CN111640443B - Audio reversible steganography method and secret information extraction method - Google Patents
Audio reversible steganography method and secret information extraction method Download PDFInfo
- Publication number
- CN111640443B CN111640443B CN202010290729.2A CN202010290729A CN111640443B CN 111640443 B CN111640443 B CN 111640443B CN 202010290729 A CN202010290729 A CN 202010290729A CN 111640443 B CN111640443 B CN 111640443B
- Authority
- CN
- China
- Prior art keywords
- value
- sampling
- block
- audio
- sub
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Classifications
-
- G—PHYSICS
- G10—MUSICAL INSTRUMENTS; ACOUSTICS
- G10L—SPEECH ANALYSIS OR SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING; SPEECH OR AUDIO CODING OR DECODING
- G10L19/00—Speech or audio signals analysis-synthesis techniques for redundancy reduction, e.g. in vocoders; Coding or decoding of speech or audio signals, using source filter models or psychoacoustic analysis
- G10L19/018—Audio watermarking, i.e. embedding inaudible data in the audio signal
Abstract
The invention relates to an audio reversible steganography method, which comprises the following steps: the audio carrier is divided into fixed length sub-blocks,the sampling values in each sub-block are arranged in ascending order; then calculating the complexity of each ascending sub-block, and continuously maintaining the original value of the sub-block according to whether the complexity is larger than a preset threshold v or not; if not, determining the actual complexity level, and calculating to obtain an optimal predicted sampling value in the ascending sub-block; finally, calculating a prediction error according to the optimal prediction sampling value, and then continuously judging whether the complexity of the ascending sub-block is larger than that of the ascending sub-block
Description
Technical Field
The invention relates to the field of audio encryption, in particular to an audio reversible steganography method and a secret information extraction method.
Background
Information hiding is to embed secret information into various digital media such as images, audio, video, etc. under the condition that the sense of a person cannot be perceived. The conventional information hiding is finished, and the carrier can generate irreversible distortion, and the methods are collectively called irreversible information hiding. In some special cases, this distortion is not allowed. To solve this problem, researchers have proposed reversible information hiding techniques. It features that after the secret information is extracted, the original carrier can be recovered completely without any distortion.
In 2003, tian et al proposed a reversible information hiding framework based on difference extensions (difference expansion, DE). The method comprises the steps of firstly, performing difference on two continuous pixel values, then converting the difference value into a binary form, and finally adding a secret information bit to the lowest bit of the binary difference value. This method is simple to operate, but the distortion is large after the difference between adjacent pixels is spread. Based on DE, thodi et al propose to use prediction errors to replace the differences of neighboring pixels for extended embedding, and experiments have shown that the prediction error extensions (prediction error expansion, PEE) have lower distortion than DE, so many of the latter work starts to turn to PEE. The essence of these methods is that the vector is greatly modified by means of the spread embedding by the difference value, the prediction error, etc., and the distortion performance is inferior to that of the Histogram Shift (HS) method. Tsai combines the prediction error with HS to effectively reduce distortion at the expense of a portion of the embedding capacity. Ou et al, in turn, utilized a plurality of histogram modifications (multiple histograms modification, MHM) to embed on the basis of Tsai, making up for the deficiency of Ou in embedding capacity. In 2013, li et al proposed a reversible steganography framework for pixel value ordering (pixel value ordering, PVO). A picture is divided into a plurality of pixel blocks with fixed sizes, then the pixel values in the pixel blocks are sequenced, the maximum (small) value is predicted through the second large (small) value, and finally the secret information is embedded into the pixel value with the prediction error value of 1 or minus 1. Because the carrier is only slightly modified, the carrier after steganography has the characteristic of high fidelity. Peng et al propose Improved Pixel Value Ordering (IPVO) based on PVO, and increase the prediction error "0" as the embedding condition, effectively improving the embedding capacity. Weng et al research on prediction mechanisms in pixel blocks, increase the number of prediction errors generated in pixel blocks, and improve the utilization rate of pixels in blocks.
At present, most of research work of reversible steganography is focused on the image field, and the reversible steganography algorithm using audio as a carrier is not so much. The DE framework was first applied to audio carriers by the applicant, confirming the versatility of the DE framework. Wang combines the features of audio to propose a new reversible audio watermarking scheme based on improved Prediction Error Extension (PEE) and Histogram Shifting (HS). Compared with the severe et al, the embedded capacity and the signal to noise ratio are further improved. The Xiang divides the audio time domain sampling value into two sets according to the odd and even of the position, a complex non-causal prediction algorithm is provided to calculate the prediction error, then the secret information is embedded into the prediction error in an expanded mode, and the prediction performance of the method is optimal at present. Thus, the capacity control is optimized on the basis of the Xiang, and the performance is improved to a certain extent under the condition of high embedding rate. In summary, current audio reversible steganography research is mostly spread around PEE, and factors limiting the performance of such algorithms mainly have two points: firstly, prediction precision and secondly, an embedding mechanism. Although the Xiang improves the prediction precision to a new height, the embedding mechanism also utilizes binary expansion of the prediction error to carry out embedding, and the modification amplitude of the carrier by the method is larger, so that the high-fidelity embedding effect is difficult to obtain. Since the secret information is often loaded into the pixel values with the prediction error values of 0 and 1 during the image steganography, the variation amplitude of the audio sampling value is larger than that of the image, and therefore if the sampling values with the prediction error values of 0 and 1 in the audio are used for embedding, the problem of insufficient embedding capacity exists at the moment, and further improvement is needed.
Disclosure of Invention
The first technical problem to be solved by the invention is to provide an audio reversible steganography method which can obtain a high-fidelity steganography effect and simultaneously improve the steganography capacity and has a better steganography effect.
The second technical problem to be solved by the invention is to provide a secret information extraction method based on the reversible steganography method aiming at the current state of the art.
The technical scheme adopted by the invention for solving the first technical problem is as follows: an audio reversible steganography method is used for embedding secret information into an audio carrier S with the length of N to obtain a secret-containing audio S'; wherein s= { x 1 ,x 2 ，...,x N }，x 1 For sample 1 in the audio carrier S, x 2 For sample 2 in the audio carrier S, x N Is the nth sample value in the audio carrier S; the method is characterized in that: the method comprises the following steps:
step 1, dividing an audio carrier S with the length of N into sub-blocks with the length of N; wherein the audio carrier s= { x after division 1 ,S 1 ,x n+2 ,S 2 ,x 2n+3 ,...S i ...,x N Expression S of the ith sub-block i ＝{x i+n*(i-1)+1 ,x i+n*(i-1)+2 ，...,x i+n*i }，Representing a downward rounding; and the ith subblock S i The next previous sample value is x i+n*(i-1) And the ith subblock S i The next sample value is x i+n*i+1 ；
Step 2, respectively carrying out ascending arrangement on sampling values in each sub-block to obtain ascending sub-blocks; wherein the ith sub-block S i Corresponding ascending sub-block X i The expression is: x is X i ＝{x σ(1) ,x σ(2) ,...,x σ(n) }，x σ(1) ≤x σ(2) ≤...≤x σ(n) The method comprises the steps of carrying out a first treatment on the surface of the Sigma (i) is the ascending sub-block X i Middle sampling value x σ(i) Corresponding to the atomic block S i Is a position in the middle; sigma (i) ∈1, n]；
Step 3, obtaining an ith subblock S according to the preset complexity level T, wherein T is a positive integer i Corresponding ascending sub-block X i Sample value set { x for computational complexity in the interior σ(T+2) ,x σ(T+3) ,...x σ(n-T-1) And calculate and get the ith ascending sub-block X i The complexity delta of (2) is calculated as:
wherein μ is x σ(T+2) x σ(T+3) ...x σ(n-T-1) 、x i+n*(i-1) And x i+n*i+1 The average value corresponding to all sampling values;
step 4, judging the ith ascending sub-block X calculated in the step 3 i If the complexity delta of (a) is greater than the preset threshold v, if so, then the current ascending sub-block X i For complex blocks, the ith sub-block S in the audio carrier S i Continuously maintaining the original value of the sampling value, and transferring to the step 9; if not, then the current ascending subBlock X i Is a smooth block and is transferred to the step 5; the initial value of i is 1;
step 5, calculating an actual complexity level k meeting the following formula, wherein k is a positive integer;and correspondingly obtaining the current ascending sub-block X according to the actual complexity level k i In the optimal predicted sample value x σ(k+1) And x σ(n-k) ；
Step 6, calculating optimal predicted sampling value x respectively σ(k+1) And { x } σ(1) ,x σ(2) ...x σ(k) Error between each sampling value in the sequence to obtain k prediction errors PE min The method comprises the steps of carrying out a first treatment on the surface of the Respectively calculating optimal predicted sampling value x σ(n-k) And { x } σ(n-k+1) ,x σ(n-k+2) ...x σ(n) Error between each sampling value in the sequence to obtain k prediction errors PE max ；
Step 7, judging the current ascending sub-block X in the step 3 i Whether the complexity delta of (a) is greater thanIf yes, then for { x } σ(1) ,x σ(2) ...x σ(k) Intermediate and optimal predicted sample value x σ(k+1) Sampling value with prediction error value D and { x }, between σ(n-k+1) ,x σ(n-k+2) ...x σ(n) Intermediate and x σ(n-k) Performing embedding operation on sampling values with the prediction error value D, and performing shifting operation on sampling values with the prediction error value greater than D, wherein D is a natural number greater than 1; and the new sampling value obtained after the embedding operation and the shifting operation is used for replacing the original sampling value, and the sampling value with the prediction error value smaller than D is continuously kept at the original value to obtain { x } σ(1) ,x σ(2) ...x σ(k) New sample value { x } 'corresponding to' σ(1) ,x′ σ(2) ...x′ σ(k) { x } σ(n-k+1) ,x σ(n-k+2) ...x σ(n) New sample value { x } 'corresponding to' σ(n-k+1) ,x′ σ(n-k+2) ...x′ σ(n) -and go to step 8; wherein, inlay the secret operation andthe calculation formula of the shift operation is:
m∈[1,k]；
m∈[n-k+1,n]；
if not, then for { x } σ(1) ,x σ(2) ...x σ(k) Intermediate and optimal predicted sample value x σ(k+1) Sampling values with prediction error values of 0 and 1 and { x }, respectively σ(n-k+1) ,x σ(n-k+2) ...x σ(n) Intermediate and x σ(n-k) Performing embedding operation on sampling values with prediction error values of 0 and 1, performing shifting operation on sampling values with prediction error values smaller than 0 or larger than 1, and replacing the original sampling values with new sampling values obtained after the embedding operation and the shifting operation to obtain { x } σ(1) ,x σ(2) ...x σ(k) New sample value { x } 'corresponding to' σ(1) ,x′ σ(2) ...x′ σ(k) { x } σ(n-k+1) ,x σ(n-k+2) ...x σ(n) New sample value { x } 'corresponding to' σ(n-k+1) ,x′ σ(n-k+2) ...x′ σ(n) -a }; and go to step 8; the calculation formulas of the embedding operation and the shifting operation are as follows:
q∈[1,k]；
q∈[n-k+1,n]；
step 8, the { x } 'obtained in the step 7' σ(1) ,x′ σ(2) ...x′ σ(k) Sum { x' σ(n-k+1) ,x′ σ(n-k+2) ...x′ σ(n) ' current sub-block S i { x in } σ(k+1) ,x σ(k+2) ...x σ(n-k) Respectively according to the current sub-block S i Is the origin of (1)Beginning order replacing ith S in audio carrier S i Sampling values at corresponding positions in the sub-blocks;
step 9, judging whether all secret information is embedded at the moment, if yes, turning to step 10; if not, making i=i+1, and turning to the step 4, and continuing to perform embedding judgment on the next sub-block;
step 10, replacing the sub-blocks in the audio carrier S by the step 4 or the step 8 to obtain a new audio carrier, wherein the obtained new audio carrier is the secret audio S' embedded with the secret information.
Specifically, in the step 6, PE min The calculation formula of (2) is as follows:
PE min ＝x s -x t
a＝1,2,...k；
PE max the calculation formula of (2) is as follows:
PE＝x-x maxuv
b＝n-k+1,n-k+2,..n；。
as an improvement, the method for determining the value D in the step 7 is as follows: firstly, presetting the value range of D as [ D ] min ,D max ]Counting k prediction errors PE in step 6 min And k prediction errors PE max The prediction error values of the two are respectively D min To D max And taking the prediction error value corresponding to the value with the largest number as D.
The invention solves the second technical problem by adopting the technical proposal that: a secret information extraction method is used for extracting secret information from a secret-containing audio S ', wherein S ' = { x ' 1 ,x′ 2 ...,x′ N }，x′ 1 Is the 1 st sampling value, x ' in the audio S ' with the secret ' 2 Is the 2 nd sampling value, x ' in the audio S ' with the secret ' N Is the nth sampling value in the encrypted audio S'; the method is characterized in that: the method comprises the following steps:
step 1, dividing the encrypted audio S ' into sampling blocks with the length of n, wherein the length of each sampling block is the same as the length of each sub-block during audio steganography, and the divided encrypted audio S ' = { x ' 1 ,S′ 1 ,x′ n+2 ,S′ 2 ,x′ 2n+3 ,...S′ i ...,x′ N Expression S 'of the ith sample block' i ＝{x′ i+n*(i-1)+1 ,x′ i+n*(i-1)+2 ...,x′ i+n*i }，Representing a downward rounding; and the ith sample block S' i The immediately preceding sample value is x' i+n*(i-1) And the ith sampling block S' i The next subsequent sample value is x' i+n*i+1 ；
Step 2, respectively carrying out ascending arrangement on sampling values in each sampling block to obtain ascending sampling blocks; wherein the ith sample block S' i Corresponding ascending sampling block X' i The expression is: x'. i ＝{x′ σ(1) ,x′ σ(2) ,...,x′ σ(n) }，x′ σ(1) ≤x′ σ(2) ≤...≤x′ σ(n) The method comprises the steps of carrying out a first treatment on the surface of the Sigma (i) is the ascending sample block X' i Middle sampling value x' σ(i) Corresponding to the original sampling block S' i Is a position in the middle; sigma' (i) ∈1, n]；
Step 3, obtaining an ith sampling block S 'according to the preset complexity level T same as that of the audio steganography' i Corresponding ascending sampling block X' i Sample value set { x } 'for computational complexity in the interior' σ(T+2) ,x′ σ(T+3) ,...x′ σ(n-T-1) And calculate the ith sample block X' i The complexity delta' of (a) is calculated as:
wherein μ 'is x' σ(T+2) ,x′ σ(T+3) ...x′ σ(n-T-1) 、x′ i+n*(i-1) And x' i+n*i+1 The average value corresponding to all sampling values;
step 4, judging the i-th sampling block X 'calculated in the step 3' i If the complexity delta 'of (a) is greater than the same preset threshold v as when the audio is steganographic, if so, then the current up-sampled block X' i For complex blocks, the ith sample block S ' in the dense audio S ' is sampled ' i Continuously maintaining the original value of the sampling value, and transferring to the step 8; if not, then the current up-sample block X' i Is a smooth block and is transferred to the step 5; the initial value of i is 1;
step 5, obtaining the current ascending sampling block X 'by using the same actual complexity level k as the audio steganography' i Is equal to the optimal predicted sample value x' σ(k+1) And x' σ(n-k) ；
Step 6, respectively calculating optimal predicted sampling values x' σ(k+1) And { x' σ(1) ,x′ σ(2) ...x′ σ(k) Error between each sample value in the sequence, k prediction errors PE 'are obtained' min The method comprises the steps of carrying out a first treatment on the surface of the And respectively calculating optimal predicted sampling value x' σ(n-k) And { x' σ(n-k+1) ,x′ σ(n -k+2)...x′ σ(n) Error between each sample value in the sequence, k prediction errors PE 'are obtained' max ；
Step 7, judging the current ascending sampling block X 'in the step 3' i Whether the complexity delta' of (a) is greater thanIf so, then for { x' σ(1) ,x′ σ(2) ...x′ σ(k) In } and x' σ(k+1) Sampling values with prediction error values of D and D+1 and { x' σ(n-k+1) ,x′ σ(n -k+2)...x′ σ(n) In } and x' σ(n-k) And carrying out secret information extraction on sampling values with the prediction error value of D and D+1, wherein the extracted secret information b has a calculation formula as follows:
b＝PE′ min -D,ifPE′ min ∈{D,D+1}
b＝PE′ max -D,ifPE′ max ∈{D,D+1}；
if not, then for { x' σ(1) ,x′ σ(2) ...x′ σ(k) Intermediate and optimal predicted sample value x' σ(k+1) Sampling values with prediction error values of-1, 0,1 and 2, { x' σ(n-k+1) ,x′ σ(n-k+2) ...x′ σ(n) In } and x' σ(n-k) The method comprises the steps that secret information extraction is carried out on sampling values with prediction error values of-1, 0,1 and 2, wherein D is the same value as that of audio steganography; the calculation formula for extracting the secret information b is as follows:
step 8, judging whether all secret information is extracted at the moment, if yes, turning to step 9; if not, making i=i+1, and turning to the step 4;
and 9, forming the extracted secret information in each sampling block into complete secret information according to the sequence of the sampling blocks, and obtaining the secret information extracted from the secret-containing audio S'.
Compared with the prior art, the invention has the advantages that: according to the audio characteristics of different types, the prediction error value D is determined, and the embedding is carried out on the sampling block with the prediction error value D, so that the embedding capacity can be obviously improved, and the secret information embedding is carried out on the sampling value with higher similarity by adaptively determining the prediction error value D, so that the steganography method has better steganography effect.
Detailed Description
The present invention is described in further detail below with reference to examples.
An audio reversible steganography method is used for embedding secret information into an audio carrier S with the length of N to obtain a secret-containing audio S'; wherein the method comprises the steps of，S＝{x 1 ,x 2 ，...,x N }，x 1 For sample 1 in the audio carrier S, x 2 For sample 2 in the audio carrier S, x N Is the nth sample value in the audio carrier S; n is a natural number, comprising the following steps:
step 1, dividing an audio carrier S with the length of N into sub-blocks with the length of N; wherein the audio carrier s= { x after division 1 ,S 1 ,x n+2 ,S 2 ,x 2n+3 ,...S i ...,x N Expression S of the ith sub-block i ＝{x i+n*(i-1)+1 ,x i+n*(i-1)+2 ，...,x i+n*i }，Representing a downward rounding; and the ith subblock S i The next previous sample value is x i+n*(i-1) And the ith subblock S i The next sample value is x i+n*i+1 ；
Step 2, respectively carrying out ascending arrangement on sampling values in each sub-block to obtain ascending sub-blocks; wherein the ith sub-block S i Corresponding ascending sub-block X i The expression is: x is X i ＝{x σ(1) ,x σ(2) ,...,x σ(n) }，x σ(1) ≤x σ(2) ≤...≤x σ(n) The method comprises the steps of carrying out a first treatment on the surface of the Sigma (i) is the ascending sub-block X i Middle sampling value x σ(i) Corresponding to the atomic block S i Is a position in the middle; sigma (i) ∈1, n]；
Step 3, obtaining an ith subblock S according to the preset complexity level T, wherein T is a positive integer i Corresponding ascending sub-block X i Sample value set { x for computational complexity in the interior σ(T+2) ,x σ(T+3) ,...x σ(n-T-1) And calculate and get the ith ascending sub-block X i The complexity delta of (2) is calculated as:
wherein μ is x σ(T+2) x σ(T+3) ...x σ(n-T-1) 、x i+n*(i-1) And x i+n*i+1 The average value corresponding to all sampling values;
the range of the complexity level T is as follows:
step 4, judging the ith ascending sub-block X calculated in the step 3 i If the complexity delta of (a) is greater than the preset threshold v, if so, then the current ascending sub-block X i For complex blocks, the ith sub-block S in the audio carrier S i Continuously maintaining the original value of the sampling value, and transferring to the step 9; if not, then the current ascending sub-block X i Is a smooth block and is transferred to the step 5; the initial value of i is 1;
step 5, calculating an actual complexity level k meeting the following formula, wherein k is a positive integer;
and correspondingly obtaining the current ascending sub-block X according to the actual complexity level k i
In the optimal predicted sample value x σ(k+1) And x σ(n-k) ；k∈[1,T]；
Step 6, calculating optimal predicted sampling value x respectively σ(k+1) And { x } σ(1) ,x σ(2) ...x σ(k) Error between each sampling value in the sequence to obtain k prediction errors PE min The method comprises the steps of carrying out a first treatment on the surface of the Respectively calculating optimal predicted sampling value x σ(n-k) And { x } σ(n-k+1) ,x σ(n-k+2) ...x σ(n) Error between each sampling value in the sequence to obtain k prediction errors PE max ；
PE min The calculation formula of (2) is as follows:
PE min ＝x s -x t
a＝1,2,...k；
PE max the calculation formula of (2) is as follows:
PE＝x-x maxuv
b＝n-k+1,n-k+2,..n；
step 7, judging the current ascending sub-block X in the step 3 i Whether the complexity delta of (a) is greater thanIf yes, then for { x } σ(1) ,x σ(2) ...x σ(k) Intermediate and optimal predicted sample value x σ(k+1) Sampling value with prediction error value D and { x }, between σ(n-k+1) ,x σ(n-k+2) ...x σ(n) Intermediate and x σ(n-k) Performing embedding operation on sampling values with the prediction error value D, and performing shifting operation on sampling values with the prediction error value greater than D, wherein D is a natural number greater than 1; and the new sampling value obtained after the embedding operation and the shifting operation is used for replacing the original sampling value, and the sampling value with the prediction error value smaller than D is continuously kept at the original value to obtain { x } σ(1) ,x σ(2) ...x σ(k) New sample value { x } 'corresponding to' σ(1) ,x′ σ(2) ...x′ σ(k) { x } σ(n-k+1) ,x σ(n-k+2) ...x σ(n) New sample value { x } 'corresponding to' σ(n-k+1) ,x′ σ(n-k+2) ...x′ σ(n) -and go to step 8; the calculation formulas of the embedding operation and the shifting operation are as follows:
m∈[1,k]；
m∈[n-k+1,n]；
if not, then for { x } σ(1) ,x σ(2) ...x σ(k) Intermediate and optimal predicted sample value x σ(k+1) Sampling value with prediction error value of 0 and 1{ x } σ(n-k+1) ,x σ(n-k+2) ...x σ(n) Intermediate and x σ(n-k) Performing embedding operation on sampling values with prediction error values of 0 and 1, performing shifting operation on sampling values with prediction error values smaller than 0 or larger than 1, and replacing the original sampling values with new sampling values obtained after the embedding operation and the shifting operation to obtain { x } σ(1) ,x σ(2) ...x σ(k) New sample value { x } 'corresponding to' σ(1) ,x′ σ(2) ...x′ σ(k) { x } σ(n-k+1) ,x σ(n-k+2) ...x σ(n) New sample value { x } 'corresponding to' σ(n-k+1) ,x′ σ(n-k+2) ...x′ σ(n) -a }; and go to step 8; the calculation formulas of the embedding operation and the shifting operation are as follows:
q∈[1,k]；
q∈[n-k+1,n]；
in this scheme, the method for determining the value D is: firstly, presetting the value range of D as [ D ] min ,D max ]Counting k prediction errors PE in step 6 min And k prediction errors PE max The prediction error values of the two are respectively D min To D max The total number corresponding to each integer, and taking the prediction error value corresponding to the value with the maximum total number as D.
In this embodiment, because different types of audio features are different, the value range of the set value D is: 2,5, respectively counting the number of the prediction error values of 2,3,4 and 5, and obtaining the prediction error value with the largest number as D;
the above formula for performing the embedding operation on the sampling values with the prediction error values of 0 and 1 is cited in the article: fei Peng, xiaolon Li, bin Yang, improved PVO-based reversible data hiding, digital Signal Processing 2014;
the embedding operation and the shifting operation obtain x' σ(q) The specific deduction steps of the calculation formula of (a) are as follows: q is E [1, k]；
Embedding bit b e {0,1} of secret information into prediction error PE min In (3) to obtain PE' min ；
Similarly, the above-described embedding operation and shifting operation result in x' σ(q) The specific deduction steps of the calculation formula of (a) are as follows: q epsilon [ n-k+1, n]
Embedding bit b e {0,1} of secret information into prediction error PE max In (3) to obtain PE' max ；
Step 8, the { x } 'obtained in the step 7' σ(1) ,x′ σ(2) ...x′ σ(k) Sum { x' σ(n-k+1) ,x′ σ(n-k+2) ...x′ σ(n) ' current sub-block S i { x in } σ(k+1) ,x σ(k+2) ...x σ(n-k) Respectively according to the current sub-block S i Is the original order of replacement of the ith S in the audio carrier S i Sampling values at corresponding positions in the sub-blocks;
step 9, judging whether all secret information is embedded at the moment, if yes, turning to step 10; if not, making i=i+1, and turning to the step 4, and continuing to perform embedding judgment on the next sub-block;
step 10, replacing the sub-blocks in the audio carrier S by the step 4 or the step 8 to obtain a new audio carrier, wherein the obtained new audio carrier is the secret audio S' embedded with the secret information.
During the embedding process, overflow problems may occur due to the modification of the sample values. To solve this problem, we build a location mapping table L M Marking blocks of samples that may cause overflow, since the 16bit quantized audio sample value range is [ -32768,32767]Thus, if there are sample values in the current block equal to-32768 or 32767, let L M (i) =1, otherwise L M (i) =0; after the embedding is finished, the obtained position mapping table L M Lossless compression is carried out to obtain L S . In practice, after the chunk size is determined, L M Is also uniquely determined. The compressed position diagram L S The method comprises the steps of sequentially connecting and combining the auxiliary information L with a block size n, a complexity threshold v, a secret communication length n and an auxiliary information end mark EOS, and embedding the auxiliary information L into the beginning part of audio in an LSB replacement mode. At the same time, in order to ensure reversibility, the part replaced by the auxiliary information L is embedded in the carrier together with the secret information.
The secret information extraction method is an inverse procedure of the steganography method, and in order to extract secret information from the audio encrypted by the reversible steganography method, after receiving the audio containing the secret information, the auxiliary information at the beginning is extracted according to the auxiliary information end mark EOS. Thus, the position mapping table L is obtained in turn S Dividing the block size into n and the cipher message length into n, and adding L S Decompressing to obtain an original position mapping table L M The method comprises the steps of carrying out a first treatment on the surface of the The secret information extraction method corresponding to the steganography method is used for extracting secret information from the secret-containing audio S ', wherein S ' = { x ' 1 ,x′ 2 ...,x′ N }，x′ 1 Is the 1 st sampling value, x ' in the audio S ' with the secret ' 2 Is the 2 nd sampling value, x ' in the audio S ' with the secret ' N Is the nth sampling value in the encrypted audio S'; the method comprises the following steps:
step 1, dividing the audio S' containing the secret into sampling blocks with the length of n, wherein the length of each sampling block is the same as the length of each sub-block in audio steganography, and the dividing is as followsThe latter dense-containing audio S '= { x' 1 ,S′ 1 ,x′ n+2 ,S′ 2 ,x′ 2n+3 ,...S′ i ...,x′ N Expression S 'of the ith sample block' i ＝{x′ i+n*(i-1)+1 ,x′ i+n*(i-1)+2 ...,x′ i+n*i }，Representing a downward rounding; and the ith sample block S' i The immediately preceding sample value is x' i+n*(i-1) And the ith sampling block S' i The next subsequent sample value is x' i+n*i+1 ；
Step 2, respectively carrying out ascending arrangement on sampling values in each sampling block to obtain ascending sampling blocks; wherein the ith sample block S' i Corresponding ascending sampling block X' i The expression is: x'. i ＝{x′ σ(1) ,x′ σ(2) ,...,x′ σ(n) }，x′ σ(1) ≤x′ σ(2) ≤...≤x′ σ(n) The method comprises the steps of carrying out a first treatment on the surface of the Sigma (i) is the ascending sample block X' i Middle sampling value x' σ(i) Corresponding to the original sampling block S' i Is a position in the middle; sigma' (i) ∈1, n]；
Step 3, obtaining an ith sampling block S 'according to the preset complexity level T same as that of the audio steganography' i Corresponding ascending sampling block X' i Sample value set { x } 'for computational complexity in the interior' σ(T+2) ,x′ σ(T+3) ,...x′ σ(n-T-1) And calculate the ith sample block X' i The complexity delta' of (a) is calculated as:
wherein μ 'is x' σ(T+2) ,x′ σ(T+3) ...x′ σ(n-T-1) 、x′ i+n*(i-1) And x' i+n*i+1 The average value corresponding to all sampling values;
since the sampling value for calculating complexity is kept unchanged before and after embedding, delta=delta', ensuring that no error occurs during decryption;
step 4, judging the i-th sampling block X 'calculated in the step 3' i If the complexity delta 'of (a) is greater than the same preset threshold v as when the audio is steganographic, if so, then the current up-sampled block X' i For complex blocks, the ith sample block S ' in the dense audio S ' is sampled ' i Continuously maintaining the original value of the sampling value, and transferring to the step 8; if not, then the current up-sample block X' i Is a smooth block and is transferred to the step 5; the initial value of i is 1;
step 5, obtaining the current ascending sampling block X 'by using the same actual complexity level k as the audio steganography' i Is equal to the optimal predicted sample value x' σ(k+1) And x' σ(n-k) ；
Step 6, respectively calculating optimal predicted sampling values x' σ(k+1) And { x' σ(1) ,x′ σ(2) ...x′ σ(k) Error between each sample value in the sequence, k prediction errors PE 'are obtained' min The method comprises the steps of carrying out a first treatment on the surface of the And respectively calculating optimal predicted sampling value x' σ(n-k) And { x' σ(n-k+1) ,x′ σ(n -k+2)...x′ σ(n) Error between each sample value in the sequence, k prediction errors PE 'are obtained' max ；
Step 7, judging the current ascending sampling block X 'in the step 3' i Whether the complexity delta' of (a) is greater thanIf so, then for { x' σ(1) ,x′ σ(2) ...x′ σ(k) In } and x' σ(k+1) Sampling values with prediction error values of D and D+1 and { x' σ(n-k+1) ,x′ σ(n -k+2)...x′ σ(n) In } and x' σ(n-k) And carrying out secret information extraction on sampling values with the prediction error value of D and D+1, wherein the extracted secret information b has a calculation formula as follows:
b＝PE′ min -D,ifPE′ min ∈{D,D+1}
b＝PE′ max -D,ifPE′ max ∈{D,D+1}；
if not, then for { x' σ(1) ,x′ σ(2) ...x′ σ(k) Intermediate and optimal predicted sample value x' σ(k+1) Sampling values with prediction error values of-1, 0,1 and 2, { x' σ(n-k+1) ,x′ σ(n -k+2)...x′ σ(n) In } and x' σ(n-k) The method comprises the steps that secret information extraction is carried out on sampling values with prediction error values of-1, 0,1 and 2, wherein D is the same value as that of audio steganography; the calculation formula for extracting the secret information b is as follows:
step 8, judging whether all secret information is extracted at the moment, if yes, turning to step 9; if not, making i=i+1, and turning to the step 4;
and 9, forming the extracted secret information in each sampling block into complete secret information according to the sequence of the sampling blocks, and obtaining the secret information extracted from the secret-containing audio S'.
The above only describes the process of extracting the secret information from the secret-containing audio, and of course, the audio carrier can be recovered from the secret-containing audio S' in the process of extracting the secret information, and the audio carrier is recovered correspondingly when the secret information is extracted from the sampling values with the prediction error values of D and d+1 in step 7 of the secret information extraction method, where the calculation formula is as follows:m∈[1,k]；
m∈[n-k+1,n]；
when secret information extraction is carried out on sampling values with prediction error values of-1, 0,1 and 2, the corresponding audio carrier is recovered, and the calculation formula is as follows:
q∈[1,k]；
the invention embeds the secret information bit into the prediction error which meets the condition, and shifts the prediction error which does not meet the requirement by adding one or subtracting one so as to ensure the accuracy of decryption. Therefore, the method slightly modifies the audio carrier, accords with the characteristic of high fidelity and can ensure the perception quality of the audio; in addition, the maximum value of the quantity is set as the prediction error value D in all the prediction error values, so that secret information is embedded into the sampling value with higher similarity in the method, and the steganography effect is improved while the embedding capacity is improved.
Claims (4)
1. An audio reversible steganography method is used for embedding secret information into an audio carrier S with the length of N to obtain a secret-containing audio S'; wherein s= { x 1 ,x 2 ，...,x N }，x 1 For sample 1 in the audio carrier S, x 2 For sample 2 in the audio carrier S, x N Is the nth sample value in the audio carrier S; the method is characterized in that: the method comprises the following steps:
step 1, dividing an audio carrier S with the length of N into sub-blocks with the length of N; wherein the audio carrier s= { x after division 1 ,S 1 ,x n+2 ,S 2 ,x 2n+3 ,...S i ...,x N Expression S of the ith sub-block i ＝{x i+n*(i-1)+1 ,x i+n*(i-1)+2 ，...,x i+n*i }， Representing a downward rounding; and the ith subblock S i The next previous sample value is x i+n*(i-1) And the ith subblock S i The next sample value is x i+n*i+1 ；
Step 2, respectively carrying out ascending arrangement on sampling values in each sub-block to obtain ascending sub-blocks; wherein the ith sub-block S i Corresponding ascending sub-block X i The expression is: x is X i ＝{x σ(1) ,x σ(2) ,...,x σ(n) }，x σ(1) ≤x σ(2) ≤...≤x σ(n) The method comprises the steps of carrying out a first treatment on the surface of the Sigma (i) is the ascending sub-block X i Middle sampling value x σ(i) Corresponding to the atomic block S i Is a position in the middle; sigma (i) ∈1, n]；
Step 3, obtaining an ith subblock S according to the preset complexity level T, wherein T is a positive integer i Corresponding ascending sub-block X i Sample value set { x for computational complexity in the interior σ(T+2) ,x σ(T+3) ,...x σ(n-T-1) And calculate and get the ith ascending sub-block X i The complexity delta of (2) is calculated as:
wherein μ is x σ(T+2) x σ(T+3) ...x σ(n-T-1) 、x i+n*(i-1) And x i+n*i+1 The average value corresponding to all sampling values;
step 4, judging the ith ascending sub-block X calculated in the step 3 i If the complexity delta of (a) is greater than the preset threshold v, if so, then the current ascending sub-block X i For complex blocks, the ith sub-block S in the audio carrier S i Continuously maintaining the original value of the sampling value, and transferring to the step 9; if not, then the current ascending sub-block X i Is a smooth block and is transferred to the step 5; the initial value of i is 1;
step 5, calculating an actual complexity level k meeting the following formula, wherein k is a positive integer;
and correspondingly obtaining the current ascending sub-block X according to the actual complexity level k i In the optimal predicted sample value x σ(k+1) And x σ(n-k) ；
Step 6, calculating optimal predicted sampling value x respectively σ(k+1) And { x } σ(1) ,x σ(2) ...x σ(k) Error between each sampling value in the sequence to obtain k prediction errors PE min The method comprises the steps of carrying out a first treatment on the surface of the Respectively calculating optimal predicted sampling value x σ(n-k) And { x } σ(n-k+1) ,x σ(n-k+2) ...x σ(n) Error between each sampling value in the sequence to obtain k prediction errors PE max ；
Step 7, judging the current ascending sub-block X in the step 3 i Whether the complexity delta of (a) is greater thanIf yes, then for { x } σ(1) ,x σ(2) ...x σ(k) Intermediate and optimal predicted sample value x σ(k+1) Sampling value with prediction error value D and { x }, between σ(n-k+1) ,x σ(n-k+2) ...x σ(n) Intermediate and x σ(n-k) Performing embedding operation on sampling values with the prediction error value D, and performing shifting operation on sampling values with the prediction error value greater than D, wherein D is a natural number greater than 1; and the new sampling value obtained after the embedding operation and the shifting operation is used for replacing the original sampling value, and the sampling value with the prediction error value smaller than D is continuously kept at the original value to obtain { x } σ(1) ,x σ(2) ...x σ(k) New sample value { x } 'corresponding to' σ(1) ,x′ σ(2) ...x′ σ(k) { x } σ(n-k+1) ,x σ(n-k+2) ...x σ(n) New sample value { x } 'corresponding to' σ(n-k+1) ,x′ σ(n-k+2) ...x′ σ(n) -and go to step 8; the calculation formulas of the embedding operation and the shifting operation are as follows:
if not, then for { x } σ(1) ,x σ(2) ...x σ(k) Intermediate and optimal predicted sample value x σ(k+1) Sampling values with prediction error values of 0 and 1 and { x }, respectively σ(n-k+1) ,x σ(n-k+2) ...x σ(n) Intermediate and x σ(n-k) Performing embedding operation on sampling values with prediction error values of 0 and 1, performing shifting operation on sampling values with prediction error values smaller than 0 or larger than 1, and replacing the original sampling values with new sampling values obtained after the embedding operation and the shifting operation to obtain { x } σ(1) ,x σ(2) ...x σ(k) New sample value { x } 'corresponding to' σ(1) ,x′ σ(2) ...x′ σ(k) { x } σ(n-k+1) ,x σ(n-k+2) ...x σ(n) New sample value { x } 'corresponding to' σ(n-k+1) ,x′ σ(n-k+2) ...x′ σ(n) -a }; and go to step 8; the calculation formulas of the embedding operation and the shifting operation are as follows:
step 8, the { x } 'obtained in the step 7' σ(1) ,x′ σ(2) ...x′ σ(k) Sum { x' σ(n-k+1) ,x′ σ(n-k+2) ...x′ σ(n) ' current sub-block S i { x in } σ(k+1) ,x σ(k+2) ...x σ(n-k) Respectively according to the current sub-block S i Is the original order of replacement of the ith S in the audio carrier S i Sub-block corresponding bitSampling values;
step 9, judging whether all secret information is embedded at the moment, if yes, turning to step 10; if not, making i=i+1, and turning to the step 4, and continuing to perform embedding judgment on the next sub-block;
step 10, replacing the sub-blocks in the audio carrier S by the step 4 or the step 8 to obtain a new audio carrier, wherein the obtained new audio carrier is the secret audio S' embedded with the secret information.
2. The audio reversible steganography method of claim 1, characterized in that: PE in the step 6 min The calculation formula of (2) is as follows:
PE min ＝x s -x t
PE max the calculation formula of (2) is as follows:
PE＝x-x maxuv
3. the audio reversible steganography method of claim 1, characterized in that: the method for determining the numerical value D in the step 7 is as follows: firstly, presetting the value range of D as [ D ] min ,D max ]Counting k prediction errors PE in step 6 min And k prediction errors PE max The prediction error values of the two are respectively D min To D max And taking the prediction error value corresponding to the value with the largest number as D.
4. A secret information extraction method is used for extracting secret information from a secret-containing audio S ', wherein S ' = { x ' 1 ,x′ 2 ...,x′ N }，x′ 1 Is the 1 st sampling value, x ' in the audio S ' with the secret ' 2 Is the 2 nd sampling value, x ' in the audio S ' with the secret ' N Is the nth sampling value in the encrypted audio S'; the method is characterized in that: the method comprises the following steps:
step 1, dividing the encrypted audio S ' into sampling blocks with the length of n, wherein the length of each sampling block is the same as the length of each sub-block during audio steganography, and the divided encrypted audio S ' = { x ' 1 ,S′ 1 ,x′ n+2 ,S′ 2 ,x′ 2n+3 ,...S′ i ...,x′ N Expression S 'of the ith sample block' i ＝{x′ i+n*(i-1)+1 ,x′ i+n*(i-1)+2 ...,x′ i+n*i }， Representing a downward rounding; and the ith sample block S' i The immediately preceding sample value is x' i+n*(i-1) And the ith sampling block S' i The next subsequent sample value is x' i+n*i+1 ；
Step 2, respectively carrying out ascending arrangement on sampling values in each sampling block to obtain ascending sampling blocks; wherein the ith sample block S' i Corresponding ascending sampling block X' i The expression is: x'. i ＝{x′ σ(1) ,x′ σ(2) ,...,x′ σ(n) }，x′ σ(1) ≤x′ σ(2) ≤...≤x′ σ(n) The method comprises the steps of carrying out a first treatment on the surface of the Sigma (i) is the ascending sample block X' i Middle sampling value x' σ(i) Corresponding to the original sampling block S' i Is a position in the middle; sigma' (i) ∈1, n]；
Step 3, obtaining an ith sampling block S 'according to the preset complexity level T same as that of the audio steganography' i Corresponding ascending sampling block X' i Sample value set { x } 'for computational complexity in the interior' σ(T+2) ,x′ σ(T+3) ,...x′ σ(n-T-1) And calculateObtaining the ith sampling block X' i The complexity delta' of (a) is calculated as:
wherein μ 'is x' σ(T+2) ,x′ σ(T+3) ...x′ σ(n-T-1) 、x′ i+n*(i-1) And x' i+n*i+1 The average value corresponding to all sampling values;
step 4, judging the i-th sampling block X 'calculated in the step 3' i If the complexity delta 'of (a) is greater than the same preset threshold v as when the audio is steganographic, if so, then the current up-sampled block X' i For complex blocks, the ith sample block S ' in the dense audio S ' is sampled ' i Continuously maintaining the original value of the sampling value, and transferring to the step 8; if not, then the current up-sample block X' i Is a smooth block and is transferred to the step 5; the initial value of i is 1;
step 5, obtaining the current ascending sampling block X 'by using the same actual complexity level k as the audio steganography' i Is equal to the optimal predicted sample value x' σ(k+1) And x' σ(n-k) ；
Step 6, respectively calculating optimal predicted sampling values x' σ(k+1) And { x' σ(1) ,x′ σ(2) ...x′ σ(k) Error between each sample value in the sequence, k prediction errors PE 'are obtained' min The method comprises the steps of carrying out a first treatment on the surface of the And respectively calculating optimal predicted sampling value x' σ(n-k) And { x' σ(n-k+1) ,x′ σ(n-k+2) ...x′ σ(n) Error between each sample value in the sequence, k prediction errors PE 'are obtained' max ；
Step 7, judging the current ascending sampling block X 'in the step 3' i Whether the complexity delta' of (a) is greater thanIf so, then for { x' σ(1) ,x′ σ(2) ...x′ σ(k) In } and x' σ(k+1) Sampling value with prediction error value of D and D+1{x′ σ(n-k+1) ,x′ σ(n-k+2) ...x′ σ(n) In } and x' σ(n-k) And carrying out secret information extraction on sampling values with the prediction error value of D and D+1, wherein the extracted secret information b has a calculation formula as follows:
b＝PE′ min -D,ifPE′ min ∈{D,D+1}
b＝PE′ max -D,ifPE′ max ∈{D,D+1}；
if not, then for { x' σ(1) ,x′ σ(2) ...x′ σ(k) Intermediate and optimal predicted sample value x' σ(k+1) Sampling values with prediction error values of-1, 0,1 and 2, { x' σ(n-k+1) ,x′ σ(n-k+2) ...x′ σ(n) In } and x' σ(n-k) The method comprises the steps that secret information extraction is carried out on sampling values with prediction error values of-1, 0,1 and 2, wherein D is the same value as that of audio steganography; the calculation formula for extracting the secret information b is as follows:
step 8, judging whether all secret information is extracted at the moment, if yes, turning to step 9; if not, making i=i+1, and turning to the step 4;
and 9, forming the extracted secret information in each sampling block into complete secret information according to the sequence of the sampling blocks, and obtaining the secret information extracted from the secret-containing audio S'.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010290729.2A CN111640443B (en) | 2020-04-14 | 2020-04-14 | Audio reversible steganography method and secret information extraction method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010290729.2A CN111640443B (en) | 2020-04-14 | 2020-04-14 | Audio reversible steganography method and secret information extraction method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN111640443A CN111640443A (en) | 2020-09-08 |
| CN111640443B true CN111640443B (en) | 2023-09-26 |
Family
ID=72333153
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010290729.2A Active CN111640443B (en) | 2020-04-14 | 2020-04-14 | Audio reversible steganography method and secret information extraction method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111640443B (en) |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106448686A (en) * | 2016-08-10 | 2017-02-22 | 中山大学 | Self-adapting audio frequency spatial domain steganography method |
| CN108766449A (en) * | 2018-05-30 | 2018-11-06 | 中国科学技术大学 | A kind of reversible water mark implementation method of audio signal |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US9396733B2 (en) * | 2014-05-06 | 2016-07-19 | University Of Macau | Reversible audio data hiding |
-
2020
- 2020-04-14 CN CN202010290729.2A patent/CN111640443B/en active Active
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106448686A (en) * | 2016-08-10 | 2017-02-22 | 中山大学 | Self-adapting audio frequency spatial domain steganography method |
| CN108766449A (en) * | 2018-05-30 | 2018-11-06 | 中国科学技术大学 | A kind of reversible water mark implementation method of audio signal |
Non-Patent Citations (1)
| Title |
|---|
| 余恒等.基于采样值排序的音频可逆隐写算法.《计算机工程》.2020,全文. * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN111640443A (en) | 2020-09-08 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Weng et al. | Reversible watermarking based on invariability and adjustment on pixel pairs | |
| Lee et al. | High-payload image hiding with quality recovery using tri-way pixel-value differencing | |
| KR100880039B1 (en) | Method and system for achieving coding gains in wavelet-based image codecs | |
| Hong | An efficient prediction-and-shifting embedding technique for high quality reversible data hiding | |
| CN101105857A (en) | High capacity reversible water mark method based on predication and companding technology | |
| Kumar et al. | A reversible data hiding scheme using pixel location. | |
| CN112714231A (en) | Robust steganography method based on DCT (discrete cosine transformation) symbol replacement | |
| CN112070647B (en) | Reversible high dynamic range image information hiding method | |
| CN111464718B (en) | Encryption image reversible information hiding method based on interpolation technology | |
| Wang et al. | An efficient reversible data hiding scheme using prediction and optimal side information selection | |
| CN111640443B (en) | Audio reversible steganography method and secret information extraction method | |
| Chu et al. | A digital image watermarking method based on labeled bisecting clustering algorithm | |
| CN110910299B (en) | Self-adaptive reversible information hiding method based on integer wavelet transform | |
| Li et al. | Data hiding in images by adaptive LSB substitution based on the pixel-value differencing | |
| CN112714230A (en) | Robust video steganography method and device based on audio side channel | |
| CN113099068B (en) | High-capacity pixel value sorting reversible information hiding method based on double images | |
| CN111816195B (en) | Audio reversible steganography method, secret information extraction and carrier audio recovery method | |
| Weng et al. | Invariability of Remainder Based Reversible Watermarking. | |
| CN108200435B (en) | Reversible information hiding method for fractal image | |
| Yang et al. | Reversible data hiding for color BMP image based on block difference histogram | |
| Alavianmehr et al. | A reversible data hiding scheme for video robust against H. 264/AVC compression | |
| Wang et al. | A novel block sorting scheme for reversible data hiding in JPEG images | |
| CN108009974B (en) | Robust reversible watermarking method resisting JPEG compression and digital television broadcasting system | |
| Ou et al. | A high payload histogram-based reversible wartermarking using linear prediction | |
| Weng et al. | Lossless data hiding based on prediction-error adjustment |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |