Summary of the invention
For the deficiencies in the prior art, the present invention is based on the principle of code division multiplexing, by mutually orthogonal one group to
The reversible insertion that information is realized on the carrier image that is added to is measured, recipient is provided after receiving the close image of load according to sender
Key, that is, insertion vector, not only can completely extract the information of insertion, and recovery original graph that can be lossless
Picture.
The present invention uses following technical solution:
A kind of reversible information hidden method based on code division multiplexing, comprising:
In information sender: multiple carrier vectors are constituted by the pixel in setting sequential selection original image, using insertion
Vector carries embedding information into carrier vector, generates and carries the close image of close vector sum load;The insertion vector is using based on code
0 mean vector of mutually orthogonal one group for dividing principle of multiplexing to generate.
In receiving party: it obtains and carries close vector sum and carry close image, using carrying close vector described in the insertion vector analysis,
The embedding information is extracted, carrier vector sum original image is restored.
Further, the carrier vector is identical as the insertion length of vector.
Further, binary system conversion and coding are successively carried out to the embedding information.
The insertion vector are as follows: sσ∈ { -1,1 },
Further, the binary system conversion and cataloged procedure are as follows: when binary number is 1, replaced with number 1, when two
When system number is 0, replaced with number -1.
The embedding information are as follows:
Wherein, w represents original binary number, and b is the data after conversion
Further, described to generate the method that close vector sum carries close image that carries are as follows: by embedding information with insertion vector to set
Fixed embedment strength coefficient is input in carrier vector, obtains carrying close vector, carries close vector according to the setting sequence for described
It rearranges, constitutes and carry close image.
Information insertion when, embedding information according to the following formula shown in mode be embedded into image:
Wherein, α is embedment strength coefficient, and b is embedding information,It is the close vector of the later load of embedding data, according to certain
Sequential selection original image in pixel constitute carrier vector Vj=[p1,p2,.....,pl](j∈{1,2,...,N×N/
L }), k indicates the data of k bit.
Further, the carrier vector includes the non-reversible insertion carrier vector of reversible insertion carrier vector sum, works as insertion
When the product of vector field homoemorphism and embedment strength coefficient is greater than carrier vector and is embedded in the mould of inner product of vectors, which is reversible
Information is embedded in carrier vector, otherwise is embedded in carrier vector for non-reversible information;It is embedded in carrier vector for non-reversible information,
Meet after embedding information
Further, close vector is carried by the way that the inner product parsing for calculating the close vector of load with being embedded in vector is described, method particularly includes:
When the mould for carrying close vector and insertion inner product of vectors is less than the product of insertion vector field homoemorphism and one times of embedment strength coefficient, extracting should
Carry the embedding information that close vector carries;When the mould for carrying close vector and insertion inner product of vectors is greater than insertion vector field homoemorphism and twice of insertion
When the product of strength factor, the close vector of the load is not embedded into information;When the mould of the close vector of load and insertion inner product of vectors is more than or equal to embedding
The product of the mould of incoming vector and one times of embedment strength coefficient, and it is less than the product of insertion vector field homoemorphism and twice of embedment strength coefficient
When, mark whether the identification close vector of the load is embedded in information by positioning.
Sequential selection pixel when being embedded according to information, which is constituted, carries close vectorClose vector is carried by calculatingIt is embedding with difference
The inner product of incoming vector is extracted and is restored original image, is shown below:
Further, the present invention extracts the embedding information by sign () function.
WhenAndWhen, recipient judges the data of insertion for 1 (corresponding original binary number
1), whenAndWhen, the data of insertion are -1 (corresponding original binary number 0).
Further, the carrier vector is constituted using neighborhood pixels, and the neighborhood pixels are under the jurisdiction of in setting range.
Further, same embedding information or different embedding informations are repeatedly carried to identical carrier using different insertion vectors
Vector generates and carries the close image of close vector sum load.
Beneficial effects of the present invention:
This patent realizes the high-capacity reversible Information hiding based on sensitive image based on code division multiplexing technology, passes through difference
Insertion vector be embedded into secret information is reversible in carrier image.In the case where carrier image pixel value changes lesser situation,
The disposable insertion of bulk information may be implemented.Simultaneously as the orthogonality of insertion vector, the recursive insertion of secret information
Into carrier image;During multi-layer information repeats to be embedded in, occur mutually to offset between the constitution element of difference insertion vector
The phenomenon that disappearing still is able to maintain good picture quality to make to carry close image after the insertion of large capacity information;Moreover, this patent
The extraction of middle secret information is not influenced by insertion sequence, and the insertion vector sum embedment strength coefficient only grasped with recipient has
It closes, this materially increases the hiding flexibility and safety of reversible information.
Specific embodiment:
The invention will be further described with embodiment with reference to the accompanying drawing:
It is noted that following detailed description is all illustrative, it is intended to provide further instruction to the application.Unless another
It indicates, all technical and scientific terms used herein has usual with the application person of an ordinary skill in the technical field
The identical meanings of understanding.
It should be noted that term used herein above is merely to describe specific embodiment, and be not intended to restricted root
According to the illustrative embodiments of the application.As used herein, unless the context clearly indicates otherwise, otherwise singular
Also it is intended to include plural form, additionally, it should be understood that, when in the present specification using term "comprising" and/or " packet
Include " when, indicate existing characteristics, step, operation, device, component and/or their combination.
A kind of typical embodiment of the present invention is as shown in Figure 1, be included in information sender: by setting sequential selection original graph
Pixel as in constitutes multiple carrier vectors, carries embedding information into carrier vector using insertion vector, generates and carry close vector
With the close image of load;Insertion vector is using 0 mean vector of mutually orthogonal one group generated based on code division multiplexing principle.
In receiving party: obtaining and carry the close image of close vector sum load, carry close vector using described in insertion vector analysis, extract
Embedding information restores carrier vector sum original image.
Insertion vector in the present embodiment derives from Hadamard (Hadamard) matrix, the characteristics of according to hadamard matrix
(Hadamard (Hadamard) matrix is made of and is met+1 and -1 element(wherein HnIt is n rank
Hadamard square matrix, InFor n rank unit square matrix)), one group of orthogonal vectors is constituted using the row or column of hadamard matrix, then just
Hand over vector element all 1 or -1, and in each vector each element and be 0;Meanwhile the inner product between any two vector is
0.It is shown below:
If S=(sσ)1×mIt is that a 1 ' m being made of the row (column) of Hadamard matrix ties up 0 mean value insertion vector, then:
The inner product of any two difference vector are as follows:
Embedding information is converted into binary stream, binary stream is recompiled, when binary number is 1, is used
Number 1 replaces, and when binary number is 0, is replaced with number -1.
Wherein, w represents original binary number, and b is the data to be embedded after conversion.If I is having a size of the original of N × N
Image selects the pixel in original image to constitute carrier vector, V in a certain orderj=[p1,p2,.....,pl](j∈
{ 1,2 ..., N × N/l }), in information insertion, a vector is selected, is operated according to the following formula, the length of carrier vector
" l " and insertion vector SiUnanimously.Data according to the following formula shown in mode be embedded into image:
Wherein, α is embedment strength, and b is embedding information.It is the close vector of the later load of embedding data.According to the shape of formula (4)
Formula, it can be seen that the data of k bit are embedded in carrier vector VjOn.Embedment strength coefficient controls the depth of information insertion, embedding
It is bigger to enter intensity, embeddable data are more, also bigger to the introducing noise of original image.It willAccording to the suitable of original vector
Sequence, which rearranges to constitute, carries close image.
In the recipient for carrying close image, the content of insertion is extracted in the following way:
IfIt is to carry close image, sequential selection pixel when being embedded according to information, which is constituted, carries close vectorCalculating can be passed through
The close vector extraction embedding data lossless from the inner product of different insertion vectors is carried, and restores original image.As shown in formula (5);
Since different insertion vectors is mutually orthogonal, so, when using vector siWhen extracting information, formula (5) can simplify are as follows:
Formula (6) shows to be independent of each other between different insertion vectors, due to it is each insertion vector mutual orthogonality,
The forward and backward information being repeatedly embedded in is independent of each other;Namely information insertion sequence does not require absolutely consistent with sequence of extraction, this is also
One of the advantages of method.On the other hand, in formula (6), be∈ { -1,1 }, α is positive integer,Always positive number, so, expression
FormulaSymbol be by beDetermining;Work as beWhen=1,Symbol be positive, work as beWhen=- 1,Symbol
It is negative.Therefore, whenWhen, expression formulaSymbol be byCome what is determined.In such case
Under, embedding data b can be extracted by sign () functione:
By formula (7) it is found that when the product of insertion vector field homoemorphism and embedment strength coefficient is greater than carrier vector and insertion vector
Inner product when, embedding data can be extracted correctly.It is specific: whenAndWhen, recipient sentences
The data of disconnected insertion are 1, whenAndWhen, the data of insertion are -1.In this algorithm, all satisfactionsCarrier vector can embedding information.Such as: when selection is embedded in factor alpha=1, insertion vector Si=(1,
1, -1, -1) when, it is all meet element between mutual deviation absolute value less than 4(0,±1,±2,±3)Carrier vector can realize that information is embedding
Enter, in information telescopiny, the change of each pixel value is not more than 1.;When selection is embedded in factor alpha=2, insertion vector Si=
When (1,1, -1, -1), it is all meet element between mutual deviation absolute value less than 8(0,±1,±2,±3,±4,±5,±6,±7)Carrier vector be ok
Realize information insertion;And in information telescopiny, the change of each pixel value is not more than 2.Thus, it is keeping carrying close image
Under the premise of quality, information insertion ability is greatly improved.
After embedding data is correctly extracted, initial carrier vector can carry out according to the following formula Distortionless:
Due to being embedded in vector SiIt is 0 mean vector, carrier vector VjIt is to be made of neighborhood pixels, the inner product of the twoDeng
Valence is in the difference calculated in carrier vector between two adjacent elements and sums, when the element for constituting carrier vector is closely related
When, the value of inner product is very small.Therefore, the adjacent pixel similarity in image is higher, and the carrier vector fluctuation of composition is smaller, more
Carrier vector can be embedded into data.According to the Close relation of adjacent pixel in natural image, neighbouring similar pixel is selected
More information can be embedded in by constituting carrier vector, promote the ability of image information insertion.
Due to the content information redundancy properties between adjacent pixel in natural image, neighborhood pixels it is general very approximate, especially
It is, texture relatively small number of region relatively flat in image, and the gray value of adjacent pixel differs very little.Therefore,Value
It is general all smaller, the information insertion of large capacity can be realized in the case where causing the smaller quality of image to decline.On the other hand,
The selection of insertion coefficient also will affect the information insertion ability of this algorithm, and the value of embedment strength coefficient is bigger,Value also
Bigger, more carrier vectors can satisfy the insertion condition of information, thus can be embedded in more information, lifting carrier image
Information embedding capacity;On the other hand, biggish embedment strength coefficient, which also results in, carries close picture quality reduction amplitude increase, carries
Close picture quality rapid decrease with the increase of insertion coefficient.Suitable insertion can be selected strong according to different application scenarios
It spends coefficient and realizes that the reversible information of secret data is hidden.Formula (7) is embedded in coefficient it is also shown that only grasping correct insertion vector sum
Recipient could extract the information of insertion from carrying in close image and restore original image, this also increases the safety of this algorithm
Property.
During information insertion, by formula (7) it is found that working asWhen, embedding information just may be implemented
Correctly extract and restore original image.Thus need to be embedded in the carrier of secret information using the method mark of the target icon note
The position of vector guarantees the correct complete recovery extracted with original image of embedding information.It can according to the operation result of formula (7)
Know, the inner product of the carrier vector after insertion and insertion vector centainly meets:
On the other hand, the inner product of carrier vector after insertion and insertion vector meetsWhen, carrier
Secret information is centainly not embedded into vector.Thus, it does not need in image recipient to all carriers for being unsatisfactory for information insertion
Vector is labeled, but only that markBetween pixel, so as to greatly reduce
The embedded quantity of accessory information is conducive to be embedded in more information.
Fig. 2, which gives, carries close framing figure information labeling.
The present embodiment is during Information hiding, on the one hand, since insertion vector is mutually orthogonal, data can be with independent more
It is secondary to be added on carrier vector, the correct extraction without influencing information.Consider the selected vector of each embedding data mutually just
The characteristic of friendship, the data of insertion do not interfere with the extraction of second of embedding data on carrier vector for the first time.It therefore, can be with
By multi-layer information be embedded in promoted information insertion capacity, different insertion vectors can carry information repeat be added to carrier to
The insertion that secret information is realized in amount, to increase the Information hiding ability of carrier image.
On the other hand, during information repeats additive embedding, it will appear between the element of different orthogonal vector and mutually support
The phenomenon that disappearing allows to carry close image and keeps higher to reduce the decrease speed of the picture quality in information telescopiny
Picture quality.Such as: when taking embedment strength factor alpha=1, embedding data 1, being embedded in vector for the first time is S1=(1,1 ,-
1, -1) when, the change of information embedded images original pixels gray value is not more than 1.When carrying out second of information insertion, if choosing
Selecting insertion vector is S2=(1, -1,1, -1), when embedding data is 1, after information insertion twice, the changing into of original pixels (2,
0,0, -2);When embedding data is 0, after information insertion twice, original pixels change into (0,2, -2,0).Due to different insertions
Offseting each other between vector element, most elements revert to the value of original pixels again.Thus, in the insertion of large capacity information,
Realize that the hiding close image of load of reversible information is still able to maintain higher picture quality based on this algorithm.
This embodiment is illustrated by way of example may be implemented the order-independency principle of considerable information hide, information superposition insertion:
If two bits to be embedded are 1 and 0 respectively, according to formula (3), two binary numbers are converted to 1 and -1.If
Initial carrier vector is V1=(166,166,165,165), insertion vector is respectively S1=(1, -1,1, -1), S2=(1,1 ,-
1, -1), strength factor α=1 of information insertion, after selecting initial carrier vector to execute the insertion of first time information, initial carrier figure
The grey scale pixel value variation of picture is respectively 1 and -1, and the value for carrying close vector isIt is embedded in second of information
Afterwards, the value for carrying close vector isThe first of carrier vector, the gray-value variation of three elements are 2, first, four
The gray-value variation of element reverts to 0.In information superposition telescopiny, the variation of most of carrier image pixel is cancelled,
Thus, in multiple information telescopiny, carrier quality downward gradient caused by information is embedded in gradually slows down, and carries close image big
There is good quality holding capacity in capacity information telescopiny.Meanwhile by table 1 it can also be seen that being based on code division multiplexing
Reversible information hidden algorithm in, the extraction process of information with hide unrelated namely embedding information the extraction of sequence not by information
It is embedded in the limitation of sequencing, and it is only related with the embedment strength of information extraction and insertion vector.This largely increases
The hiding flexibility of reversible information;Meanwhile recipient's ability of correct insertion vector sum embedment strength coefficient is only grasped
Hiding secret information is extracted, this also improves the safety of this algorithm.
Multi-layer information insertion and extraction process of the table 1. based on code division multiplexing
The foregoing is merely preferred embodiment of the present application, are not intended to limit this application, for the skill of this field
For art personnel, various changes and changes are possible in this application.Within the spirit and principles of this application, made any to repair
Change, equivalent replacement, improvement etc., should be included within the scope of protection of this application.