CN106169189B - STC coding method based on embedded quantity bilayer Optimal Ratio - Google Patents
STC coding method based on embedded quantity bilayer Optimal Ratio Download PDFInfo
- Publication number
- CN106169189B CN106169189B CN201610471816.1A CN201610471816A CN106169189B CN 106169189 B CN106169189 B CN 106169189B CN 201610471816 A CN201610471816 A CN 201610471816A CN 106169189 B CN106169189 B CN 106169189B
- Authority
- CN
- China
- Prior art keywords
- vector
- embedded
- cost
- carrier
- secret information
- 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.)
- Expired - Fee Related
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T9/00—Image coding
Abstract
The present invention relates to a kind of STC coding methods based on embedded quantity bilayer Optimal Ratio.Embedding information is embedded into the lowest bit position and time low bit position of carrier image by this method in an optimal manner according to message sink coding, equally extracts secret information from the two private communication channels during the extraction process.During to amending image, other than carrying out ± 1 change to pixel, allow partial pixel to carry out ± 2 change, can greatly reduce insertion modification points, the safety after guaranteeing steganography insertion.This method not only increases Embedding efficiency, and can be reasonably resistant to the existing attack based on high dimensional feature steganalysis method, is particularly suitable for carrying out the covert communications of high security.
Description
Technical field
The invention belongs to information security field, it is related in the adaptive Steganography of spatial domain picture based on two layers of most optimal sorting of embedded quantity
Application of the grid code encoding method and this method of proportion in covert communications.
Background technique
Steganography is the important branch of Information Hiding Techniques in information security field, and how Steganography will be secret if mainly being studied
The carriers such as information embedded images, audio, video are to achieve the purpose that covert communications.As the countermove of Steganography, steganography point
Analysis uses the methods of machine learning, pattern-recognition then to distinguish general carrier and the file Jing Guo steganography.
Generally, when designing steganographic algorithm, need to comprehensively consider the following aspects factor:
1) imperceptibility: it is required that secret information is embedded in digital carrier, the subjective quality without will affect original vector,
Observed person is not easy to discover;
2) robustness: it is required that hide after Digital Media in transmittance process, although by it is multiple be not intended to or intentional letter
Number processing, but remains to be restored secret information;
3) hidden capacity: it is required that under the premise of guaranteeing imperceptibility and robustness to a certain degree, it should as more as possible
Ground is embedded in secret information into carrier;
4) safety: it is required that even if possessing the priori knowledge to the steganographic algorithm, existing steganalysis method can not yet
With sufficiently high accuracy differentiation general carrier and the file Jing Guo the algorithm steganography.
The medium with wide-scale distribution in network is widely used as people in daily life, image has as steganography carrier
The incomparable advantage of other bearer documents.Image latent writing art at home and abroad under the common concern of experts and scholars, has considerable
Progress and development.Some algorithms applied in image latent writing at the same time can also obtain effectively in other carriers
Application.
Image latent writing is divided into two kinds according to carrier type, airspace steganography and JPEG domain steganography.Wherein airspace steganography is repaired
The object changed is the pixel of image, and the object that JPEG domain steganography is modified is the DCT coefficient of jpeg image.But due to JPEG
Integrally visual disturbance is larger to image for DC coefficient and the modification of AC zero coefficient in the DCT coefficient of image, generally to jpeg image
Steganography do not consider DC coefficient and AC zero coefficient.Therefore for JPEG domain steganography, the embedding capacity of airspace steganography is higher.
Image latent writing is roughly divided into two stages according to the time, with improve traditional Steganography stage based on Embedding efficiency and
In conjunction with the adaptive Steganography stage of picture material.
Secret information is mainly embedded into the lowest bit position of image pixel or DCT coefficient by traditional Steganography, and to the greatest extent may be used
It can ground raising Embedding efficiency.It is visually not variant with carrier image using image embedded by traditional Steganography, but without
Method resists the attack based on high-dimensional feature steganalysis method.
The steganographic algorithm of current main-stream is the adaptive Steganography in conjunction with picture material.Adaptive Steganography be specifically divided into
Lower two parts:
1) cost calculating section: for the texture for preferably protruding image, image pixel modification influences image bring
It is indicated in the form of numerical value cost, such as HUGO, UNIWARD, HILL etc..From the point of view of analysis, image cost is actually hidden
The embodiment of characteristic value in analysis is write, adaptive Steganography resists attacking for steganalysis by reducing the disturbance to these characteristic values
It hits.For example, the filtering core that HUGO is used when calculating cost is the Spam submodel in Rich Model, and HILL is calculating generation
The filtering core used when valence is the KB submodel in Rich Model.
2) coded portion: being effectively embedding the lesser position of cost value for preembedded information, and guarantees that insertion is repaired
The cost changed the time is cumulative and minimum, such as Wet Paper Code (WPC), Syndrome-Trellis Codes (STCs).Currently
The encoding scheme generally used is STCs.
Traditional Steganography has been studied people due to can not preferably resist the attack based on high-dimensional feature steganalysis
Member is abandoned.Adaptive Steganography is accepted by preferable safety by domestic and international experts and scholars.About adaptive Steganography,
It is studied relative to cost function like a raging fire, coded portion is by the effect that construction process is excessively complicated and STCs is realized
It is optimal already close to theory, it stagnated always about the research of encoding context both at home and abroad in recent years.And the coding of better performances
It can be described as the basis of cost function research, therefore the research about coding needs to cause enough attention.
Summary of the invention
The object of the present invention is to provide a kind of STC coding methods based on embedded quantity bilayer Optimal Ratio, can
Embedding information is embedded into the lowest bit position and time low bit position of carrier image in an optimal manner according to message sink coding, no
Embedding efficiency is improved only, and the existing attack based on high dimensional feature steganalysis method can be reasonably resistant to.
The present invention is encoded compared to original STCs, using complete on image lowest bit position and time low bit position information distribution ratio
The new more accurate method of salary distribution, during to amending image, other than carrying out ± 1 change to pixel, to part picture
Element allows to carry out ± 2 change.Insertion modification points, the safety after guaranteeing steganography insertion can greatly be reduced.As it can be seen that this hair
The encoding scheme of bright proposition is different from original STCs coding, is particularly suitable for carrying out the covert communications of high security.
The present invention uses the embedding strategy of current image adaptive Steganography generality in telescopiny, and specific difference exists
The scheme used by coding module.Secret information can be embedded into image lowest bit position and put down by the present invention in telescopiny
Face and time low bit two private communication channels of bit plane extract secret letter during the extraction process equally from the two private communication channels
Breath.
The technical solution adopted by the present invention is based under image adaptive Steganography frame, secret in image adaptive Steganography
The telescopiny of information is as shown in Figure 1, include the following steps (coding mode is by taking single layer STC as an example):
1) image preprocessing: under image present channel, image present channel is expressed as vector x, by vector x with 8
Binary form is indicated, and carrier lowest bit position is taken out, and is expressed as vector x1;
2) cost calculates: carrying out generation according to pre-set cost calculation to vector x obtained in step 1)
Valence calculates, and obtained cost value is expressed as vector ρ;
3) information is embedded in advance: secret information m is embedded into carrier vector x using single layer STC coding1In, so that Η1x1'=
M, wherein vector x1' indicate the carrier after embedding information, Η1For the parity matrix generated according to shared parameter;Single layer STC
It can guarantee to be embedded in and minimum cumulative to the cost of carrier modifying point while embedding information, i.e.,
4) it is embedded in secret information: vector feedback to be modified in step 3) is modified into carrier, after guaranteeing modification
Pixel value in the range of 0~255;Such as pixel value, if it is 255, obtaining 256 to its modification for carrying out+1 can be more than picture
The upper limit of element, can carry out -1 modification to it in this case, and obtaining pixel value is 254, equally can satisfy embedding information
Demand;
5) to other channels of image, the insertion of secret information is all carried out according to the sequence of step 1) to step 4), most
It obtains being embedded in modified carrier image eventually.
Further, the above method further includes the extraction process of secret information, as shown in Figure 2, comprising the following steps:
1) image preprocessing: under image present channel, image present channel is expressed as vector x, by vector x with 8
Binary form is indicated, and carrier lowest bit position is taken out, and is expressed as vector x1;
2) carry out information extraction according to embedded coding mode: the parameter according to shared transmitting generates parity matrix Η1,
Utilize formula Η1x1=m extracts secret information contained in carrier;
3) under other channels of image, all secret informations are extracted according to the sequence of step 1) to step 4), so
All information is combined to the message being embedded in afterwards.
In the above scheme, the present invention is embedded in the more original embedded mode of module to information in advance further improvement.Tool
Body can be divided into two o'clock, and first, present invention employs the allocation strategies based on two layers of Optimal Ratio of embedded quantity, by secret information point
It is fitted on the lowest bit bit plane and time low bit bit plane of carrier vector;Second, present invention employs the insertion of optimization modifications
Strategy, comprehensive three classes insertion modification strategy, so that reaching minimum to the disturbance of image.Wherein it is based on two layers of optimum allocation of embedded quantity
The allocation strategy of ratio is used and is derived from basic binary system insertion, determines that each layer of private communication channel modifying point is then
Using single layer STC.It is summarized as follows about the two:
It is closely related between embedding information length and cost in being actually embedded in, and this relationship is specifically by binary system
Insertion is described.It is assumed herein that known preembedded information m and by the calculated cost ρ of associated cost function.Then according to letter
The relevant knowledge of source code can obtain following relevant information.
βi,jFor in the position (i, j) cost ρi,jUnder modification probability, λ is unknown parameter, and occurrence combines subsequent formula to ask
Solution.Under conditions of known each pixel modification cost and modification probability, theoretic overall disturbance can be further shown that
D:
In the case where knowing each pixel modification Probability Condition, it can be shown that pre- embedding information length | m |:
Wherein H (βi,j) indicate in βi,jUnder comentropy, can be obtained by the following formula:
H(βi,j)=- βi,jlog(βi,j)-(1-βi,j)log(1-βi,j), (4)
(1), (4) relevant parameter, which are replaced and are updated in formula (3), to obtain following formula:
For actual telescopiny, embedding information length | m | and all by the calculated cost value ρ of associated cost function
Know only have a unknown parameter λ in formula (5).To formula (5), further analysis can be obtained, | m | with the increase monotone decreasing of λ, institute
By can by Binary searches in a manner of (5) are solved.After obtaining λ, so far listed formula can be solved.
Single layer STC coding is a kind of specific implementation form of binary system insertion, and it is long that the result of insertion meets embedding information
Degree | m | and the relationship of cost ρ.And single layer STC enables to modified carrier when secret information m is embedded into carrier x
X' and embedding information m meet Η x'=m, and wherein Η is by shared parameter parity matrix generated.At the same time, STC
Enable to whole insertion cost minimum.
Pre- embedding information m is divided into two parts according to the allocation strategy based on two layers of Optimal Ratio of embedded quantity by the present invention
m1And m2, and using the modification strategy that disturbance is minimum in actual modification.Distribution wherein based on two layers of Optimal Ratio of embedded quantity
Tactful principle is as follows:
It is assumed that pre- embedding information m is divided into two parts m1And m2.The lowest bit bit plane of known image carrier x to
Measure x1With secondary low bit bit plane vector x2, it is ρ by the calculated cost value of associated cost function institute.To lowest bit bit plane
Vector x1In conjunction with cost ρ embedding information m1When, every corresponding insertion modification probability is as follows:
Wherein λ1For unknown parameter, occurrence combines subsequent equations.It will here according to every insertion modification probability
Every change is defined as a stochastic variable ci,j, and the modification situation of stochastic variable obeys two Distribution values:
It according to every modification probability, can be derived that ideally, insertion modification points N1.Energy is combined with formula (5)
Enough by embedding information length m1It further indicates that.
After the completion of the insertion of first layer private communication channel, insertion modification point set C can be obtained1,|C1|=E (N1).In order to make
The insertion of two channels can be completed by ± 1 modification, the present invention is by the cost of first layer private communication channel insertion modifying point the
It is set as 0 before two layers of private communication channel insertion, obtains new cost ρ '.When being embedded in second layer private communication channel, second layer insertion letter
Cease length | m2| it can be indicated according to the form of formula (5), ideally, the point of the second layer can have according to whether in first layer
Modification is divided into C1WithTwo set.Then | m2| it can further segment expression:
Due to ρ 'i,j=0,It carries it into formula (11) replacement and abbreviation obtains:
Ideally, in order to realizing the insertions of two private communication channels by ± 1, where it is desirable to whole embedding of the second layer
Enter information and is embedded into set C1In, i.e., the embedding information of latter is 0 in requirement (12).Because generally speaking latter is with λ2
Increase successively decrease, it is possible to pass through setting λ2→ ∞ is realized.Therefore ideal second layer embedding information amount | m2| it can be as
Lower expression:
|m2|=| C1|。 (13)
Overall embedding information amount | m | can be by | m1| and | m2| it indicates, wherein | m1| it can be indicated by (10), and | m2| foundation
(13) can be by | C1| replacement, and then can be indicated with (9).
Unique unknown parameter is λ in above formula1, can be obtained by derivation analysis | m | with λ1Increase and reduce, therefore can
Formula (14) is solved in a manner of through Binary searches.And the λ found out1It is it is assumed that second layer private communication channel modifying point is complete
Concentrate on entirely it is ideally obtained in first layer private communication channel modifying point, at the same time when being integrated in practice,
Possess correct implementation strategy.Obtaining λ1Afterwards, first layer private communication channel can be obtained ideally by carrying it into formula (10)
Embedding information amount, and then pass through | m2|=| m |-| m1| obtain the embedding information amount of second layer private communication channel ideally.
However, when being embedded in using single layer STC to each private communication channel, since single layer STC does not reach theoretical
Value, thus it is inevitably more higher than theoretical value for the Section 2 in (14), therefore a tune is introduced in actual use
Integral divisor τ.Expression formula is solved used by being actually embedded in are as follows:
Wherein τ is a value close to 1, and best performance value can be obtained by experiment.
Two layers of optimum allocation implementation strategy of embedded quantity proposed by the present invention, as shown in figure 3, detailed process includes following
Step:
1) parameter inputs: total embedding information length | m |, cost ρ;
2) overturning λ is calculated1: by total embedding information length | m | and insertion cost ρ brings formula (15) into, using Binary searches
Mode solves formula (15), obtains λ1;
3) the corresponding embedding information length of two layers of private communication channel is calculated: by insertion cost ρ and overturning λ1Bring formula (10) into,
Calculate first layer private communication channel and be correspondingly embedded in message length | m1|, and then pass through | m2|=| m |-| m1| calculate the hidden letter of the second layer
The corresponding embedding information length in road | m2|。
Grid code encoding method proposed by the present invention based on two layers of Optimal Ratio of embedded quantity, specific telescopiny packet
Include following steps (unless otherwise specified, following steps have computer execution):
1) image preprocessing: under image present channel, being expressed as vector x for image present channel, by carrier vector x with
8 binary forms indicate, and lowest bit bit plane and time low bit bit plane are taken out respectively, are expressed as vector x1With
Vector x2;
2) cost calculates: carrying out generation according to pre-set cost calculation to vector x obtained in step 1)
Valence calculates, and obtained cost value is expressed as vector ρ;
3) information is embedded in advance: secret information is determined according to cost vector ρ to carrier vector x1And x2Modification, it is specific really
Determining process includes the following steps:
A) one or two layers of private communication channel information distribution ratio are determined: being incited somebody to action according to the step of two layers of optimum allocation implementation strategy of embedded quantity
Total secret information m is divided into m1And m2, respectively correspond in carrier vector x1And x2The information of middle insertion;
B) to x1Pre- insertion: by gained vector x in step 1)1As the carrier of first layer private communication channel, according to single layer STC
Working principle, by secret information m1It is that insertion foundation is embedded into carrier vector x with cost vector ρ1In.Guarantee Η simultaneously1x1'=
m1, wherein Η1For the parity matrix generated according to shared parameter, x1' indicate first layer private communication channel insertion secret information after
Carrier vector;
C) to x2Pre- insertion: by gained vector x in step 1)2Carrier as second layer private communication channel.According to embedding in b)
Enter as a result, by x1The cost of the middle point modified is set to zero, obtains new cost vector ρ '.By secret information m2With cost vector
ρ ' is that insertion foundation is embedded into carrier vector x2In.Guarantee Η simultaneously2x2'=m2, wherein Η2For the surprise generated according to shared parameter
Even parity check matrix, x2' indicate the carrier vector after second layer private communication channel insertion secret information;
4) be embedded in secret information: according in 3) to the coincidence situation of first layer and second layer private communication channel modifying point, to figure
The disturbance situation of picture can be divided into three classes, a) x1Modification, x2It does not modify;b)x1It does not modify, x2Modification;c)x1, x2Full modification.It is practical
It can be directly by the lowest bit position of x and time low bit position x in modification1' and x2' replace, but 0 and 255 picture can be made in this way
There is for ± 3 the case where in vegetarian refreshments, therefore the pixel modified for boundary needs to carry out the amendment of opposite direction ± 4;
5) to other channels of image, the insertion of secret information is all carried out according to the sequence of step 1) to step 4).Most
It obtains being embedded in modified carrier image eventually.
Further, the above method further includes the extraction process of secret information, comprising the following steps:
1) image preprocessing: under image present channel, image present channel is expressed as vector x, by carrier vector x
Lowest bit bit plane and time low bit bit plane take out respectively, are expressed as vector x1And vector x2;
2) first layer private communication channel secret information m is extracted1: parity matrix Η is generated according to shared parameter1, by Η1According to
According to formula Η1x1=m1With lowest bit bit vector x1Matrix multiple obtains insertion ciphertext m1;
3) second layer private communication channel secret information m is extracted2: parity matrix Η is generated according to shared parameter2, by Η2According to
According to formula Η2x2=m2With lowest bit bit vector x2Matrix multiple obtains insertion ciphertext m2;
4) information combines: the secret information that step 2) and step 3) extract is combined into insertion ciphertext m;
5) under other channels of image, all secret informations are extracted according to the sequence of step 1) to step 4).
Spatial domain picture grid code encoding method based on two layers of embedded quantity Optimal Ratio of the invention is led in Information hiding
Domain has the beneficial effect that:
1) embedding capacity is big.The coding method that current image adaptive Steganography uses has single layer STC coding, double-deck
STC coding.It is encoded relative to single layer STC, embedding capacity can be greatly improved in the double-deck STC coding and the present invention, so that insertion
Capacity is promoted by about one time compared with single layer STC coding.In the biggish situation of embedding capacity, encoded relative to single layer STC, energy of the present invention
The insertion picture amount needed is set to reduce nearly half, enough so as to reduce covert communications bring risk.It is lower in embedding capacity
In the case where, it is encoded relative to single layer STC, the present invention can reduce the points of the disturbance to image, to guarantee safety.
2) insertion failure rate is reduced.The reason that single layer STC coding is 0.5 due to the load factor upper limit, embedding capacity is limited, when
When further combined with cost, making some images, there is a situation where insertion failures.And the present invention improves embedding capacity, makes it
It is enough preferably to combine cost, to greatly reduce insertion failure rate.
3) safety.The present invention uses the strategy based on two layers of Optimal Ratio of embedded quantity, while allowing the mistake in modification
A few pixels point carries out ± 2 modification in journey, under identical insertion rate, relative to single layer STC coding and bilayer STC coding, originally
Invention greatly reduces insertion modification points, and is embedded in the texture region that modifying point position is more concentrated on image, Jin Ercong
It fundamentally ensure that the safety of adaptive Steganography.At the same time, so that based on the hidden of the adaptive Steganography of spatial domain picture
Communications security is further guaranteed.
4) versatility.The starting point that the present invention designs is the general embedded coding method that design can be combined with cost,
Individually consider there is no certain cost function is directed in the design process, independent place is not done to the format of spatial domain picture yet
Reason, to ensure that preferable versatility.It can be with WOW (the Wavelet Obtained generally popular at this stage
Weights)、SUNIWARD(Spatial Universal Wavelet Relative Distortion)、HILL(High
Low Low) scheduling algorithm wants to combine.
Detailed description of the invention
Fig. 1 is insertion block schematic illustration of the present invention;
Fig. 2 is extraction block schematic illustration of the present invention;
Fig. 3 is two layers of Optimal Ratio strategy schematic diagram of embedded quantity of the invention;
Fig. 4 is secret information insertion flow chart of the invention;
Fig. 5 is that secret information of the invention extracts flow chart;
Fig. 6 is τ value search schematic diagram of the invention;
Fig. 7 is that the present invention is combined with SUNIWARD cost, is shown using the result that SRM and maxSRMd2 feature is attacked
It is intended to;
Fig. 8 is that the present invention is combined with HILL cost, is illustrated using the result that SRM and maxSRMd2 feature is attacked
Figure;
Fig. 9 is that the present invention is combined with WOW cost, the result schematic diagram attacked using SRM and maxSRMd2 feature;
Figure 10 is image binary representation of the present invention and the one or two layer of private communication channel building schematic diagram.
Specific embodiment
The invention will be further described by 4-10 with reference to the accompanying drawing.
Before carrying out the insertion of secret information steganography using the present invention, the data that can be first embedded in needs are encrypted, and are obtained
To random binary data stream.It is proposed by the present invention to use the grid code encoding method based on two layers of Optimal Ratio of embedded quantity
Adaptive Steganography, secret information insertion process as shown in figure 4, concrete operations details is as follows:
1) image preprocessing: under image present channel, being expressed as vector x for image present channel, by carrier vector x with
8 binary forms indicate and take out lowest bit bit plane and time low bit bit plane respectively, are expressed as vector x1With
Vector x2(Figure 10).The wherein numerical value vector that vector x is 0~255, vector x1And vector x2The bit vectors for being 0 and 1 for value;
2) cost calculates: carrying out generation according to pre-set cost calculation to vector x obtained in step 1)
Valence calculates, and obtained cost value is expressed as vector ρ;
3) information is embedded in advance: secret information is determined according to cost vector ρ to carrier vector x1And x2Modification, it is specific really
Determine process include it is several under several steps;
A) one or two layers of private communication channel information distribution ratio are determined: by by embedding information total length | m | and cost vector ρ is brought into
Formula (15) solution obtains insertion overturning λ1, then by λ1Formula (10) is brought into cost vector ρ finds out the insertion of first layer private communication channel
Information content | m1|.Finally according to formula | m2|=| m |-| m1| calculate second layer private communication channel embedding information length | m2|.Specific to
Embedding information sequentially intercepts directly from m | m1| and | m2| the information of length is as one or two layers of private communication channel embedding information m1With
m2;
B) to x1Pre- insertion: by gained vector x in step 1)1It is hidden to first layer as the carrier of first layer private communication channel
Channel application single layer STC coding, by secret information m1It is that insertion foundation is embedded into carrier vector x with cost vector ρ1In, guarantee embedding
It is cumulative and minimum to enter to modify cost.It is at the same time the information for guaranteeing the insertion of normal extraction first layer private communication channel, application
Single layer STC coding, which can make to be embedded in modified carrier in telescopiny, meets Η1x1'=m1, wherein Η1For according to shared ginseng
The parity matrix that number generates, x1' indicate the carrier vector after first layer private communication channel insertion secret information;
C) to x2Pre- insertion: by gained vector x in step 1)2Carrier as second layer private communication channel.According to embedding in b)
Enter two layers of Optimal Ratio derivation process of result and embedded quantity, by x1The cost of the middle point modified is set to zero, obtains new generation
Valence vector ρ '.Second layer private communication channel application single layer STC is encoded, by secret information m2It is insertion according to embedding with cost vector ρ '
Enter to carrier vector x2In, guarantee that insertion modification cost is cumulative and minimum.It is at the same time guarantee to second layer private communication channel energy
Enough normal extraction embedding informations, can make to be embedded in modified carrier in telescopiny and meet Η using single layer STC coding2x2'=
m2, wherein Η2For the parity matrix generated according to shared parameter, x2' indicate the second layer private communication channel insertion secret information after
Carrier vector;
4) it is embedded in secret information: by the lowest bit position of x and time low bit position x1' and x2' replacement, it is 0 He to pixel value
The point that 255 lowest bit position and time low bit position are modified carries out reversed ± 4 amendment;
5) to other channels of image, the insertion of secret information is all carried out according to the sequence of step 1) to step 4).Most
It obtains being embedded in modified carrier image eventually.
The adaptive Steganography of grid code encoding method proposed by the present invention based on two layers of Optimal Ratio of embedded quantity,
Process is extracted as shown in figure 5, concrete operations details is as follows:
1) image preprocessing: under image present channel, being expressed as vector x for image present channel, by carrier vector x with
8 binary forms indicate and take out lowest bit bit plane and time low bit bit plane respectively, are expressed as vector x1With
Vector x2(Figure 10).The wherein numerical value vector that vector x is 0~255, vector x1And vector x2The bit vectors for being 0 and 1 for value;
2) first layer private communication channel secret information m is extracted1: parity matrix Η is generated according to shared parameter1, by Η1According to
According to formula Η1x1=m1With lowest bit bit vector x1Matrix multiple obtains insertion ciphertext m1;
3) second layer private communication channel secret information m is extracted2: parity matrix Η is generated according to shared parameter2, by Η2According to
According to formula Η2x2=m2With lowest bit bit vector x2Matrix multiple obtains insertion ciphertext m2;
4) information combines: the secret information that step 2) and step 3) extract is combined into insertion ciphertext m;
5) under other channels of image, all secret informations are extracted according to the sequence of step 1) to step 4).
Can be seen that the present invention from the above specific embodiment can have preferable logical in conjunction with different Cost Models
With property, the obtained information content distribution condition base of strategy of two layers of Optimal Ratio of embedded quantity is furthermore used for different pictures
This difference, but can equally meet safer Embedded Pointers, therefore the present invention has stronger adaptability.
Before security test, it is thus necessary to determine that specific τ value, specific search result as shown in fig. 6, its abscissa is τ value,
Ordinate is detection error rate, and NDSTCs indicates STC coding method of the invention in figure, and ternary STCs indicates original bilayer
The binary version of STC coding method.As shown in table 1,10000 BOSSbase v1.01 (512 × 512) gray level images are chosen
Carry out the test of safe performance indexes.The cost function wherein combined with coding has HILL, SUNIWARD, WOW algorithm.As peace
The coding of full property comparison has chosen ternary domain the domain (ternary) He Wuyuan (pentary) version of the double-deck STC coding, in addition makees
It is also compared for theoretical optimal simulator (simulator) embedded mode.Insertion rate selection 0.05,0.1,0.2,0.3,
0.4,0.5,0.6 7 insertion rate.Security performance test is attacked using the steganalysis based on high dimensional feature, analyzes result
For OOB, numerical value shows that more greatly safety is higher.Wherein high dimensional feature uses generally by academia is approved and uses SRM
Feature and maxSRMd2 feature.By table 1, it can be seen that the present invention combines the phase under different characteristic analysis with different cost functions
The ternary domain and five yuan of domain versions of double-deck STC coding are all improved, while security performance is already close to theoretical optimal.It is this
Promote the display that can be more clear in figures 7-9.Abscissa is insertion load factor in Fig. 7-9, and ordinate is detection error rate.
Wherein, Fig. 7 is that the present invention is combined with SUNIWARD cost, is illustrated using the result that SRM and maxSRMd2 feature is attacked
Figure;Fig. 8 is that the present invention is combined with HILL cost, the result schematic diagram attacked using SRM and maxSRMd2 feature;Fig. 9
It is that the present invention is combined with WOW cost, the result schematic diagram attacked using SRM and maxSRMd2 feature.
1. security performance evaluating result of table
SRM
The above embodiments are merely illustrative of the technical solutions of the present invention rather than is limited, the ordinary skill of this field
Personnel can be with modification or equivalent replacement of the technical solution of the present invention are made, without departing from the spirit and scope of the present invention, this
The protection scope of invention should subject to the claims.
Claims (6)
1. a kind of STC coding method based on embedded quantity bilayer Optimal Ratio, which is characterized in that will be secret using STC coding
Information is embedded into image, in telescopiny by secret information be assigned to image-carrier vector lowest bit bit plane and time
Two private communication channels of low bit bit plane;Secret information is distributed using the allocation strategy based on two layers of Optimal Ratio of embedded quantity
To the lowest bit bit plane and time low bit bit plane, the allocation strategy based on two layers of Optimal Ratio of embedded quantity
It is:
A) by total embedding information length | m | and insertion cost ρ brings following formula into, is solved using Binary searches mode, obtains λ1;
Wherein, τ is Dynamic gene;ρi,jFor in the insertion cost of position (i, j), i, j indicate transverse and longitudinal coordinate, λ1It indicates by current total
Embedding information length | m | and the intermediate parameter that insertion cost ρ is obtained;
B) cost ρ and λ will be embedded in1Bring following formula into, calculate first layer private communication channel and be correspondingly embedded in message length | m1|, in turn
Pass through | m2|=| m |-| m1| calculate the corresponding embedding information length of second layer private communication channel | m2|;
2. the method as described in claim 1, which is characterized in that when being embedded in secret information, be embedded in modification strategy by optimization
So that reaching minimum to the disturbance of image.
3. method according to claim 2, which is characterized in that the insertion modification strategy of use is: by image-carrier vector x
Lowest bit position and time low bit position x1' and x2' replacement, the lowest bit position and time low bit position for being 0 and 255 to pixel value
The point modified carries out reversed ± 4 amendment;Wherein x1' indicate the carrier after first layer private communication channel insertion secret information
Vector, x2' indicate the carrier vector after second layer private communication channel insertion secret information.
4. a kind of method for being embedded in secret information in the picture, which comprises the following steps:
1) image present channel is expressed as carrier vector x, carrier vector x is indicated in the form of 8 binary, and will be minimum
Bit bit plane and time low bit bit plane take out respectively, are expressed as vector x1And vector x2;
2) cost calculating is carried out according to preset cost calculation to carrier vector x, obtained cost value is expressed as
Vector ρ;
3) according to cost vector ρ, secret information m is divided into m1And m2, respectively correspond in carrier vector x1And x2The information of middle insertion;
Secret information is embedded by carrier vector x using the allocation strategy based on two layers of Optimal Ratio of embedded quantity1And x2In, the base
It is in the allocation strategy of two layers of Optimal Ratio of embedded quantity:
A) by total embedding information length | m | and insertion cost ρ brings following formula into, is solved using Binary searches mode, obtains λ1;
Wherein, τ is Dynamic gene;ρi,jFor in the insertion cost of position (i, j), i, j indicate transverse and longitudinal coordinate, λ1It indicates by current total
Embedding information length | m | and the intermediate parameter that insertion cost ρ is obtained;
B) cost ρ and λ will be embedded in1Bring following formula into, calculate first layer private communication channel and be correspondingly embedded in message length | m1|, in turn
Pass through | m2|=| m |-| m1| calculate the corresponding embedding information length of second layer private communication channel | m2|;
4) by the lowest bit position of image-carrier vector x and time low bit position x1' and x2' replacement, it is 0 and 255 to pixel value
The point that lowest bit position and time low bit position are modified carries out reversed ± 4 amendment;Wherein x1' indicate the hidden letter of first layer
Road is embedded in the carrier vector after secret information, x2' indicate the carrier vector after second layer private communication channel insertion secret information;
5) to other channels of image, the insertion of secret information is all carried out according to the sequence of step 1) to step 4), it is final to obtain
To the modified carrier image of insertion.
5. method as claimed in claim 4, which is characterized in that step 3) is in carrier vector x1And x2Middle insertion secret information
Method is:
A) by vector x1As the carrier of first layer private communication channel, according to single layer STC working principle, by secret information m1With cost
Vector ρ is that insertion foundation is embedded into carrier vector x1In;Guarantee Η simultaneously1x1'=m1, wherein Η1To be generated according to shared parameter
Parity matrix, x1' indicate the carrier vector after first layer private communication channel insertion secret information;
B) by vector x2As the carrier of second layer private communication channel, according to the insertion in step a) as a result, by x1The middle point modified
Cost be set to zero, obtain new cost vector ρ ';By secret information m2It is that insertion foundation is embedded into carrier with cost vector ρ '
Vector x2In;Guarantee Η simultaneously2x2'=m2, wherein Η2For the parity matrix generated according to shared parameter, x2' indicate second
Carrier vector after layer private communication channel insertion secret information.
6. method as described in claim 4 or 5, which is characterized in that further include the extraction process of secret information, including following step
It is rapid:
A) under image present channel, image present channel is expressed as vector x, by the lowest bit bit plane of carrier vector x
It is taken out respectively with secondary low bit bit plane, is expressed as vector x1And vector x2;
B) parity matrix Η is generated according to shared parameter1, by Η1According to formula Η1x1=m1With lowest bit bit vector x1
Matrix multiple obtains insertion ciphertext m1;
C) parity matrix Η is generated according to shared parameter2, by Η2According to formula Η2x2=m2With lowest bit bit vector x2
Matrix multiple obtains insertion ciphertext m2;
D) step b) and step c) secret information extracted are combined into insertion ciphertext m;
E) under other channels of image, all secret informations are extracted according to the sequence of step a) to step d).
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201610471816.1A CN106169189B (en) | 2016-06-24 | 2016-06-24 | STC coding method based on embedded quantity bilayer Optimal Ratio |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201610471816.1A CN106169189B (en) | 2016-06-24 | 2016-06-24 | STC coding method based on embedded quantity bilayer Optimal Ratio |
Publications (2)
Publication Number | Publication Date |
---|---|
CN106169189A CN106169189A (en) | 2016-11-30 |
CN106169189B true CN106169189B (en) | 2019-08-27 |
Family
ID=58065924
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201610471816.1A Expired - Fee Related CN106169189B (en) | 2016-06-24 | 2016-06-24 | STC coding method based on embedded quantity bilayer Optimal Ratio |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN106169189B (en) |
Families Citing this family (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110365864B (en) * | 2018-04-10 | 2020-09-04 | 北京大学 | Image steganography method, image steganography system, computer device and computer-readable storage medium |
CN108898538A (en) * | 2018-05-08 | 2018-11-27 | 天津大学 | A kind of image airspace steganographic algorithm merging LBP feature |
CN110059754B (en) * | 2019-04-22 | 2019-12-24 | 厦门大学 | Batch data steganography method, terminal equipment and storage medium |
CN112714231A (en) * | 2020-12-28 | 2021-04-27 | 杭州电子科技大学 | Robust steganography method based on DCT (discrete cosine transformation) symbol replacement |
CN115049715B (en) * | 2022-05-19 | 2024-04-09 | 辽宁石油化工大学 | Image steganography method based on texture fusion feature block matching and large embedding rate product code |
Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101582157A (en) * | 2009-03-06 | 2009-11-18 | 中山大学 | Adaptive spatial steganographic method based on adjacent pixel difference |
CN104010193A (en) * | 2014-05-29 | 2014-08-27 | 中国科学院信息工程研究所 | Video steganographic method based on macro block partition mode disturbance |
-
2016
- 2016-06-24 CN CN201610471816.1A patent/CN106169189B/en not_active Expired - Fee Related
Patent Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101582157A (en) * | 2009-03-06 | 2009-11-18 | 中山大学 | Adaptive spatial steganographic method based on adjacent pixel difference |
CN104010193A (en) * | 2014-05-29 | 2014-08-27 | 中国科学院信息工程研究所 | Video steganographic method based on macro block partition mode disturbance |
Non-Patent Citations (3)
Title |
---|
An improved steganographic method of product perfect codes;Ze-li Zhao 等;《2011 IEEE International Conference on Signal Processing, Communications and Computing》;20111027;第1-5页 * |
一种融合湿纸码和STC码的高效隐写算法;刘华 等;《计算机应用研究》;20120630;第29卷(第6期);摘要、第1-2节 * |
基于像素分类的图像隐写分析算法;陈世媛 等;《万方数据库》;20160621;第39-46页 * |
Also Published As
Publication number | Publication date |
---|---|
CN106169189A (en) | 2016-11-30 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN106169189B (en) | STC coding method based on embedded quantity bilayer Optimal Ratio | |
Yang et al. | An embedding cost learning framework using GAN | |
CN103345767B (en) | A kind of JPEG image steganography method of high security | |
CN102917227B (en) | Compressive sensing-based adaptive video information hiding method | |
He et al. | Reversible data hiding in JPEG images based on negative influence models | |
CN104010193B (en) | Video steganographic method based on macro block partition mode disturbance | |
CN102938841B (en) | Method for hiding information in bearing image, image quality evaluation method and information transmission method | |
CN101419702B (en) | Semi-fragile watermarking method for color image tamper localization and recovery | |
Li et al. | Tamper detection and self-recovery of biometric images using salient region-based authentication watermarking scheme | |
CN101201939A (en) | Method and system for generating picture identifying code | |
CN107274334A (en) | A kind of color image-adaptive water mark method based on Quaternion Transformation | |
CN110457996B (en) | Video moving object tampering evidence obtaining method based on VGG-11 convolutional neural network | |
CN106097241A (en) | Reversible information hidden method based on eight neighborhood pixel | |
CN102073978B (en) | Method and system for identifying and recovering digital images by utilizing irregular region segmentation | |
CN112070647A (en) | Reversible high dynamic range image information hiding method | |
CN115018688A (en) | Network model watermark generation method based on DCT (discrete cosine transformation) coefficient and application | |
CN101923700B (en) | Double-effect digital watermarking method | |
CN105741222B (en) | A kind of steganography information locating method based on the estimation of pixel subset insertion rate | |
CN116757909B (en) | BIM data robust watermarking method, device and medium | |
CN109920014B (en) | 3D grid model steganography method | |
Li et al. | A new robust watermarhing scheme based on shuffled frog leaping algorithm | |
CN103440616B (en) | High volume reversible watermarking method based on self-adaptive prediction model | |
CN102156956A (en) | High robustness watermark method based on singular value decomposition and discrete cosine transform | |
CN106295704B (en) | Image hashing acquisition methods based on image library airspace and frequency domain character | |
Jiang et al. | A high capacity steganographic method based on quantization table modification |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant | ||
CF01 | Termination of patent right due to non-payment of annual fee | ||
CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20190827 Termination date: 20210624 |