Dynamic reversible information hiding method based on asymmetric digital system coding
By introducing a dynamic reversible information hiding method in asymmetric digital system encoding, using reversible dynamic probability predictor and ANS-Variant encoder, the problem of increasing storage burden of transmission carrier sequence distribution is solved, and efficient information hiding and low-rate distortion is achieved Effect, suitable for memory-constrained environments.
Patent Information
- Application Number
- CN202411001758.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-07-25
- Publication Date
- 2024-11-29
AI Technical Summary
In resource-constrained environments, transmitting the complete carrier sequence distribution increases additional storage and transmission burden, limiting the actual use value of the reversible information hidden encoding method in memory-constrained scenarios.
Using a dynamic reversible information hiding method based on asymmetric digital system encoding, by designing reversible dynamic probability predictors and ANS-Variant encoders, the changing carrier sequence distribution is used to replace the traditional static distribution, avoiding the dependence on prior knowledge and reducing the The overhead of storage space.
It realizes efficient information hiding in memory-constrained environments, reduces storage resource requirements, maintains a high embedded capacity and is close to the theoretical rate distortion boundary, and is suitable for scenarios such as wireless sensor networks.
Smart Images

Figure CN119046901A_ABST
Abstract
Description
Technical Field
[0001] The invention belongs to the technical field of information security, and in particular relates to a dynamic reversible information hiding method based on asymmetric digital system encoding. Background Art
[0002] With the rapid advancement of network technology and the popularity of digital multimedia in daily life, the importance of information security has become increasingly prominent, especially in the protection of digital product copyrights and data security. To meet these challenges, researchers have proposed reversible information hiding technology, which is an effective information protection strategy. This technology embeds secret information into the original carrier through a data embedding algorithm to generate a secret carrier, and then sends the secret carrier embedded with secret information to the receiver. The receiver uses an extraction algorithm to extract the embedded secret data and restore the original carrier without loss. The uniqueness of reversible information hiding technology lies in its ability to embed and extract information without losing any information. Therefore, reversible information hiding technology is widely used in fields that require extremely high losslessness of the original carrier. For example, in the field of military communications, reversible information hiding technology can ensure the secret transmission of command information while maintaining the transparency of communication channels. Another example is the field of judicial archive management, where reversible information hiding technology can embed sensitive case information into electronic documents, ensuring the security of the files while maintaining the originality and integrity of the files.
[0003] However, in reversible information hiding technology, many efficient algorithms require a certain degree of prior knowledge, such as the statistical properties of the carrier, the parameters of the embedding strategy, or the data structure used for encoding and decoding. This prior knowledge helps to improve embedding efficiency, reduce error rate, and ensure reversibility. Among them, the carrier sequence distribution describes the probability distribution pattern of the original carrier and is an important consideration in the design of many efficient reversible information hiding systems, especially when it is necessary to maximize the embedding capacity and maintain a low distortion level. However, transmitting the complete carrier sequence distribution will increase the additional storage and transmission burden, which is particularly challenging in resource-constrained environments. For example, in wireless sensor networks, the storage and computing resources of nodes are usually limited, and transmitting a large amount of prior knowledge may cause the storage space to be quickly exhausted and have a negative impact on the performance and life of the network. So far, although some reversible information hiding coding methods can approach the rate-distortion bound, they often require the transmission of carrier sequence distribution data in advance, which limits their practical use value in memory-constrained scenarios. Therefore, it is crucial to study methods to reduce the transmission of prior knowledge and develop more efficient encoding and extraction algorithms to reduce the demand for storage resources while maintaining reversibility. Summary of the invention
[0004] The purpose of the present invention is: in view of the constraint that the receiving end needs to obtain the carrier sequence distribution in advance, the present invention provides a dynamic reversible information hiding method based on asymmetric digital system encoding. The method uses a reversible dynamic probability calculator to replace the traditional static carrier sequence distribution with a changing carrier sequence distribution, thereby eliminating the dependence on prior knowledge and reducing the storage space overhead. It is suitable for memory-constrained environments, such as wireless sensor networks.
[0005] The technical solution of the present invention is:
[0006] A dynamic reversible information hiding method based on asymmetric digital system encoding, comprising:
[0007] A reversible dynamic probability predictor is designed to ensure that the carrier sequence distribution used when processing the same signal at the embedding end and the receiving end is exactly the same, avoiding the storage overhead caused by prior transmission;
[0008] A variant of asymmetric number system encoding, ANS-Variant, is used as the underlying reversible data hiding coding framework to achieve data embedding and extraction.
[0009] Preferably, the embedding and extraction process of the data is performed based on an assumed frequency distribution, and the frequency distribution is updated after processing each carrier signal to achieve dynamic adjustment of the distribution.
[0010] Preferably, the assumed frequency distribution is an initial distribution, which is known in advance to both the data embedding end and the receiving end. By designing a reversible dynamic probability predictor, when the encoder and decoder process the same signal in opposite order, a completely consistent carrier sequence distribution is used, thereby achieving synchronization of the carrier sequence distribution.
[0011] Preferably, an ANS-Variant encoder is used to embed the secret message, a decoder is used to generate the corresponding secret carrier, and the output of the encoder is communicated with the decoder as temporary information.
[0012] Preferably, in the data embedding stage, three data: carrier sequence X = (x1, x2, ···, x N ), secret message, assumed frequency distribution F X ={f X (x)|x∈Σ B} as input, where f X (x) represents the frequency of the carrier signal x.
[0013] Preferably, in the data embedding stage, the following steps are specifically included:
[0014] Step 1: traverse the vector sequence X=(x1,x2,···,x N) to calculate its frequency distribution; specifically, when encountering the carrier signal x i When the corresponding signal x i The frequency f X (x i ) is increased by 1, that is, f is updated X (x i ) has a value of f X (x i )+1; continue to process the next carrier signal and update its frequency in the same way until all carrier signals are processed; finally, the latest frequency distribution F is obtained. X ;
[0015] Step 2: Let F X i ={f X (x)|x∈Σ B} represents the frequency distribution at step i; when processing each carrier signal x i When the carrier signal x i The frequency f X (x i ) is reduced by 1, that is, f is updated X (x i ) has a value of f X (x i )-1; according to the updated F X i ={f X (x)|x∈Σ B}Calculate the cumulative frequency distribution of vector sequences Distribution of vector sequences CX ={cf X (k) / cf X (B)|k∈Σ B+1 Next, we apply the estimation algorithm to obtain the optimal marginal distribution P of the encrypted sequence. CY = {P CY (k)|k∈Σ B+1}; Based on the latest P CY ,pass With f Y (k) = cf Y (k+1)-cf Y (k) Calculate the statistical information F Y ={f Y (k)|k∈Σ B}and
[0016]
[0017] Step 3, let s i-1Indicates that the ANS-Variant encoder processes the i-th carrier signal x i Previous state, interval I: = [2 T-v×n ,2 T ), And satisfy the condition T,v∈Ν + , Tv×n≥0; in the data embedding stage, according to the current state s i-1 The value of , determines whether to embed the secret message; if the state value s i-1 Below the set interval I xi The lower limit of s i-1 <f X (x i )×2 T-v×n , then read v×n bits from the secret message, convert them into a decimal integer D, and pass them through s i-1 =(s i-1 <<v×n)+D updates the state to increase the state value; conversely, if the state is exactly within the interval, there is no need to change the state value;
[0018] Step 4, when state s i-1 Adjust to the specified range Based on the vector sequence distribution P CX , using the ANS-Variant encoder to encode the carrier signal x i Encode; the encoder outputs n bits of temporary information Where Rem(s i-1 ,f X (x i )) means s i-1 Divide by f X (x i ) and set the state Then, the temporary information output by the encoder is Transmitted to the decoder; the decoder uses temporary information Process and find a value y such that Decoder output y i =y; then, by Update the state value; if the state s i Exceeds the upper limit set during encoding 2 T , then s i The lowest v×n bits of the i =s i >>v×n updates the state to reduce the state value; if the state s i If it is exactly in interval I, there is no need to change the state value;
[0019] Repeat steps 2-4 for all carrier signals, and finally obtain a carrier sequence Y = (y1, y2, ···, y N ) and the final state s N ; In addition, if data is written to the secret message in step N, these v×n bits need to be stored additionally to ensure that the original secret message can be accurately restored during the extraction process.
[0020] Preferably, in the data extraction stage, first, the same assumed frequency distribution as in the data embedding stage is used, and the carrier sequence distribution P is calculated based on the current frequency distribution. CX ; Next, apply the estimation algorithm, through P CX Estimate the optimal marginal distribution P CY , and based on this distribution, the ANS-Variant encoder is used to encrypt the encrypted signal y i Encode; after the output of the encoder is transmitted to the decoder, the decoder is based on P CX Decode carrier signal x i , and update the assumed frequency distribution after decoding is completed; Similar to the above data embedding stage, in the decoding process of the carrier signal, the state value is also used as an indicator to extract the secret message;
[0021] In order to recover the secret message, the data extraction phase needs to input the encryption sequence Y=(y1,y2,···,y N ), final state s N , the same frequency distribution F as the initial assumption in the data embedding stage X′ ={f X′ (x)|x∈Σ B}, and possibly additional v×n bits of storage;
[0022] The data extraction stage processes the encrypted signals in reverse, starting from the last signal and processing them one by one.
[0023] Preferably, in the data extraction stage, the following steps are specifically included:
[0024] Step 1: Let F X′ i ={f X′ (x)|x∈Σ B} represents the frequency distribution at step i; according to the current frequency distribution F X′ i ={f X′ (x)|x∈Σ B}, you can get the statistical information of the vector sequence CF X ={cf X (k)|k∈Σ B+1} and vector sequence distribution P CX={P CX (k)|k∈Σ B+1}, where P CX (k)=cf X (k) / cf X (B);
[0025] Step 2, when processing each encrypted signal, first apply the estimation algorithm on the current carrier sequence distribution P CX to estimate the optimal marginal distribution P CY ; on this basis, through and f Y (k)=cf Y (k + 1)-cf Y (k)calculate the statistical information F Y ={f Y (k)|k∈Σ B}and
[0026]
[0027] Step 3, let s i ' represent the state before processing the i-th encrypted signal y i , the interval I:=[2 T-v×n , 2 T ), If the state s i ' is less than the lower bound of the interval , that is, s i ' < f Y (y i )×2 T-v×n , then read v×n bits from the secret message, represent it as decimal D, and update the value of s i ' to (s i ' << v×n)+D to increase the state s i '; if the state s i ' is exactly within the interval , then there is no need to change the state value; next, based on the optimal marginal distribution P CY , use the ANS-Variant encoder to encode the encrypted signal y i ; the encoder outputs a temporary information of n bits where Rem(s i ', f Y (y i )) represents the remainder of s i ' divided by f Y (y i ), and update the state
[0028] Step 4: Output the temporary information of the encoder Transmitted to the decoder; for temporary information Find a value x such that Decoder output x i = x, and set the state s i ' -1 The value of In decoding each carrier signal x i After that, it is necessary to increase its frequency f X′ (x i ) to update the statistics, that is, update f X′ (x i ) = f X′ (x i )+1;
[0029] Step 5, pass state s i ' -1 The value of determines whether to read the secret message; if the state s i ' -1 The value exceeds the upper bound 2 selected during encoding T , then s i ' -1 The lowest v×n bits of the secret message are written into the secret message, and the state s is updated. i ' -1 =s i ' -1 >>v×n to reduce the state value; if the state s i ' -1 If it is exactly within interval I, there is no need to change the state value;
[0030] Repeat the above five steps for all encrypted signals, and finally obtain the carrier sequence X=(x1,x2,···,x N ) and secret messages.
[0031] The advantages of the present invention are:
[0032] In order to reduce the extra overhead required for reversible information hiding, the present invention designs a dynamic reversible information hiding method based on asymmetric digital system encoding. The method uses a reversible dynamic probability calculator to replace the traditional static carrier sequence distribution with a changing carrier sequence distribution. This design eliminates the reliance on prior knowledge, reduces the overhead of storage space, and is suitable for memory-constrained environments, such as wireless sensor networks.
[0033] Experimental results show that the present invention achieves satisfactory embedding capacity while reducing overhead, which is close to the theoretical rate-distortion bound. Although the quality of the embedded image is reduced, this is due to the use of a changing frequency distribution rather than a real frequency distribution, but the present invention can avoid the additional overhead caused by the advance transmission of the probability distribution. This dynamic method achieves a good balance between storage space and embedding capacity. The present invention can reduce the demand for storage space while maintaining a high embedding capacity, which is of great significance for meeting the requirements for data privacy protection in memory-constrained scenarios. BRIEF DESCRIPTION OF THE DRAWINGS
[0034] The present invention will be further described below in conjunction with the accompanying drawings and embodiments;
[0035] Figure 1 is a flow chart of the data embedding stage in the present invention;
[0036] Figure 2 It is a flow chart of the data extraction stage in the present invention. DETAILED DESCRIPTION
[0037] In order to make the technical means, creative features, objectives and effects of the present invention easy to understand, the following embodiments and accompanying drawings specifically illustrate the high-capacity reversible information hiding method designed by the present invention.
[0038] Aiming at the constraint that the receiving end needs to obtain the carrier sequence distribution in advance, the present invention proposes a dynamic reversible information hiding method based on asymmetric digital system coding. The method designs a reversible dynamic probability predictor to ensure that the carrier sequence distribution used is exactly the same when the embedding end and the receiving end process the same signal, thereby avoiding the storage overhead caused by prior transmission. Then, the variant ANS-Variant of asymmetric digital system coding is used as the underlying reversible data hiding coding framework to realize data embedding and extraction.
[0039] In the present invention, the data embedding and extraction process is based on an assumed frequency distribution. In view of the fact that the assumed distribution may have certain errors, the method updates the frequency distribution after processing each carrier signal to achieve dynamic adjustment of the distribution, avoiding the problem of low embedding capacity caused by directly using uncertain distribution. Specifically, firstly, a frequency distribution is assumed to be the initial distribution, which is known in advance to both the data embedding end and the receiving end. By cleverly designing a reversible dynamic probability predictor, when the encoder and decoder process the same signal in opposite order, they use a completely consistent carrier sequence distribution, thereby achieving synchronization of the carrier sequence distribution. On this basis, the ANS-Variant encoder is used to embed the secret message, and the decoder is used to generate the corresponding secret carrier. The output of the encoder is used as temporary information to communicate with the decoder.
[0040] In the data embedding stage, three key data are required as input: carrier sequence X = (x1, x2, ···, x N ), secret message, assumed frequency distribution F X ={f X (x)|x∈Σ B}, where f X (x) represents the frequency of the carrier signal x.
[0041] This method updates the frequency distribution after processing each carrier signal and embeds the secret message using the ANS-Variant encoder and decoder based on the latest distribution.
[0042] The data embedding process is as follows Figure 1 As shown, assuming an initial frequency distribution of F X ={f X (x)=1|x∈Σ B The method processes the carrier signal from front to back in a signal-by-signal manner, converts each carrier signal into a secret carrier signal, and continuously modifies the assumed distribution in this process, and finally completes the embedding of the secret message into the secret carrier. Specifically, it includes the following steps:
[0043] The first step is to traverse the vector sequence X = (x1, x2, ···, x N ), for each carrier signal x i , update f X (x i ) = f X (x i )+1. After processing all carrier signals, the latest frequency distribution F is obtained. X ;
[0044] In the second step, when processing the first carrier signal x1, first convert the frequency f of the carrier signal x1 X The value of (x1) decreases by 1, that is, operation f is performed X (x1) = f X (x1)-1. According to the updated F X 1 Calculate the cumulative frequency distribution CF of the vector sequence X Distribution of vector sequences CX Next, we apply the estimation algorithm to obtain the optimal marginal distribution P of the encrypted sequence. CY On this basis, the statistical information F is calculated Y With CF Y ;
[0045] Step 3: Set the initial state value s0 of the ANS-Variant encoder. If the state value is too small, enlarge it by reading v×n bits from the secret message; otherwise, no changes are required to the state value.
[0046] The fourth step is based on P CX , perform ANS-Variant encoding on x1. During the encoding process, if the state value is too small, it is enlarged by reading v×n bits from the secret message; otherwise, no change is made to the state value. Next, based on P CY , perform decoding of the ANS-Variant encoder to generate the corresponding encrypted signal y1, and update the state value at the same time. If the state value is too large, write the lowest v×n bits of the state value into the secret message and reduce the state value; otherwise, no change is required to the state value.
[0047] After repeating the above steps 2 to 4 for all subsequent carrier signals, we can finally obtain a secret carrier and the final state s N And there may be additional v×n bits to be stored (in subsequent experiments, the parameters v=1 and n=16 are set, so that this overhead can be ignored).
[0048] The above is the main content of the data embedding stage in the present invention, and the following is an introduction to the data extraction stage.
[0049] In the data extraction stage, in order to ensure that the frequency distribution used for encoding and decoding the same signal is consistent, this method performs the inverse operation of the frequency distribution adjustment during the data embedding process. Specifically, first, the process uses the same assumed frequency distribution as the data embedding stage and calculates the carrier sequence distribution P based on the current frequency distribution. CX Next, we apply the estimation algorithm, by P CX Estimate the optimal marginal distribution P CY , and based on this distribution, the ANS-Variant encoder is used to encrypt the encrypted signal y i After the encoder output is transmitted to the decoder, the decoder is based on P CX Decode carrier signal x i , and updates the assumed frequency distribution after decoding is completed. Similar to the above data embedding stage, in the decoding process of the carrier signal, this method also uses the state value as an indicator to extract the secret message.
[0050] In order to recover the secret message, the data extraction phase needs to input the encryption sequence Y=(y1,y2,···,y N ), final state s N , the same frequency distribution F as the initial assumption in the data embedding stage X′ ={fX′ (x)|x∈Σ B}, and possibly additional v×n bits of storage.
[0051] The data extraction stage processes the encrypted signals in reverse, starting from the last signal and processing them one by one.
[0052] The data extraction process is as follows Figure 2 As shown, we first use the same assumed frequency distribution as in the data embedding stage, namely F X ={f X (x)=1|x∈Σ B}, and then process the secret signal in reverse order, restore each secret signal to the carrier signal one by one, and constantly modify the assumed distribution in this process, and finally successfully extract the secret message. The specific steps include:
[0053] The first step is to use the current frequency distribution F X N ={f X (x)=1|x∈Σ B}, obtain the cumulative frequency distribution CF of the vector sequence X Distribution of vector sequences CX ;
[0054] The second step is based on the P obtained in the previous step. CX , apply the estimation algorithm to estimate the optimal marginal distribution P CY , and then calculate the statistical information F Y With CF Y ;
[0055] The third step is to calculate the final state s. N , if the state value is too small, v×n bits are read from the secret message to enlarge the state value; otherwise, no change is required to the state value. Next, based on the optimal marginal distribution P of the secret carrier CY , for y N Execute the ANS-Variant encoding and update the status value;
[0056] The fourth step is to transmit the temporary information output by the encoder to the decoder based on the carrier sequence distribution P CX Perform decoding of the ANS-Variant to generate the corresponding carrier signal x N , and update the state value at the same time. In addition, it is necessary to increase the carrier signal x N The frequency f X (x N ) to update the statistics, that is, update f X (x N ) = f X (x N)+1, thus the next carrier signal x can be processed N-1 The frequency distribution F used X N-1 ;
[0057] Step 5. If the current state value is too large, write the lowest v×n bits of the state value into the secret message and reduce the state value; otherwise, no change is required to the state value.
[0058] After repeating the above five steps for all subsequent secret signals, the original carrier and secret message can be reconstructed losslessly.
[0059] Functions and Effects of the Embodiments
[0060] Comparison of experimental results is shown in Table 1, which compares the embedding performance of the present invention with other four state-of-the-art information hiding schemes on a standard 512×512 grayscale image. These information hiding schemes include: LPVO scheme (T.Zhang, histograms modification,"IEEE Transactions on Circuits and Systems for Video Technology,vol.30,no.8,pp.2329–2342,2020.),AC-RDH scheme (S.-J.Lin and W.-H.Chung,"The scalar scheme forreversible information-embedding in gray-scale signals:Capacity evaluationand code constructions,"IEEE Transactions on Information Forensics andSecurity,vol.7,no.4,pp.1155–1167,2012.) and IT-RDH scheme (Y.Jia,Z.Yin,X.Zhang,andY.Luo,"Reversibledata hiding based on reducing invalid shifting of pixels inhistogram shifting,"Signal Processing,vol.163,pp.238–246,2019.). As shown in Table 1, compared with the four prior arts, the scheme of the present invention is second only to the AC-RDH scheme in terms of average maximum embedding capacity, but is significantly better than the other three techniques. In addition, it is worth noting that the present invention does not require the carrier sequence distribution to be transmitted to the receiving end in advance, which is particularly important for memory-constrained environments.
[0061] Table 1 Comparison of experimental results between the method of the present invention and the existing information hiding technology
[0062]
[0063] The above embodiments are only for illustrating the technical concept and features of the present invention, and their purpose is to enable people familiar with the technology to understand the content of the present invention and implement it accordingly, and they cannot be used to limit the protection scope of the present invention. Any modifications made according to the spirit of the main technical solution of the present invention should be included in the protection scope of the present invention.
Claims
1. A dynamic reversible information hiding method based on asymmetric digital system encoding, It is characterized in that include: A reversible dynamic probability predictor is designed to ensure that the carrier sequence distribution used when processing the same signal at the embedding end and the receiving end is exactly the same, avoiding the storage overhead caused by prior transmission; A variant of asymmetric number system encoding, ANS-Variant, is used as the underlying reversible data hiding coding framework to achieve data embedding and extraction.
2. The dynamic reversible information hiding method according to claim 1, It is characterized in that The data embedding and extraction process is performed based on an assumed frequency distribution, which is updated after processing each carrier signal to achieve dynamic adjustment of the distribution.
3. The dynamic reversible information hiding method according to claim 2, It is characterized in that The assumed frequency distribution is an initial distribution, which is known in advance to both the data embedding end and the receiving end. By designing a reversible dynamic probability predictor, when the encoder and decoder process the same signal in opposite order, they use a completely consistent carrier sequence distribution, thereby achieving synchronization of the carrier sequence distribution.
4. The dynamic reversible information hiding method according to claim 3, It is characterized in that The secret message is embedded using the ANS-Variant encoder, and the corresponding secret carrier is generated using the decoder. The output of the encoder is used as temporary information to communicate with the decoder.
5. The dynamic reversible information hiding method according to claim 4, It is characterized in that In the data embedding stage, three data are embedded: carrier sequence X = (x 1 ,x 2 ,···,x N ), secret message, assumed frequency distribution F X ={f X (x)|x∈Σ B } as input, where f X (x) represents the frequency of the carrier signal x.
6. The dynamic reversible information hiding method according to claim 5, It is characterized in that In the data embedding stage, the following steps are included: Step 1, traverse the vector sequence X=(x 1 ,x 2 ,···,x N ) to calculate its frequency distribution; specifically, when encountering the carrier signal x i When the corresponding signal x i The frequency f X (x i ) is increased by 1, that is, f is updated X (x i ) has a value of f X (x i )+1; continue to process the next carrier signal and update its frequency in the same way until all carrier signals are processed; finally, the latest frequency distribution F is obtained. X ; Step 2, let F X i ={f X (x)|x∈Σ B } represents the frequency distribution at step i; when processing each carrier signal x i When the carrier signal x i The frequency f X (x i ) is reduced by 1, that is, update f X (x i ) has a value of f X (x i )-1; according to the updated F X i ={f X (x)|x∈Σ B }Calculate the cumulative frequency distribution of vector sequences Distribution of vector sequences CX ={cf X (k) / cf X (B)|k∈Σ B+1 Next, we apply the estimation algorithm to obtain the optimal marginal distribution P of the encrypted sequence. CY = {P CY (k)|k∈Σ B+1 }; Based on the latest P CY ,pass With f Y (k) = cf Y (k+1)-cf Y (k) Calculate the statistical information F Y ={f Y (k)|k∈Σ B } and CF Y ={cf Y (k)|k∈Σ B+1 }; Step 3, let s i-1 Indicates that the ANS-Variant encoder processes the i-th carrier signal x i Previous state, interval I: = [2 T -v×n ,2 T ), And satisfy the condition T,v∈Ν + , Tv×n≥0; in the data embedding stage, according to the current state s i-1 The value of , determines whether to embed the secret message; if the state value s i-1 Below the set range The lower limit of s i-1 <f X (x i )×2 T-v×n , then read v×n bits from the secret message, convert them into a decimal integer D, and pass them through s i-1 =(s i-1 <<v×n)+D updates the state to increase the state value; On the contrary, if the state happens to be within the interval, there is no need to change the state value; Step 4, when state s i-1 Adjust to the specified range Based on the vector sequence distribution P CX , using the ANS-Variant encoder to encode the carrier signal x i Encode; the encoder outputs n bits of temporary information Where Rem(s i-1 ,f X (x i )) means s i-1 Divide by f X (x i ) and set the state Then, the temporary information output by the encoder is Transmitted to the decoder; the decoder uses temporary information Process and find a value y such that Decoder output y i =y; then, by Update the state value; if the state s i Exceeds the upper limit set during encoding 2 T , then s i The lowest v×n bits of the i =s i >>v×n updates the state to reduce the state value; if the state s i If it is exactly in interval I, there is no need to change the state value; Repeat steps 2-4 for all carrier signals, and finally obtain a carrier sequence Y=(y 1 ,y 2 ,···,y N ) and the final state s N ; In addition, if data is written to the secret message in step N, these v×n bits need to be stored additionally to ensure that the original secret message can be accurately restored during the extraction process.
7. The dynamic reversible information hiding method according to claim 6, It is characterized in that In the data extraction phase, first, the same assumed frequency distribution as in the data embedding phase is used, and the vector sequence distribution P is calculated based on the current frequency distribution. CX ; Next, apply the estimation algorithm, through P CX Estimate the optimal marginal distribution P CY , and based on this distribution, the ANS-Variant encoder is used to encrypt the encrypted signal y i Encode; after the output of the encoder is transmitted to the decoder, the decoder is based on P CX Decode carrier signal x i , and update the assumed frequency distribution after decoding is completed; Similar to the above data embedding stage, in the decoding process of the carrier signal, the state value is also used as an indicator to extract the secret message; In order to recover the secret message, the data extraction phase requires the input of the encryption sequence Y = (y 1 ,y 2 ,···,y N ), final state s N , the same frequency distribution F as the initial assumption in the data embedding stage X′ ={f X′ (x)|x∈Σ B }, and possibly additional v×n bits of storage; The data extraction stage processes the encrypted signals in reverse, starting from the last signal and processing them one by one.
8. The dynamic reversible information hiding method according to claim 7, It is characterized in that The data extraction phase includes the following steps: Step 1: Let F X′ i ={f X′ (x)|x∈Σ B } represents the frequency distribution at step i; according to the current frequency distribution F X′ i ={f X′ (x)|x∈Σ B }, you can get the statistical information of the vector sequence CF X ={cf X (k)|k∈Σ B+1 } and vector sequence distribution P CX = {P CX (k)|k∈Σ B+1 },in P CX (k) = cf X (k) / cf X (B); Step 2: When processing each encrypted signal, first distribute P in the current carrier sequence CX Apply the estimation algorithm to estimate the optimal marginal distribution P CY On this basis, through With f Y (k) = cf Y (k+1)-cf Y (k) Calculate the statistical information F Y ={f Y (k)|k∈Σ B } and CF Y ={cf Y (k)|k∈Σ B+1 }; Step 3, let s i ′ denote the state before processing the i-th encrypted signal y i , the interval I := [2 T-v×n , 2 T ). If the state s i ′ is less than the lower bound of the interval , that is, s i ′ < f Y (y i ) × 2 T-v×n , then read v × n bits from the secret message, represent them as the decimal D, and increase the state s i ′ by updating its value to (s i ′ << v × n) + D; if the state s i ′ is exactly within the interval i , then there is no need to change the state value; next, based on the optimal marginal distribution P , use the ANS-Variant encoder to encode the encrypted signal y CY ; the encoder outputs a temporary message of n bits i where Rem(s ′, f i (y Y )) represents the remainder of s i ′ divided by f i (y Y ), and update the state i ) and update the state Step 4: Output the temporary information of the encoder Transmitted to the decoder; for temporary information Find a value x such that Decoder output x i = x, and set the state s i ' -1 The value of In decoding each carrier signal x i After that, it is necessary to increase its frequency f X′ (x i ) to update the statistics, that is, update f X′ (x i ) = f X′ (x i )+1; Step 5, pass state s i ' -1 The value of determines whether to read the secret message; if the state s i ' -1 The value exceeds the upper bound 2 selected during encoding T , then s i ' -1 The lowest v×n bits of the secret message are written into the secret message, and the state s is updated. i ' -1 =s i ' -1 >>v×n to reduce the state value; if the state s i ' -1 If it is exactly within interval I, there is no need to change the state value; Repeat the above five steps for all encrypted signals, and finally obtain the carrier sequence X=(x 1 ,x 2 ,···,x N ) and secret messages.
