram of the encryption principle of the traversal matrix according to the embodiment of the present invention.

Fig. 5 is a ciphertext image of an embodiment of the present invention.

FIG. 6 is a histogram of an original image according to an embodiment of the present invention.

Fig. 7 is a histogram of a ciphertext image of an embodiment of the invention.

Fig. 8 is a watermark image of an embodiment of the present invention.

Fig. 9 is an encryption diagram of the Hilbert curve according to the embodiment of the present invention.

FIG. 10 is a schematic diagram of all-plane scrambling according to an embodiment of the present invention.

Fig. 11 is a bit-plane block diagram of an embodiment of the present invention.

FIG. 12 is a block classification diagram according to an embodiment of the present invention.

FIG. 13 is a block retention location diagram of an embodiment of the present invention

FIG. 14 is a diagram of non-contiguous blocks and embedding locations according to an embodiment of the present invention.

Fig. 15 is a decrypted image of an embodiment of the present invention.

Fig. 16 shows an original image restored after a salt and pepper noise attack and an extracted watermark image according to an embodiment of the present invention.

Fig. 17 shows the ciphertext image, the restored original image, and the extracted watermark image after the cut attack according to the embodiment of the present invention.

Detailed Description

The separable ciphertext domain reversible data hiding method based on bit plane segmentation is shown in fig. 1 and sequentially comprises the following steps:

step 1: carrying out block-by-block classified scrambling encryption and traversal matrix encryption on original images

Step 1.1 (a-f)6 standard grayscale raw images of fig. 2 with dimensions of 128 × 128: lena, Baboon, Boat, Pepper, barbarbara, airplan, divided into p × q subblocks of size M × N without overlap, p ═ M/M, q ═ N/N;

step 1.2, dividing the sub-blocks into a brightness block and a darkness block according to a formula (1), and marking the brightness block with 1 and the darkness block with 0 to generate a classification chart B;

in the formula (1), num represents the number of pixels with the gray value being equal to or greater than 127 in the sub-block, i belongs to (1,2,3 … p), and j belongs to (1,2,3 … q);

step 1.3, performing Huffman compression coding on the classification chart B, and taking the coded data as a secret key K₁Storing;

step 1.4, traversing the classification graph B to rearrange, placing all the brightness blocks in front of the darkness blocks, rearranging the brightness blocks according to the original sequence, and rearranging the darkness blocks in the reverse sequence to obtain the image after the block classification scrambling;

for example, the 16 × 16 original image shown in fig. 3a is divided into non-overlapping 4 × 4 sub-blocks, and the classification matrix and the scrambled matrix are respectively shown in formula B, B', so that the scrambled image is shown in fig. 3b, where white represents that the gray value of the pixel is greater than or equal to 127 (luminance block), and black represents that the gray value is less than 127 (darkness block);

step 1.5 as shown in fig. 4a and 4b, matrix traversal encryption is performed by using the scrambled image sub-blocks as a unit, and a random number sequence L with the length of m × n is generated by using a formula (2)₁In the formula 3.5699456<μ≤4，0<x<1, the random number sequence L₁It should be unordered and not repeated, so in the process of generating random numbers, if the same data is encountered, the random numbers are deleted and then generated again, and the initial values mu and x in the formula (2) are used₀As a key K₂Storing;

x_k+1＝μx_k(1-x_k) (2)

for example, for 4 × 4 sub-blocks, in K₂Generating a random sequence L of length 16 for the key₁Wherein each subsequent (i) represents its position:

L₁＝{0.4523(1),0.9537(2),0.1698(3),0.5429(4),0.9554(5),0.1639(6),0.5277(7),0.9595(8),0.1495(9),0.4894(10),0.9621(11),0.1405(12),0.4649(13),0.9578(14),0.1557(15),0.5062(16)}；

step 1.6 sequence of random numbers L₁Sorting, the number sequence after sorting is marked as L₂In each (i, j) after the number, i represents the position of the value in the original sequence, and j represents the position in the sorted sequence:

L₂＝{0.1405(12,1),0.1495(9,2),0.1557(15,3),0.1639(6,4),0.1698(3,5),0.4523(1,6),0.4649(13,7),0.4894(10,8),0.5062(16,9),0.5277(7,10),0.5429(4,11),0.9537(2,12),0.9554(5,13),0.9578(14,14),0.9595(8,15),0.9621(11,16)}；

according to L₁And L₂To generate pairs of (1,6) (2,12) (3,5) (4,11) (5,13) (6,4) (7,10) (8,15) (9,2) (10,8) (11,16) (12,1) (13,7) (14,14) (15,3) (16,9)

According to the corresponding relation of the data pairs, the constructed traversal matrix is as follows:

step 1.7, performing matrix traversal encryption for multiple times by taking the image subblocks after block classification scrambling as a unit to obtain a ciphertext image, wherein the matrix traversal encryption is to rearrange pixels in the subblocks according to a constructed traversal matrix, and a primary matrix traversal encryption matrix and a secondary matrix traversal encryption matrix are respectively shown in fig. 4c and 4d, using the traverse encryption times of the matrix as a secret key K₃Storing;

as shown in fig. 5, the obtained ciphertext image is similar to noise and has low correlation in terms of encryption effect, and original image detail information cannot be obtained from the ciphertext image; FIG. 6 is a histogram of an original image, FIG. 7 is a histogram of a ciphertext image, wherein the ciphertext image is uniformly distributed in terms of pixel spatial distribution and cannot be fitted to the pixel distribution of the original image;

table 1 shows the information entropy, NPCR, UACI of the original image and the ciphertext image, and the closer the information entropy value is to 8, the better the randomness of the image is, and the weaker the pixel correlation is.

TABLE 1

As can be seen from table 1, the information entropy of the original image is about 6.7, and the information entropy of the ciphertext image is about 7.9; the difference between the original image and the ciphertext image is measured by using NPCR and UACI, wherein the NPCR value is more than 99%, and the UACI value is less than 17%; in summary, the difference between the ciphertext image and the original image is large, and the pixel distribution randomness of the ciphertext image is stronger, which shows that the encryption used for the original image has strong security.

And 2, step: watermark image encryption

Dividing the gray watermark image with the size of 64 x 64 shown in fig. 8 into non-overlapping 8 x 8 sub-blocks, as shown in fig. 9a-d, traversing all sub-blocks in a snake shape in a horizontal direction first, and traversing all pixel points in each sub-block in a non-crossing manner according to a Hilbert curve to generate one-dimensional data;

and step 3: the method for hiding the reversible data in the ciphertext domain comprises the following specific steps:

step 3.1, as shown in fig. 10a-c, performing all-bit plane scrambling on the ciphertext image, namely combining the highest bits of 8 different pixels into a new 1 × 8 matrix, combining the seventh bits of 8 different pixels into a new 1 × 8 matrix, and so on, finally combining the lowest bits of 8 different pixels into a new 1 × 8 matrix, and recombining the 8 new 1 × 8 matrices into an 8 × 8 matrix to obtain the ciphertext image after all-bit plane scrambling;

step 3.2, as shown in fig. 11, respectively forming the first 4 elements and the last 4 elements of each row with the same bits in the new bit plane of the ciphertext image after the all-bit plane scrambling into 2 × 2 sub-blocks;

step 3.3 as shown in fig. 12, scanning all the generated 2 × 2 sub-blocks, and classifying all the 2 × 2 sub-blocks into a continuous block and a non-continuous block, where the continuous block is a2 × 2 sub-block in which all pixels in the block are 1 or 0, and the non-continuous block is a2 × 2 sub-block in which pixels in the block have 0 and 1; using 1 to represent continuous block and 0 to represent discontinuous block, generating block label matrix as key K₄Storing;

step 3.4, rearranging all the 2 x 2 sub-blocks according to the sequence of the preceding non-continuous block and the succeeding non-continuous block of the continuous block to obtain 8 new bit plane matrixes rearranged based on the 2 x 2 sub-blocks;

and 3.5, for the continuous block in the new bit plane matrix, setting the position of the pos in the block as a reserved position, embedding 1-bit data into the rest three positions respectively to obtain the continuous block after data embedding, wherein the pos is calculated according to a formula (3), MOD (DEG) is a modulus function, INT (DEG) is a downward integer function, and K is a K₅Is a random number as a key K₅Storing;

pos＝MOD(INT(K₅ ³),4) (3)

the remaining positions when pos is 0,1,2,3 are shown in fig. 13;

step 3.6 for non-consecutive blocks in the new bit-plane matrix, the position of the pos in the block is the predicted position, the value of the predicted position is the predicted value, the pos is calculated according to equation (3), assuming K ₅17, pos is 1;

step 3.7 determines the value of the predicted position, i.e. the predicted value p, according to the formula (4),

when the predicted value is the same as the predicted value, the non-continuous block is an embeddable block and is embedded at the predicted position1 bit of data, otherwise, the discontinuous block is a non-embeddable block, and a discontinuous block with embedded data is obtained; marking the embeddable blocks with 1 and the non-embeddable blocks with 0, generating a sequence L with the same length as the number of blocks, and using L as a secret key K₆Storing;

in fig. 14a, the black frame is the predicted position, and the sum of three pixels around the element is 1, then the predicted value is p equals to 0, and the predicted value is the same as the position value, so the block is an embeddable block, that is, the position can be used to embed information, and when recovering, the block can be completely recovered by using the three pixels around the block; in fig. 14b, the black frame is the predicted position and the predicted value is 0, but since the sum of the three surrounding pixels is 1, the predicted value is 1, and the predicted value is 0, the black frame is an embeddable block; FIG. 14c shows the black border in a position where it can be embedded;

step 3.8, all the one-dimensional data are embedded into the sub-blocks, and the recording block is embedded into the end position of the label and used as a secret key K₇Storing, namely recombining 8bit planes subjected to one-dimensional data embedding into an image to obtain a ciphertext image containing a watermark;

table 2 shows the results of the volume analysis of 6 images, and the data in the table is in bpp. The capacity of the algorithm is one of important indexes for measuring the performance of the algorithm, the higher the capacity is, the more the requirements of more users can be met, the advantage of the full bit plane embedding strategy is obvious in embedding capacity, theoretically, any pixel is converted into an 8-bit binary system, every 4 bits form 12 multiplied by 2 sub-blocks, 2 sub-blocks are in total, if 2 sub-blocks are continuous blocks, 2 multiplied by 3 to 6bit information can be embedded, and therefore the theoretically highest capacity can reach 6 bpp.

TABLE 2

As can be seen from table 2: the theoretical capacity and the actual capacity of the invention are both higher.

Table 3 shows the embedding capacity of 6 images in different algorithms.

TABLE 3

As can be seen from table 3, the embedding capacity of the present invention is much higher than that of documents [1,2,3,4], and the average values of the embedding capacities are about 70 times, 41 times, 2.83 times, and 1.27 times of those algorithms, respectively.

Step 4 image recovery and data extraction

Using an encryption key K₁-K₄Extracting a decrypted image from the ciphertext image containing the watermark, which comprises the following specific steps:

step 4.1 Using Key K₄Obtaining a block label matrix, rearranging blocks to obtain a new bit plane;

step 4.2, the new bit plane is rotated by 90 degrees in an inverse manner to obtain an original bit plane;

step 4.3, each row of 8-bit binary system of the original bit plane is converted into a decimal gray value, and all the gray values are combined into a ciphertext image;

step 4.4 Using Key K₂Constructing a traversal matrix;

step 4.5 Using Key K₃Carrying out inverse matrix decryption on the ciphertext image to obtain a decrypted image matrix;

step 4.6 Using Key K₁Rearranging the luminance block and the darkness block of the obtained classification map to obtain a decrypted image as shown in FIG. 15;

from the visual effect of fig. 15, it is difficult to see the difference between the decrypted image and the original image.

Table 4 shows the PNSR values of the decrypted image and the original image according to the embodiment of the present invention.

TABLE 4

As can be seen from table 4, the peak signal-to-noise ratio of the decrypted image to the original image is high, wherein the Airplane image PNSR is up to 69.9710, and the lowest value babon image also reaches 55.5628.

Table 5 shows the image quality results of the present invention with different algorithms.

TABLE 5

As can be seen from Table 5, the average PSNR of the invention is as high as 64.64dB, which is much higher than the algorithm of the documents [1,2,4,5 and 6], and the invention has good decrypted image quality.

Using an encryption key K₅-K₇Extracting a watermark image from a ciphertext image containing the watermark, which comprises the following specific steps:

step 4.7 utilizing the Key K₅Determining a data embedding position;

step 4.8 Using Key K₇Reading the end position of the block embedded label;

step 4.9, for the continuous block, sequentially reading all values except the determined position to obtain data embedded into the continuous block;

step 4.10 Using Key K₆Dividing the non-continuous blocks into embeddable blocks and non-embeddable blocks;

step 4.11, extracting the value of the determined data position in the embeddable block to obtain the data in the discontinuous block;

step 4.12, decrypting the data based on the Hilbert curve to obtain a watermark image;

step 4.13 Using Key K₅Determining the data embedding position and the reserved position by using the secret key K₇Reading the end position of the block embedded label;

step 4.14, for the continuous block, reading the value of the reserved position, and replacing all the values in the current block with the values of the reserved position;

step 4.15, for the non-continuous blocks, dividing the non-continuous blocks into embeddable blocks and non-embeddable blocks;

step 4.16, for the embeddable block, reading values of other three positions except the embedded position, if the number of 1 in the three values is more, setting the embedded position value to be 1, otherwise, setting the embedded position value to be 0, and obtaining a recovered ciphertext image;

the method adopts a double secret key to extract a watermark image from a ciphertext image containing the watermark and obtain an original image, and comprises the following specific steps:

step 4.17 Using Key K₅Determining a data embedding position;

step 4.18 Using Key K₇Reading the end position of the block embedded label;

step 4.19, for the continuous block, sequentially reading all values except the determined position to obtain data embedded into the continuous block;

step 4.20 Using Key K₆Dividing the non-continuous blocks into embeddable blocks and non-embeddable blocks;

step 4.21, extracting the value of the determined data position in the embeddable block to obtain the data in the discontinuous block;

step 4.22, decrypting the data based on the Hilbert curve to obtain a watermark image;

step 4.23 with the Key K₅Determining the data embedding position and the reserved position, and using the secret key K₇Reading the end position of the block embedded label;

step 4.24, for the continuous block, reading the value of the reserved position, and replacing all the values in the current block with the values of the reserved position;

step 4.25, for the non-continuous blocks, dividing the non-continuous blocks into embeddable blocks and non-embeddable blocks;

step 4.26, for the embeddable block, reading values of other three positions except the embedded position, if the number of 1 in the three values is more, setting the embedded position value to be 1, otherwise, setting the embedded position value to be 0, and obtaining a recovered ciphertext image;

step 4.27 Using Key K₄Obtaining a block label matrix, rearranging blocks of the recovered ciphertext image to obtain a new bit plane;

step 4.28, the new bit plane is rotated by 90 degrees in an inverse manner to obtain an original bit plane;

step 4.29, each row of 8-bit binary system of the original bit plane is converted into a decimal gray value, and all the gray values are combined into a ciphertext image;

step 4.30 Using Key K₂Constructing a traversal matrix;

step 4.31 utilize secretKey K₃Carrying out inverse matrix decryption on the ciphertext image to obtain a decrypted image matrix;

step 4.32 Using Key K₁Rearranging the brightness blocks and the darkness blocks of the obtained classification graph to obtain an original image;

table 6 shows the extracted watermark-to-original watermark similarity value NCC and the peak signal-to-noise ratio PSNR of the restored image to the original image.

TABLE 6

As can be seen from table 6, the similarity between the watermark extracted from the 6 ciphertext images with watermark and the original watermark is 1, and the PSNR of the recovered image and the original image is + ∞, which realizes 100% reversibility.

Fig. 16 shows the original image restored after salt and pepper noise attack and the extracted watermark image with a coefficient of 0.05 added, fig. 16a1-f1 show the original image restored after attack, and fig. 16a2-f2 show the watermark image extracted after attack. Under the attack of noise, the watermark image extracted and the restored original image are clear and visible, and certain robustness is achieved.

Table 7 shows the NCC values of the original watermark image and the extracted watermark image under the attack of salt and pepper noise.

TABLE 7

As can be seen from Table 7, the NCC values of the original watermark image and the extracted watermark image in the embodiment of the invention are both above 0.99, the similarity is extremely high, and the invention can effectively resist salt and pepper noise attack.

Fig. 17a1-f1 show ciphertext images with different positions subjected to shearing attack in the embodiment of the present invention, fig. 17a2-f2 show restored original images, and fig. 17a3-f3 show extracted watermark images, which show that both the restored original images and the extracted watermark images are clear and visible under shearing attack.

Table 8 shows the NCC values of the original watermark image and the extracted watermark image under the shearing attack at different positions and different proportions.

TABLE 8

As can be seen from table 8, the NCC values of the original watermark image and the extracted watermark image are both above 0.98, which indicates that the present invention can effectively resist shearing attack.

Reference documents:

[1]Qin Chuan,Zhang Wei,Cao Fang,et al.Separable reversible data hiding in encrypted images via adaptive embedding strategy with block selection[J].Signal Processing,2018,153:109-122

[2] separable encrypted domain reversible information hiding against ciphertext-only attacks [ J ] CAD and graphics bulletins 2020,32(6):874-

[3] Yang, Li Xin, Hu jin Si Chuan encrypted domain reversible information hiding based on block reconstruction [ J ] applied science bulletin 2021,39(06): 906-.

[4] Image encryption reversible data hiding based on prediction error adaptive coding [ J ] computer research and development, 2021,58(06):1340 + 1350.

[5]Ren Honglin Lu Wei,Chen Bing.Reversible data hiding in encrypted binary images by pixel prediction[J].Signal Processing,2019(07):268-277

[6] Li jin Wei, Zhang Xiao ya, Yao Yuan Zhi, Shu Nenera, ciphertext domain image reversible information hiding method [ J ] based on fine-grained embedded space reservation, network and information security bulletin, 2022,8(01): 106-.

Claims (1)

1. A separable ciphertext domain reversible data hiding method based on bit plane segmentation is characterized by comprising the following steps in sequence:

step 1: carrying out block-by-block classified scrambling encryption and traversal matrix encryption on original images

Step 1.1 divides an original image of size M × N into p × q sub-blocks of size M × N that do not overlap, where p is M/M and q is N/N;

step 1.2, dividing the sub-blocks into a brightness block and a darkness block according to a formula (1), and marking the brightness block with 1 and the darkness block with 0 to generate a classification chart B;

in the formula (1), num represents the number of pixels with the gray value being equal to or greater than 127 in the sub-block, i belongs to (1,2,3 … p), and j belongs to (1,2,3 … q);

step 1.3, performing Huffman compression coding on the classification chart B, and taking the coded data as a secret key K₁Storing;

step 1.4, traversing the classification graph B to rearrange, placing all the brightness blocks in front of the darkness blocks, rearranging the brightness blocks according to the original sequence, and rearranging the darkness blocks in the reverse sequence to obtain the image after the block classification scrambling;

step 1.5 Generation of a random number sequence L of length m n by equation (2)₁In the formula 3.5699456<μ≤4，0<x<1, the random number sequence L₁Disorder and no repetition, and the initial values of μ and x in formula (2)₀As a key K₂Storing;

x_k+1＝μx_k(1-x_k) (2)

step 1.6 sequence of random numbers L₁Sorting, the number sequence after sorting is marked as L₂According to L₁And L₂Generating a number pair according to the one-to-one correspondence relationship of the number pair, and constructing a traversal matrix according to the correspondence relationship of the number pair;

step 1.7, performing matrix traversal encryption for multiple times by taking the image subblocks after block classification scrambling as a unit to obtain a ciphertext image, wherein the matrix traversal encryption is to rearrange pixels in the subblocks according to a constructed traversal matrix, and take the number of times of the matrix traversal encryption as a secret key K₃Storing;

step 2: watermark image encryption

Dividing the watermark image into non-overlapping 8 x 8 sub-blocks, traversing all the sub-blocks in a snake shape in a horizontal direction first, and traversing all pixel points in each sub-block in a non-crossing manner according to a Hilbert curve to generate one-dimensional data;

and step 3: the method for hiding the reversible data in the ciphertext domain comprises the following specific steps:

step 3.1: carrying out all-position plane scrambling on the ciphertext image, namely combining the highest bits of 8 different pixels into a new 1 x 8 matrix, combining the seventh bits of 8 different pixels into a new 1 x 8 matrix, repeating the above steps, finally combining the lowest bits of 8 different pixels into a new 1 x 8 matrix, and recombining the 8 new 1 x 8 matrices into an 8 x 8 matrix to obtain the ciphertext image subjected to all-position plane scrambling;

step 3.2, respectively forming the front 4 elements and the rear 4 elements of each row with the same position in the new bit plane of the ciphertext image after the all-bit plane scrambling into 2 multiplied by 2 subblocks;

step 3.3, all the 2 × 2 sub-blocks generated by scanning are classified into continuous blocks and non-continuous blocks, wherein the continuous blocks are 2 × 2 sub-blocks with all pixels being 1 or 0, and the non-continuous blocks are 2 × 2 sub-blocks with pixels being 0 and pixels being 1 in the blocks; using 1 to represent continuous block and 0 to represent discontinuous block, generating block label matrix as key K₄Storing;

step 3.4, rearranging all the 2 x 2 sub-blocks according to the sequence of the preceding non-continuous block and the succeeding non-continuous block of the continuous block to obtain 8 new bit plane matrixes rearranged based on the 2 x 2 sub-blocks;

and 3.5, for the continuous block in the new bit plane matrix, setting the position of the pos in the block as a reserved position, embedding 1-bit data into the rest three positions respectively to obtain the continuous block after data embedding, wherein the pos is calculated according to a formula (3), MOD (DEG) is a modulus function, INT (DEG) is a downward integer function, and K is a K₅Is a random number as a key K₅Storing;

pos＝MOD(INT(K₅ ³),4) (3)

step 3.6, for the non-continuous block in the new bit plane matrix, the position of the pos in the block is a predicted position, the value of the predicted position is a predicted value, and the pos is calculated according to the formula (3);

step 3.7 determines the value of the predicted position, i.e. the predicted value p, according to equation (4),

when the predicted value is the same as the predicted value, the non-continuous block is an embeddable block, 1-bit data is embedded in the predicted position, otherwise, the non-continuous block is a non-embeddable block, and the non-continuous block with embedded data is obtained; marking the embeddable blocks with 1 and the non-embeddable blocks with 0, generating a sequence L with the same length as the number of blocks, and using L as a secret key K₆Storing;

step 3.8, all the one-dimensional data are embedded into the sub-blocks, and the recording block is embedded into the end position of the label and used as a secret key K₇Storing, namely recombining 8bit planes subjected to one-dimensional data embedding into an image to obtain a ciphertext image containing a watermark;

step 4 image recovery and data extraction

Using an encryption key K₁-K₄Extracting a decrypted image from the ciphertext image containing the watermark, which comprises the following specific steps:

step 4.1 Using Key K₄Obtaining a block label matrix, rearranging blocks to obtain a new bit plane;

step 4.2, the new bit plane is rotated by 90 degrees in an inverse manner to obtain an original bit plane;

step 4.3, each row of 8-bit binary system of the original bit plane is converted into a decimal gray value, and all the gray values are combined into a ciphertext image;

step 4.4 Using Key K₂Constructing a traversal matrix;

step 4.5 Using Key K₃Carrying out inverse matrix decryption on the ciphertext image to obtain a decrypted image matrix;

step 4.6 Using Key K₁Rearranging the brightness blocks and the darkness blocks of the obtained classification image to obtain a decrypted image;

using an encryption key K₅-K₇Extracting watermark from ciphertext image containing watermarkThe image comprises the following specific steps:

step 4.7 Using Key K₅Determining a data embedding position;

step 4.8 Using Key K₇Reading the end position of the block embedded label;

step 4.9, for the continuous block, sequentially reading all values except the determined position to obtain data embedded into the continuous block;

step 4.10 utilizing the Key K₆Dividing the non-continuous blocks into embeddable blocks and non-embeddable blocks;

step 4.11, extracting the value of the determined data position in the embeddable block to obtain the data in the discontinuous block;

step 4.12, decrypting the data based on the Hilbert curve to obtain a watermark image;

step 4.13 Using Key K₅Determining the data embedding position and the reserved position by using the secret key K₇Reading the end position of the block embedded label;

step 4.14, for the continuous block, reading the value of the reserved position, and replacing all the values in the current block with the values of the reserved position;

step 4.15, for the non-continuous blocks, dividing the non-continuous blocks into embeddable blocks and non-embeddable blocks;

step 4.16, for the embeddable block, reading values of other three positions except the embedded position, if the number of 1 in the three values is more, setting the embedded position value to be 1, otherwise, setting the embedded position value to be 0, and obtaining a recovered ciphertext image;

the method adopts a double secret key to extract a watermark image from a ciphertext image containing the watermark and obtain an original image, and comprises the following specific steps:

step 4.17 Using Key K₅Determining a data embedding position;

step 4.18 Using Key K₇Reading the end position of the block embedded label;

step 4.19, for the continuous block, sequentially reading all values except the determined position to obtain data embedded into the continuous block;

step 4.20 Using Key K₆Dividing the non-continuous blocks into embeddable blocks and non-embeddable blocks;

step 4.21, extracting the value of the determined data position in the embeddable block to obtain the data in the discontinuous block;

step 4.22, decrypting the data based on the Hilbert curve to obtain a watermark image;

step 4.23 Using Key K₅Determining the data embedding position and the reserved position, and using the secret key K₇Reading the end position of the block embedded label;

step 4.24, for the continuous blocks, reading the values of the reserved positions, and replacing all the values in the current block with the values of the reserved positions;

step 4.25, for the non-continuous blocks, dividing the non-continuous blocks into embeddable blocks and non-embeddable blocks;

step 4.26, for the embeddable block, reading values of other three positions except the embedded position, if the number of 1 in the three values is more, setting the embedded position value to be 1, otherwise, setting the embedded position value to be 0, and obtaining a recovered ciphertext image;

step 4.27 Using Key K₄Obtaining a block label matrix, rearranging blocks of the recovered ciphertext image to obtain a new bit plane;

step 4.28, the new bit plane is rotated by 90 degrees in an inverse manner to obtain an original bit plane;

step 4.29, each row of 8-bit binary system of the original bit plane is converted into a decimal gray value, and all the gray values are combined into a ciphertext image;

step 4.30 Using Key K₂Constructing a traversal matrix;

step 4.31 Using Key K₃Carrying out inverse matrix decryption on the ciphertext image to obtain a decrypted image matrix;

step 4.32 Using Key K₁And rearranging the brightness block and the darkness block of the obtained classification image to obtain an original image.

Images