CN106600518A - Image encryption method and image decryption method with visual security and data security based on compressed sensing - Google Patents

Abstract

The invention relates to an image encryption method and an image decryption method with visual security and data security based on compressed sensing. The image encryption method comprises the steps of: firstly, utilizing an SHA 256 hash function to obtain a 256-bit hash value of a plaintext image as an image secret key, and calculating initial numerical values of one-dimensional skew tent chaotic mapping and zigzag scrambling; carrying out sparse processing on the plaintext image, and carrying out zigzag scrambling on a coefficient matrix; and then utilizing the one-dimensional skew tent chaotic mapping to generate a measurement matrix, measuring and quantifying a scrambling matrix to obtain a compressed and encrypted image, and embedding the image into a carrier image with visual significance to obtain a final ciphertext image with visual significance. The image encryption method realizes the visual security and data security of the plaintext image, has large secret key space, is highly sensitive to plaintext, has higher capacity of resisting brute-force attack, chosen-plaintext attack and known-plaintext attack, does not need an additional storage space, and can transmit and store the ciphertext image quickly and effectively.

Description

Image encryption, the decryption method of visual security and data safety based on compressed sensing

Technical field

The present invention relates to image Encrypt and Decrypt technical field, more particularly to a kind of visual security sum based on compressed sensing According to the image encryption of safety, decryption method.

Background technology

As the fast development of digital technology and network technology, increasing multi-medium data are produced, passed by network It is defeated and stored in platforms such as Cloud Servers.Numerical data includes bulk information, and for example, a military oil depot picture not only may be used To teach that its size and number, but also its Position Approximate can be provided；One width human face photo not only can expose His or her appearance, can also provide general age and health.Therefore, in medical image system, military picture system In video conference, the safety for protecting view data causes extensive concern, and image encryption is that one kind can be with effective protection number According to the method for safety.

At present, many resume images are suggested, and they have used chaos system, DNA calculating, cellular automata, light Learn the technologies such as conversion, Brownian movement, wave function transmission.One most important feature of known most of resume images is just It is：Plaintext image is converted into a width noise like ciphertext, the rectangular histogram of ciphertext graph picture be it is smooth, pixel value [0,255] it Between be uniformly distributed, the comentropy close 8 of ciphertext graph picture, ciphertext graph picture is delivered to recipient by sender, and network attack person pass through Analysis ciphertext hardly results in key and corresponding plaintext.Most of AESs can effective protection view data safety, Some of them algorithm due to key space is little or algorithm and plaintext dependency it is little and be broken.Additionally, these algorithms have one Common drawback, is exactly that the ciphertext of noise like is easy to cause the attention of attacker in transmission and storage, so as to being found and attacking Hit.Based on this, if we design a kind of ciphertext of visual security, i.e. ciphertext graph picture has significant outward appearance, in their quilts In storage and transmission, it is difficult to be found, so as to avoid the possibility attacked, it is ensured that the safety of itself.Therefore, design tool The resume image for having visual security and data safety is a job highly significant.Simultaneously, it is contemplated that transmission bandwidth, deposit Storage space and the consideration of real-time application, the size that we are not intended to the ciphertext graph picture obtained after image encryption is bright more than corresponding Texts and pictures picture.

2006, Candes and Donoho proposed the concept of compressed sensing, and its core concept is tied using information space Structure, merging of compressing and sample is carried out, and breaches the restriction of Shannon's sampling theorem, can greatly reduce sampling number, in compression Perception concept occurs shortly having researcher to consider compressed sensing is used in image encryption.For example, H.Fang etc. is zigzag Scramble and compressed sensing are used in the compression encryption of image and video, find zigzag scrambles can effectively improve reconstruction image and The signal to noise ratio of video.Zhang Yushu etc. is used in the compression of image encryption using chaos system and compressed sensing, and first image is dilute Sparse coefficient is sorted scramble by chaos sequence, then measures encryption, the ciphertext graph of last noise like using compressed sensing As being obtained.Zhou Nanrun etc. devises a kind of new image using non-linear fractional order Mellin transform and two dimensional compaction cognition technology Compression AES, the tested moment matrix of plaintext image is measured from both direction, is then become using non-linear fractional order plum forests The secondary encryption of swap-in row.In these AESs, the combination of scrambling process and compressed sensing process can effectively loosen compression The limited equidistant restriction for perceiving, reduces the computation complexity of algorithm, while enhancing compression and the encryption performance of algorithm.But It is that in AES, scrambling process and plaintext are unrelated, this means that identical scramble vector adds for different plaintext images It is close, the anti-chosen -plain attact of algorithm and known plain text attack it is poor.Therefore, increase the association between algorithm and plaintext, contribute to The further safety of boosting algorithm.

The content of the invention

For deficiency of the prior art, the present invention proposes a kind of visual security and data safety based on compressed sensing Image encryption, decryption method, height is relied in plain text, the energy of anti-selection plaintext and known plain text attack in raising encryption process Power, using so that key space greatly improves for the hash functions of SHA 256, strengthens the opposing brute force attack of image encrypting and decrypting technology Ability.

According to design provided by the present invention, a kind of image of the visual security and data safety based on compressed sensing Encryption method, comprises the steps of：

Plaintext image P of the step 1. using the hash functions of SHA 256 to size for m × n is calculated, and obtains image key Key, and calculate the initial value of one-dimensional skew tent chaotic maps and zigzag scrambles；

Step 2. carries out sparse process using wavelet transform to plaintext image P, obtains coefficient matrix P1；To coefficient square Battle array P1 enters line shuffle according to zigzag paths, obtains the matrix P2 after scramble；By predetermined threshold value TS, matrix P2 is repaiied Just, revised matrix P3 is obtained；

Step 3. utilizes one-dimensional skew tent chaotic maps to generate the calculation matrix Φ for compressed sensing, and to matrix P3 is measured, and obtains measuring value matrix P4, wherein, the compression ratio of plaintext image P is expressed as CR, and the size of calculation matrix Φ is The size of M × N, then M=CR × m, N=m, matrix P4 is M × n；

Step 4. quantifies to the element of matrix P4, obtains matrix P5, as the ciphertext graph picture after compression encryption；

It is the significant image of vision of m × n as carrier image CI that step 5. selects size, and to it implement from Scattered wavelet transformation, obtains four matrixes：Low frequency part decomposition coefficient LL, horizontal direction decomposition coefficient LH, vertical direction resolving system Number HL and diagonal decomposition coefficient HH, is embedded matrix P5, and discrete wavelet inverse transformation is carried out again, obtains final The significant ciphertext graph picture of vision.

Above-mentioned, the initial value of one-dimensional skew tent chaotic maps and zigzag scrambles is calculated in step 1, comprising：Will Image key key is scaled 32 decimal number k₁, k₂..., k₃₂, and calculate the parameter and just of one-dimensional skewtent chaotic maps Initial value, and the initial position co-ordinates of zigzag scrambles.

Preferably, one-dimensional skew tent chaotic maps are expressed as：

,

Wherein, r ∈ (0, be 1) systematic parameter, z ∈ (0,1) be chaos system state variable；And according to formula：

To calculate 4 intermediate parameters t_i(i=1,2,3,4), wherein, l_i(i=1,2,3, are 4) preliminary setting parameters, respectively Representation parameter x '₀, y '₀, r ', z '₀, calculated t₃And t₄Parameter r and just of one-dimensional skew tent chaotic maps is represented respectively Initial value z₀；The initial position co-ordinates x of zigzag scrambles₀And y₀By formula：

It is calculated, wherein, abs (x) represents the absolute value for seeking x, and mod is modulo operation.

Above-mentioned, matrix P2 is modified in the step 2, specifically include：By predetermined threshold value TS, to the unit in P2 Element is modified, and the element less than or equal to TS is changed into 0.

Above-mentioned, the one-dimensional skew tent chaotic maps that utilize in step 3 generate the calculation matrix for compressed sensing Φ, comprising following content：

Step 3.1, orderWherein, d represents sampling Interval, r, z₀The parameter mapped for one-dimensional skew tent and initial value, l represents sampling length, and n represents Sampling starting point, will be step Parameter r and initial value z of rapid 1 calculated one-dimensional skew tent chaotic maps₀Substitute into one-dimensional skew tent chaotic maps In, generate chaos sequence；

Step 3.2, according to formulaLine translation is entered to the numerical value of chaos sequence；

Step 3.3, generateArranged by row, obtained calculation matrix, size is M × N,

In formula,For normalized.

Preferably, following content is also included in step 3.1：The front n-1 numerical value of chaos sequence is abandoned, from nth Value starts, and at interval of d numbers, takes out one, obtains new sequence of the length for MN, used as the new chaos sequence replaced.

Above-mentioned, the element in step 4 to size for the matrix P4 of M × n quantifies, specifically comprising following content：Will Element in matrix P4 is converted into a vector according to order from left to right, from top to bottom, according to formula：

p_5i=floor (255 × (p_4i-min)/(max-min))

Vector element is quantified one by one, wherein, min represents the minima of element in matrix P4, and max represents matrix P4 The maximum of middle element, floor (x) is represented and is taken the no more than maximum integer of x, p_4iRepresent matrix P4 according to from left to right, from upper I-th element (1≤i≤Mn) in the vector that order under is obtained, by p_5iThe vector of composition is converted into size for M × n's Matrix P5.

Above-mentioned, step 5 specifically includes following content：

Step 5.1, according to formula：

D1 (i, j)=P5 (i, j) mod10, D2 (i, j)=floor (P5 (i, j)/10)

Element in matrix P5 is processed one by one, matrix D 1 and D2 is obtained, wherein, D1 (i, j), D2 (i, j) and P5 (i, j) is represented respectively in matrix D 1, D2 and P5 positioned at the element at (i, j) place, 1≤i≤M, 1≤j≤n；

Step 5.2, according to compression ratio CR of plaintext image P, the size of matrix D 1, D2 and P5 is set；

Step 5.3, wavelet transform is carried out to carrier image CI, obtain four matrixes, be expressed as：Low frequency part Decomposition coefficient LL, horizontal direction decomposition coefficient LH, vertical direction decomposition coefficient HL and diagonal decomposition coefficient HH, they Size is identical with matrix D 1, D2 sizes；

Step 5.4, one size of definition and matrix D 1, D2 size identical empty matrix IT, calculate LL squares Meansigma methodss MV of all elements in battle array, according to rule：

Rule 1：If LL (i, j) >=MV, then IT (i, j)=1, LH (i, j)=D1 (i, j), HL (i, j)=D2 (i, j),

Rule 2：If LL (i, j)<MV, then IT (i, j)=0, LH (i, j)=D2 (i, j), HL (i, j)=D1 (i, j), In D1, D2 embeded matrix LH and HL, wherein, 1≤i≤m/2,1≤j≤n/2；

Step 5.5, to be embedded in after four matrix Ls L, LH, HL and HH carry out discrete wavelet inverse transformation, obtain vision intentional The ciphertext graph picture of justice, ciphering process terminates.

A kind of image decryption method of the visual security and data safety based on compressed sensing, concrete decryption step is as follows：

Step a：Obtain image key key of plaintext image P, and parameter x '₀、y′₀、r′、z′₀, min and max；

Step b：Image key key is scaled into 32 decimal number k₁, k₂..., k₃₂, and calculate one-dimensional skew tent The parameter and initial value of chaotic maps, and the initial position co-ordinates of zigzag scrambles；One-dimensional skewtent chaotic maps are expressed as：

,

Wherein, r ∈ (0, be 1) systematic parameter, z ∈ (0,1) be chaos system state variable；And according to formula：

To calculate 4 intermediate parameters t_i(i=1,2,3,4), wherein, l_i(i=1,2,3, are 4) preliminary setting parameters, respectively Representation parameter x '₀, y '₀, r ', z '₀, calculated t₃And t₄Parameter r and just of one-dimensional skew tent chaotic maps is represented respectively Initial value z₀；The initial position co-ordinates x of zigzag scrambles₀And y₀By formula：

It is calculated, wherein, abs (x) represents the absolute value for seeking x, and mod is modulo operation；

Step c：To size to be decrypted for m × n ciphertext graph as C, carry out wavelet transform, obtaining 4 sizes is (m/2) matrix of × (n/2), respectively：Low frequency part decomposition coefficient LL1, horizontal direction decomposition coefficient LH1, vertical direction point Solution coefficient HL1 and diagonal decomposition coefficient HH1；

Step d：Define empty matrix IT1 of the size for (m/2) × (n/2), the meansigma methodss of all elements of calculating matrix LL1 MV1, according to rule：

Regular A：If LL1 (i, j) >=MV1, then IT1 (i, j)=1, D11 (i, j)=LH1 (i, j), D21 (i, j)= HL1 (i, j),

Regular B：If LL1 (i, j)<MV1, then IT1 (i, j)=0, D21 (i, j)=LH1 (i, j), D11 (i, j)=HL1 (i, j), extracts cleartext information from LH1 and HL1, obtains matrix D 11 and D21, wherein, 1≤i≤m/2,1≤j≤n/ 2；

Step e：Matrix D 11, D21 is transformed to into the matrix that size is (m/4) × n, and by formula：P51 (i, j)= D21 (i, j) × 10+D11 (i, j), obtains matrix P51, wherein, D11 (i, j), D21 (i, j) and P51 (i, j) represent respectively square Battle array D11, D21 and P51 are located at the element at (i, j) place, 1≤i≤m/4,1≤j≤n；

Step f：Matrix P51 according to order from left to right, from top to bottom be converted into length for 1 × (mn/4) to Amount, p_51i(1≤i≤mn/4) is i-th element of vector, according to formula：To vector Element carries out one by one inverse quantization operation, obtains p_41i(1≤i≤mn/4), vector the matrix that size is (m/4) × n is changed into P41；

Step g：The calculation matrix Φ 1 for compressed sensing is generated using one-dimensional skew tent chaotic maps, using compression Perceive algorithm for reconstructing to be reconstructed P41, obtain matrix P31 of the size for m × n；

Step h：Using the initial position co-ordinates x of the calculated zigzag scrambles of step b₀And y₀Inverse is implemented to matrix P31 Zigzag is converted, and obtains matrix P11, discrete wavelet is carried out to matrix P11 and is operated against sparse transformation, the plaintext figure after being decrypted As P.

Above-mentioned, the one-dimensional skew tent chaotic maps that utilize in step g generate the calculation matrix Φ for compressed sensing 1, comprising following content：

Step g1, orderWherein, d is represented between sampling Every r, z₀The parameter mapped for one-dimensional skew tent and initial value, l represents sampling length, and n represents Sampling starting point, by step b Parameter r and initial value z of calculated one-dimensional skew tent chaotic maps₀In substituting into one-dimensional skew tent chaotic maps, Formation sequence, the front n-1 numerical value of sequence is abandoned, and from the beginning of n-th numerical value, at interval of d numbers, takes out one, obtains length The new sequence for MN is spent, as chaos sequence；

Step g2, according to formulaLine translation is entered to the numerical value of chaos sequence；

Step g3, generateArranged by row, obtained calculation matrix, size is M × N,

In formula,For normalized.

Beneficial effects of the present invention：

1st, the present invention proposes a kind of image encryption method of visual security, by using zigzag scrambles and compressed sensing skill Art is compressed and encrypts to plaintext image, obtains the ciphertext of width compression, and it is embedded into the significant carrier of a width vision In image, the significant ciphertext graph picture of a width vision is obtained；Compared with conventional noise like ciphertext graph picture, significant ciphertext graph As being not easy to be found by attacker, the safety of image is effectively ensured.

2nd, image encryption method proposed by the present invention, plaintext image is sparse first, and the sparse coefficient for obtaining is according to one kind Zigzag paths scramble, the initial position of zigzag scrambles is controlled by plaintext, and it is adjacent that scrambling process not only can reduce image Dependency between pixel, and after can strengthening compressed sensing under identical compression ratio image reconstruction effect；Using in plain text Related key generates calculation matrix, and measurement encryption is compressed to the matrix after scramble；In telescopiny, first to carrier Image implements wavelet transform, obtains 4 matrix Ls L, LH, HL and HH, according to the element value of LL matrixes, after compression encryption Ciphertext matrix element be embedded into one by one in LH the and HL matrixes of carrier image, the data safety of plaintext image has been effectively ensured.

3rd, image encryption method proposed by the present invention is extremely sensitive to plaintext, using the Hash letters of SHA 256 of plaintext image Numerical value increased key space as key, while calculating the parameter of one-dimensional skew tent chaotic maps and initial using it Value, and the initial position co-ordinates x that zigzag scrambles need₀And y₀, change in elevation sensitivity of the whole ciphering process to plaintext, increasing Strong encryption method opposing brute force attack, chosen -plain attact, the ability of known plain text attack, further increase image encryption side The safety of method, effectiveness.

4th, carrier image of the present invention can choose at random, and increased the key space of image encryption, while obtaining after encryption Ciphertext graph picture it is identical with plaintext picture size, it is not necessary to extra memory space, ciphertext graph picture can be passed fast and effectively Defeated and storage, algorithm is convenient to carry out.

Description of the drawings：

Fig. 1 is one of image encryption method flow chart of the present invention；

Fig. 2 is the two of the image encryption method flow chart of the present invention；

Fig. 3 is the decrypting process flow chart of resume image of the present invention；

In Fig. 4：A () is original image used by example one, (b) compress the ciphertext graph picture after encryption for example one, (c) is The carrier image of example one, is (d) the ciphertext graph picture of the visual security of example one, (e) is the correct key decrypted image of example one, (f) Decrypted image when being decrypted using wrong key for example one；

Fig. 5 is zigzag scramble schematic diagrams in example one；

Fig. 6 be in example one initial position co-ordinates for (1, the matrix before and after zigzag scrambles 1) (a) is put for zigzag Matrix before unrest, is (b) matrix after zigzag scrambles；

Fig. 7 be in example one initial position co-ordinates for (3, the matrix before and after zigzag scrambles 4) (a) is put for zigzag Matrix before unrest, is (b) matrix after zigzag scrambles；

In Fig. 8：A () is original image used by example two, be (b) the ciphertext graph picture after the correspondence compression encryption of example two, C () is the carrier image of example two, be (d) the ciphertext graph picture of the visual security of example two, is (e) the correct key of example two decryption figure Picture, decrypted image when (f) being decrypted using wrong key for example two.

Specific embodiment：

Below in conjunction with the accompanying drawings the present invention is further detailed explanation with technical scheme, and detailed by preferred embodiment Describe bright embodiments of the present invention in detail, but embodiments of the present invention are not limited to this.

Embodiment one, a kind of shown in Figure 1, image encryption side of the visual security and data safety based on compressed sensing Method, comprises the steps of：

Plaintext image P of the step 1. using the hash functions of SHA 256 to size for m × n is calculated, and obtains image key Key, and calculate the initial value of one-dimensional skew tent chaotic maps and zigzag scrambles；

Step 2. carries out sparse process using wavelet transform to plaintext image P, obtains coefficient matrix P1；To coefficient square Battle array P1 enters line shuffle according to zigzag paths, obtains the matrix P2 after scramble；By predetermined threshold value TS, matrix P2 is repaiied Just, revised matrix P3 is obtained；

Step 3. utilizes one-dimensional skew tent chaotic maps to generate the calculation matrix Φ for compressed sensing, and to matrix P3 is measured, and obtains measuring value matrix P4, wherein, the compression ratio of plaintext image P is expressed as CR, and the size of calculation matrix Φ is The size of M × N, then M=CR × m, N=m, matrix P4 is M × n；

Step 4. quantifies to the element of matrix P4, obtains matrix P5, as the ciphertext graph picture after compression encryption；

It is the significant image of vision of m × n as carrier image CI that step 5. selects size, and to it implement from Scattered wavelet transformation, obtains four matrixes：Low frequency part decomposition coefficient LL, horizontal direction decomposition coefficient LH, vertical direction resolving system Number HL and diagonal decomposition coefficient HH, is embedded matrix P5, and discrete wavelet inverse transformation is carried out again, obtains final The significant ciphertext graph picture of vision.

The present invention can simultaneously realize the data safety and visual security of plaintext image to be encrypted, effectively increase image The anti-selection of encryption in plain text and known plain text attack ability, the hash functions of SHA 256 use the key space for causing algorithm Greatly improve, enhance the ability that algorithm resists brute force attack.

Embodiment two, a kind of shown in Figure 2, image encryption side of the visual security and data safety based on compressed sensing Method, comprising following content：

One) the plaintext image P using the hash functions of SHA 256 to size for m × n is calculated, and obtains image key Key, and calculate the initial value of one-dimensional skew tent chaotic maps and zigzag scrambles；Calculate one-dimensional skew tent chaos to reflect The initial value with zigzag scrambles is penetrated, comprising：Image key key is scaled into 32 decimal number k₁, k₂..., k₃₂, and count Calculate the parameter and initial value of one-dimensional skew tent chaotic maps, and the initial position co-ordinates of zigzag scrambles.

Preferably, one-dimensional skew tent chaotic maps are expressed as：

Wherein, r ∈ (0, be 1) systematic parameter, z ∈ (0,1) be chaos system state variable；And according to formula：

To calculate 4 intermediate parameters t_i(i=1,2,3,4), wherein, l_i(i=1,2,3, are 4) preliminary setting parameters, respectively Representation parameter x '₀, y '₀, r ', z '₀, calculated t₃And t₄Parameter r and just of one-dimensional skew tent chaotic maps is represented respectively Initial value z₀；The initial position co-ordinates x of zigzag scrambles₀And y₀By formula：

It is calculated, wherein, abs (x) represents the absolute value for seeking x, and mod is modulo operation.

Two) sparse process is carried out to plaintext image P using wavelet transform, obtains coefficient matrix P1；To coefficient matrix P1 enters line shuffle according to zigzag paths, obtains the matrix P2 after scramble；By predetermined threshold value TS, the element less than or equal to TS It is changed into 0, matrix P2 is modified, obtains revised matrix P3.

Three) the calculation matrix Φ for compressed sensing is generated using one-dimensional skew tent chaotic maps, in following Hold：

1), makeWherein, d represents the sampling interval, r,z₀The parameter mapped for one-dimensional skew tent and initial value, l represents sampling length, and n represents Sampling starting point, will be step one) Parameter r and initial value z of calculated one-dimensional skew tent chaotic maps₀In substituting into one-dimensional skew tent chaotic maps, Formation sequence, in order to avoid the transient effect of chaos, the front n-1 numerical value of sequence is abandoned, from the beginning of n-th numerical value, per Every d numbers, one is taken out, new sequence of the length for MN is obtained, as chaos sequence；

2), according to formulaLine translation is entered to the numerical value of chaos sequence；

3), generationArranged by row, obtained calculation matrix, size is M × N,

In formula,For normalized.

And matrix P3 is measured according to calculation matrix Φ, obtain measuring value matrix P4, wherein, the pressure of plaintext image P Shrinkage is expressed as CR, and the size of calculation matrix Φ is M × N, then M=CR × m, and the size of N=m, matrix P4 is M × n.

Four) element in matrix P4 is converted into into a vector according to order from left to right, from top to bottom, according to public affairs Formula：

p_5i=floor (255 × (p_4i-min)/(max-min))

Vector element is quantified one by one, wherein, min represents the minima of element in matrix P4, and max represents matrix P4 The maximum of middle element, floor (x) is represented and is taken the no more than maximum integer of x, p_4iRepresent matrix P4 according to from left to right, from upper I-th element (1≤i≤Mn) in the vector that order under is obtained, by p_5iThe vector of composition is converted into size for M × n's Ciphertext graph picture after matrix P5, as compression encryption.

Five) size is selected to be the significant image of vision of m × n as carrier image CI and discrete to its enforcement Wavelet transformation, obtains four matrixes：Low frequency part decomposition coefficient LL, horizontal direction decomposition coefficient LH, vertical direction decomposition coefficient HL and diagonal decomposition coefficient HH, is embedded matrix P5, and discrete wavelet inverse transformation is carried out again, obtains final regarding Significant ciphertext graph picture is felt, specifically comprising following content：

1st, according to formula：

D1 (i, j)=P5 (i, j) mod10, D2 (i, j)=floor (P5 (i, j)/10)

Element in matrix P5 is processed one by one, matrix D 1 and D2 is obtained, wherein, D1 (i, j), D2 (i, j) and P5 (i, j) is represented respectively in matrix D 1, D2 and P5 positioned at the element at (i, j) place, 1≤i≤M, 1≤j≤n；

2nd, according to compression ratio CR of plaintext image P, the size of matrix D 1, D2 and P5 is set；

3rd, wavelet transform is carried out to carrier image CI, obtains four matrixes, be expressed as：Low frequency part resolving system Number LL, horizontal direction decomposition coefficient LH, vertical direction decomposition coefficient HL and diagonal decomposition coefficient HH, their size is big It is little identical with matrix D 1, D2 sizes；

4th, a size and matrix D 1, D2 size identical empty matrix IT are defined, calculates institute in LL matrixes There are meansigma methodss MV of element, according to rule：

Rule 1：If LL (i, j) >=MV, then IT (i, j)=1, LH (i, j)=D1 (i, j), HL (i, j)=D2 (i, j),

Rule 2：If LL (i, j)<MV, then IT (i, j)=0, LH (i, j)=D2 (i, j), HL (i, j)=D1 (i, j), In D1, D2 embeded matrix LH and HL, wherein, 1≤i≤m/2,1≤j≤n/2；

5th, to be embedded in after four matrix Ls L, LH, HL and HH carry out discrete wavelet inverse transformation, obtain vision significant close Texts and pictures picture, ciphering process terminates.

By a kind of zigzag paths scramble, scrambling process is subject to plaintext to the small echo sparse coefficient of plaintext image in the present invention Control, is compressed and is encrypted using compressed sensing to scramble image, and one-dimensional skew tent chaotic maps are used to produce measurement Matrix, using cleartext information the parameter and initial value of chaotic maps are produced, in the embedded width carrier image of the ciphertext graph picture for obtaining, The ciphertext graph picture of final width visual security is obtained；Based on compressed sensing and zigzag scrambles, plaintext image is compressed simultaneously And encryption, random zigzag scramble strategies select can effectively reducing compressed perception theory limited equidistant restriction, raising The compression performance of algorithm, the ciphertext graph picture for obtaining is identical with plaintext picture size, it is not necessary to which extra transmission bandwidth and storage is empty Between, ciphertext graph picture can fast and effectively be transmitted and stored；Again, the hash functions of SHA 256 of plaintext image are used to meter Some parameters in encryption method are calculated, algorithm is highly relied in plain text, improve the anti-selection plaintext and known plain text attack of algorithm Ability, using so that the key space of algorithm greatly improves for the hash functions of SHA 256 enhance algorithm opposing brute force attack Ability.

Image decryption process is the inverse process of ciphering process.Before image decryption, decryption side needs to obtain plaintext image P Image key key and parameter x '₀、y′₀、r′、z′₀, min and max, these are given datas, and this case application also provides one kind The image decryption method of visual security and data safety based on compressed sensing, shown in Figure 3, concrete decryption step is as follows：

Step a：Obtain image key key of plaintext image P, and parameter x '₀、y′₀、r′、z′₀, min and max；

Step b：Image key key is scaled into 32 decimal number k₁, k₂..., k₃₂, and calculate one-dimensional skew tent The parameter and initial value of chaotic maps, and the initial position co-ordinates of zigzag scrambles.

One-dimensional skew tent chaotic maps are expressed as：

,

Wherein, r ∈ (0, be 1) systematic parameter, z ∈ (0,1) be chaos system state variable；And according to formula：

To calculate 4 intermediate parameters t_i(i=1,2,3,4), wherein, l_i(i=1,2,3, are 4) preliminary setting parameters, respectively Representation parameter x '₀, y '₀, r ', z '₀, calculated t₃And t₄Parameter r and just of one-dimensional skew tent chaotic maps is represented respectively Initial value z₀；The initial position co-ordinates x of zigzag scrambles₀And y₀By formula：

It is calculated, wherein, abs (x) represents the absolute value for seeking x, and mod is modulo operation.

Step c：To size to be decrypted for m × n ciphertext graph as C, carry out wavelet transform, obtaining 4 sizes is (m/2) matrix of × (n/2), respectively：Low frequency part decomposition coefficient LL1, horizontal direction decomposition coefficient LH1, vertical direction point Solution coefficient HL1 and diagonal decomposition coefficient HH1；

Step d：Define empty matrix IT1 of the size for (m/2) × (n/2), the meansigma methodss of all elements of calculating matrix LL1 MV1, according to rule：

Regular A：If LL1 (i, j) >=MV1, then IT1 (i, j)=1, D11 (i, j)=LH1 (i, j), D21 (i, j)= HL1 (i, j),

Regular B：If LL1 (i, j)<MV1, then IT1 (i, j)=0, D21 (i, j)=LH1 (i, j), D11 (i, j)=HL1 (i, j), extracts cleartext information from LH1 and HL1, obtains matrix D 11 and D21, wherein, 1≤i≤m/2,1≤j≤n/ 2；

Step e：Matrix D 11, D21 is transformed to into the matrix that size is (m/4) × n, and by formula：P51 (i, j)= D21 (i, j) × 10+D11 (i, j), obtains matrix P51, wherein, D11 (i, j), D21 (i, j) and P51 (i, j) represent respectively square Battle array D11, D21 and P51 are located at the element at (i, j) place, 1≤i≤m/4,1≤j≤n；

Step f：Matrix P51 according to order from left to right, from top to bottom be converted into length for 1 × (mn/4) to Amount, p_51i(1≤i≤mn/4) is i-th element of vector, according to formula：To vector Element carries out one by one inverse quantization operation, obtains p_41i(1≤i≤mn/4), vector the matrix that size is (m/4) × n is changed into P41；

Step g：Method according to calculation matrix is produced in encrypting step, is generated using one-dimensional skew tent chaotic maps For the calculation matrix Φ 1 of compressed sensing, P41 is reconstructed using compressed sensing reconstruction algorithm, obtains size for m × n's Matrix P31；

Step h：The initial position co-ordinates x of the zigzag scrambles obtained using step b₀And y₀Inverse is implemented to matrix P31 Zigzag is converted, and obtains matrix P11, discrete wavelet is carried out to matrix P11 and is operated against sparse transformation, the plaintext figure after being decrypted As P.

Further to verify effectiveness of the invention, with reference to image encrypting and decrypting method of the instantiation to this case application It is further explained explanation：

Example one, referring to shown in Fig. 4～7, in the present embodiment, the programming software for adopting is Matlab R2014a, chooses Size for 512 × 512 Lena gray level images as experimental subject, concrete ciphering process is as follows：Step 1：Input original size For 512 × 512 Lena gray level images, image information is read with P=imread (' Lena.bmp '), using the Hash of SHA 256 Function pair plaintext image P is calculated, and the cryptographic Hash of a group 256 is obtained and using it as image key key, then by 256 The key of position is converted into 32 decimal number k₁, k₂..., k₃₂, then calculate the parameter and just of one-dimensional skew tent chaotic maps Initial value, and the initial position co-ordinates that zigzag scrambles need comprise the following steps that：

1.1) expression formula of one-dimensional skew tent chaotic maps that the present invention is adopted for：

Wherein, r ∈ (0, be 1) systematic parameter, z ∈ (0,1) be chaos system state variable.

1.2) plaintext image is calculated using the hash functions of SHA 256, the cryptographic Hash of a group 256 can be obtained (hexadecimal representation is：[b32171537de519d7c1c83ff25989a559532c6b269a303ac19d3d0a906bb5 E250]), and using it as image key key, then by the cryptographic Hash of 256 be converted into 32 decimal numbers (179,33, 113,83,125,229,25,215,193,200,63,242,89,137,165,89,83,44,107,38,154), and determined Justice is k₁, k₂..., k₃₂。

1.3) for the ease of the parameter and initial value of the one-dimensional skew tent mappings of calculating, and zigzag scrambles needs Initial position co-ordinates, are first according to following formula and are calculated 4 intermediate parameters t_i(i=1,2,3,4),

Wherein, l_i(i=1,2,3, are 4) given parameters, respectively representation parameter x '₀、y′₀、r′、z′₀, now, x '₀= 0.2796、y′₀=0.7531, r '=0.5678, z '₀=0.8652.Calculated t₃And t₄One-dimensional skew is represented respectively Parameter r and initial value z of tent mappings₀, now, r=0.2122, z₀=0.3539.

1.4) the initial position co-ordinates x that zigzag scrambles need₀And y₀Can be obtained by formula below,

Wherein, abs (x) represents the absolute value for seeking x, and mod is a modulo operation symbol, and result of calculation is as follows：x₀=177 And y₀=260.

Step 2：Sparse process is carried out to plaintext image P using wavelet transform, coefficient matrix P1 is obtained；Then to being Matrix number P1 enters line shuffle according to zigzag paths, and the matrix after scramble is P2；Then one threshold value TS is set, to the unit in P2 Element is modified, and the element less than or equal to TS is changed into 0, and the matrix after process is denoted as comprising the following steps that for P3：

2.1) according to following program, sparse process is carried out to plaintext image P using wavelet transform, obtains coefficient matrix P1；

Ww=DWT (512)；

P1=ww*P*ww'；

2.2) a kind of zigzag paths as shown in Figure 5 are chosen, then according to the initial position co-ordinates that step 1 is calculated (x₀=177 and y₀=260), and line shuffle is entered according to zigzag paths to coefficient matrix P1, the matrix after scramble is P2；

Scramble schematic diagram is as shown in figure 5, zigzag scrambling process is illustrated below：It is (a) before scramble shown in Fig. 6 Matrix, (b) be from (1,1) position that position i.e. element 1 are located come into effect the matrix obtained after zigzag scrambles；Fig. 7 In, it is matrix before scramble to scheme (a), figure (b) be from (3,4) position that position i.e. element 160 are located come into effect The matrix obtained after zigzag scrambles.

2.3) threshold value TS=50 is set, the element in P2 is modified, the element less than or equal to 50 is changed into 0, Matrix after process is denoted as P3.

Step 3：The calculation matrix Φ produced for compressed sensing is mapped using one-dimensional skew tent, then to matrix P3 Measure, obtain measuring comprising the following steps that for value matrix P4：

If 3.1) compression ratio of plaintext image selects to be CR=0.25 in compressed sensing, it is assumed that the size of calculation matrix Φ is M × N, then M=CR × m=128, N=m=512.

3.2)Wherein, sampling interval d=20, r, z₀For the parameter and initial value of one-dimensional skew tent mappings, parameter l=65535, n=500.Specifically, step 1 is counted first Parameter r and initial value z of the one-dimensional skew tent mappings for obtaining₀In substituting into one-dimensional skew tent mappings, iterate, it is raw Into a chaos sequence, in order to avoid the transient effect of chaos, front 499 numerical value of chaos sequence is dropped, then from the 500th Individual numerical value starts, and at interval of 20 numbers, takes out one, obtains the new chaos sequence that length is 65536.

3.3) then according to formulaLine translation is entered to the numerical value of new chaos sequence.

3.4) and then generationArranged by row, it is possible to obtain following calculation matrix Φ, size is 128 × 512,

In formula,For normalized.

3.5) then according to formula below, using calculation matrix Φ, matrix P3 is measured, obtains measuring value matrix P4, size is 128 × 512.

P4=Φ * P3

Step 4：The element of matrix P4 is quantified, element value the integer between 0～255 is converted to, after conversion Matrix is denoted as comprising the following steps that for P5：

4.1) size is turned for the element in 128 × 512 matrix P4 according to order from left to right, from top to bottom first Change a vector into；

4.2) then according to equation below quantifies one by one to vector element,

p_5i=floor (255 × (p_4i-min)/(max-min))

Wherein, min represents the minima of element in matrix P4, is -843.7286；Max represents in matrix P4 element most Big value, is that 891.1269, floor (x) expressions take the no more than maximum integer of x.p_4iRepresent matrix P4 according to from left to right, from upper I-th element (1≤i≤65536) in the vector that order under is obtained, finally p_5iThe vector of composition is converted into matrix P5, size is 128 × 512.

Step 5：The significant image of vision that a size is 512 × 512 is selected as carrier image CI, and it is real to it Wavelet transform is applied, four matrixes are obtained, is respectively low frequency part decomposition coefficient LL, horizontal direction decomposition coefficient LH, vertical Directional Decomposition coefficient HL and diagonal decomposition coefficient HH, is then embedded matrix P5, then carries out discrete wavelet to it Inverse transformation, obtains comprising the following steps that for the significant ciphertext graph picture of final vision：

5.1) element in matrix P5 is processed one by one according to equation below, obtains two matrix Ds 1 and D2.

D1 (i, j)=P5 (i, j) mod10

D2 (i, j)=floor (P5 (i, j)/10)

Wherein, D1 (i, j), D2 (i, j) and P5 (i, j) are represented respectively in matrix D 1, D2 and P5 positioned at the unit at (i, j) place Element, 1≤i≤128,1≤j≤512.

(2) select the Baboon images that size is 512 × 512 as carrier image CI, then matrix D 1, D2 is turned respectively It is changed to the matrix that size is 256 × 256.

(3) Haar wavelet functions are selected, wavelet transform is carried out to carrier image CI according to following program, obtain four Matrix, is respectively low frequency part decomposition coefficient LL, horizontal direction decomposition coefficient LH, vertical direction decomposition coefficient HL and diagonal side To decomposition coefficient HH, their size is all 256 × 256.

[LL, LH, HL, HH]=dwt2 (CI, ' haar')；

(4) the empty matrix IT that size is for 256 × 256 is defined, the average of all elements in LL matrixes is then calculated Value MV=258.2940, then according to following rule is in D1, D2 embeded matrix LH and HL.

Rule 1：If LL (i, j) >=MV, then IT (i, j)=1, LH (i, j)=D1 (i, j), HL (i, j)=D2 (i, j).

Rule 2：If LL (i, j)<MV, then IT (i, j)=0, LH (i, j)=D2 (i, j), HL (i, j)=D1 (i, j).

In above-mentioned rule, 1≤i≤256,1≤j≤256.

(5) Haar wavelet functions are then used by, discrete is implemented to four matrix Ls L, LH, HL and HH according to following program Inverse wavelet transform, obtains the significant ciphertext graph of a pair vision as C, and ciphering process terminates.

C=idwt2 (LL, LH, HL, HH, ' haar').

Image decryption process is the inverse process of ciphering process.Before image decryption, decryption side needs to obtain decryption key, Specifically include：(hexadecimal representation is 256 cryptographic Hash key of plaintext image

[b32171537de519d7c1c83ff25989a559532c6b269a303ac19d3d0a90 6bb5e250]), 6 Individual parameter x '₀=0.2796, y '₀=0.7531, r '=0.5678, z '₀=0.8652, min=-843.7286 and max= 891.1269.Decryption step is as follows：

Step 1：256 cryptographic Hash key are scaled into 32 decimal number k₁, k₂..., k₃₂(179,33,113,83, 125,229,25,215,193,200,63,242,89,137,165,89,83,44,107,38,154), then according to following formula meter Calculation obtains 4 intermediate parameters t_i(i=1,2,3,4),

Wherein, l_i(i=1,2,3, are 4) given parameters, respectively representation parameter x '₀、y′₀、r′、z′₀, now, x '₀= 0.2796、y′₀=0.7531, r '=0.5678, z '₀=0.8652.Calculated t₃And t₄One-dimensional skew is represented respectively Parameter r and initial value z of tent mappings₀, now, r=0.2122, z₀=0.3539.

Step 2：The initial position co-ordinates x that zigzag scrambles need₀And y₀Can be obtained by formula below,

Wherein, abs (x) represents the absolute value for seeking x, and mod is a modulo operation symbol, and result of calculation is as follows：x₀=177 And y₀=260.

Step 3：Assume that ciphertext graph picture to be decrypted is C, size is 512 × 512, Haar wavelet functions is selected, according to such as Lower program implements wavelet transform to ciphertext graph picture, obtains the matrix that 4 sizes are 256 × 256, is respectively low frequency part point Solution coefficient LL1, horizontal direction decomposition coefficient LH1, vertical direction decomposition coefficient HL1 and diagonal decomposition coefficient HH1.

[LL1, LH1, HL1, HH1]=dwt2 (C, ' haar')；

Step 4：The empty matrix IT1 that a size is for 256 × 256 is defined, then all elements of calculating matrix LL1 is flat Average MV1=258.2940, extracts plaintext relevant information from LH1 and HL1 according to following rule, obtains matrix D 11 And D21.

Rule 1：If LL1 (i, j) >=MV1, then IT1 (i, j)=1, D11 (i, j)=LH1 (i, j), D21 (i, j)= HL1(i,j).Rule 2：If LL1 (i, j)<MV1, then IT1 (i, j)=0, D21 (i, j)=LH1 (i, j), D11 (i, j)= HL1(i,j).In above-mentioned rule, 1≤i≤256,1≤j≤256.

Step 5：Matrix D 11, D21 is transformed to the matrix that size is 128 × 512, equation below is then passed through and is obtained square Battle array P51,

P51 (i, j)=D21 (i, j) × 10+D11 (i, j)

Wherein, D11 (i, j), D21 (i, j) and P51 (i, j) represent respectively matrix D 11, D21 and P51 positioned at (i, j) place Element, 1≤i≤128,1≤j≤512.

Step 6：Matrix P51 is converted into the vector that length is 1 × 65536 according to order from left to right, from top to bottom p_51i(1≤i≤65536), inverse quantization operation is implemented one by one then according to equation below to vector element, obtains p_41i(1≤i≤ 65536), then vector changed into 128 × 512 matrix P41.

Wherein, min and max are the keys obtained from sender.

Step 7：Parameter r and initial value z of the one-dimensional skew tent mappings obtained using step 1₀, according to encrypting step 3 Identical method obtains calculation matrix Φ 1, then selects a kind of conventional algorithm for reconstructing of compressed sensing：Orthogonal matching pursuit is calculated Method (OMP) obtains matrix P31 from P41 reconstruction signals, and size is 512 × 512.Step 8：The zigzag obtained using step 1 is put Random initial position co-ordinates x₀And y₀Implement inverse zigzag conversion to matrix P31, obtain matrix P11.

Step 9：Implement discrete wavelet according to following program to matrix P11 to operate against sparse transformation, it is possible to decrypted Plaintext image P afterwards, size is 512 × 512.

Ww=DWT (512)；

P=ww'*P1*ww；

The AES that the present invention is provided has very big key space, it is sufficient to resist any brute force attack.Specifically, it is secret Key includes：(1) 256 cryptographic Hash produced by the hash functions of SHA 256；(2) given parameter x '₀, y '₀, r ', z '₀.Additionally, Can choose at random with carrier image equivalently-sized in plain text, when carrying out wavelet transform to carrier image, existing small echo Function has 37 kinds of different selections.If the computational accuracy of computer is 10^-14, then the key space of this paper algorithms be at least 37 ×2²⁹⁶, it is seen that key space is sufficiently large, can effectively resist any brute force attack.

In Fig. 4：A () is original image used by example one, (b) be the ciphertext graph picture after compression encryption, (c) is carrier figure Picture, is (d) the ciphertext graph picture of visual security, (e) is correct key decrypted image, (f) is solution when being decrypted using wrong key Close image.As shown in Figure 4, after being encrypted to plaintext image (a) according to encipherment scheme of the present invention, ciphertext graph picture (d) for obtaining It is the significant ciphertext graph picture of a width vision, when it is in transmission over networks and storage, visually with carrier image (c) phase Together, found and cracked by attacker so as to avoid, increased transmission and the safety for storing.Simultaneously according to decryption side of the present invention Case, using correct key, the decrypted image (e) for obtaining is essentially the same with plaintext image (a), can correctly decrypt, and uses wrong Miss key (x₀'=0.2796+10^-14, other keys are constant), decrypted image (f) class for obtaining is noise-like, and this illustrates this The encryption method that bright scheme is proposed has very high susceptiveness to key, and algorithm has very high safety.

Example two, shown in Figure 8, in the present embodiment, the programming software for adopting is Matlab R2014a, chooses big It is little be 256 × 256 Finger gray level images as experimental subject, concrete ciphering process is as follows：

Step 1：Size is calculated for 256 × 256 plaintext image P using the hash functions of SHA 256, obtains one group (hexadecimal representation is the cryptographic Hash of 256：

[0f2f217b7ac7eabce5c26ed2333b7b44d8f8c0ff889e05aa0918d fa715e3396d]), And using it as image key key, then by key be scaled 32 decimal numbers (15,47,33,123,122,199,234, 188,229,194,110,210,51,59,123,68,216,248,192,255,136,158,5,170,9,24,223,167, 21,227,57,109), and it is defined as k₁, k₂..., k₃₂.4 key parameters are set to, x '₀=0.2796, y '₀= 0.7531, r '=0.5678, z '₀=0.8652, parameter r=0.3602 of one-dimensional skew tent chaotic maps is then calculated, Initial value z₀=0.189, and the initial position co-ordinates that zigzag scrambles need, x₀=153, y₀=227.

Step 2：Sparse process is carried out to plaintext image P using wavelet transform, coefficient matrix P1 is obtained；Then to being Matrix number P1 enters line shuffle according to zigzag paths, and the matrix after scramble is P2；Then one threshold value TS=50 is set, in P2 Element be modified, being changed into 0 less than or equal to the element of TS, the matrix after process is denoted as P3.

Step 3：The compression ratio for arranging plaintext image is CR=0.25, is produced for pressing using one-dimensional skew tent mappings The calculation matrix Φ that contracting is perceived, size is 64 × 256, and then matrix P3 is measured, and obtains measuring value matrix P4, and size is 64×256。

Step 4：The minima for calculating element in matrix P4 is min=-933.5953, and the maximum of element is max= =805.7708, then the element of matrix P4 is quantified, element value is converted to the integer between 0～255, after conversion Matrix is denoted as P5, and size is 64 × 256, as the ciphertext graph picture after compression encryption.

Step 5：Select the Cameraman that size is 256 × 256 as carrier image CI, discrete wavelet transformer is carried out to it Change, obtain four matrixes, be respectively low frequency part decomposition coefficient LL, horizontal direction decomposition coefficient LH, vertical direction decomposition coefficient HL and diagonal decomposition coefficient HH, is then embedded matrix P5, then carries out discrete wavelet inverse transformation to it, just obtains The significant ciphertext graph of final vision is as C.

Image decryption process is the inverse process of ciphering process.Before image decryption, decryption side needs to obtain decryption key, Specifically include：256 cryptographic Hash key of plaintext image, 6 parameter x '₀、y′₀、r′、z′₀, min and max.Decryption step is as follows：

Step 1：By cryptographic Hash key of 256 be scaled 32 decimal numbers (15,47,33,123,122,199,234, 188,229,194,110,210,51,59,123,68,216,248,192,255,136,158,5,170,9,24,223,167, 21,227,57,109), and it is defined as k₁, k₂..., k₃₂.4 key parameters are, x '₀=0.2796, y '₀=0.7531, R '=0.5678, z '₀=0.8652, then calculate parameter r=0.3602 of one-dimensional skew tent chaotic maps, initial value z₀ =0.189, and the initial position co-ordinates that zigzag scrambles need, x₀=153, y₀=227.

Step 2：Assume that ciphertext graph picture to be decrypted is C, size is 256 × 256, and to it wavelet transform is implemented, and is obtained To 4 sizes for 128 × 128 matrix, be respectively low frequency part decomposition coefficient LL1, horizontal direction decomposition coefficient LH1, vertical Directional Decomposition coefficient HL1 and diagonal decomposition coefficient HH1.

Step 3：The empty matrix IT1 that a size is for 128 × 128 is defined, then all elements of calculating matrix LL1 is flat Average MV1, according to following rule matrix D 11 and D21 are obtained from LH1 and HL1.

Rule 1：If LL1 (i, j) >=MV1, then IT1 (i, j)=1, D11 (i, j)=LH1 (i, j), D21 (i, j)= HL1(i,j)。

Rule 2：If LL1 (i, j)<MV1, then IT1 (i, j)=0, D21 (i, j)=LH1 (i, j), D11 (i, j)=HL1 (i,j).In above-mentioned rule, 1≤i≤128,1≤j≤128.

Step 4：Matrix D 11, D21 is transformed to the matrix that size is 64 × 256, equation below is then passed through and is obtained matrix P51,

P51 (i, j)=D21 (i, j) × 10+D11 (i, j)

Wherein, D11 (i, j), D21 (i, j) and P51 (i, j) represent respectively matrix D 11, D21 and P51 positioned at (i, j) place Element, 1≤i≤64,1≤j≤256.

Step 5：Matrix P51 is converted into the vector that length is 1 × 16384 according to order from left to right, from top to bottom p_51i(1≤i≤16384), inverse quantization operation is implemented one by one then according to equation below to vector element, obtains p_41i(1≤i≤ 16384), then vector changed into 64 × 256 matrix P41.

Wherein, min and max are the keys obtained from sender.

Step 6：Parameter r and initial value z of the one-dimensional skew tent mappings obtained using step 1₀, according to encrypting step 3 Identical method obtains calculation matrix Φ 1, followed by orthogonal matching pursuit algorithm (OMP) from P41 reconstruction signals, obtains matrix P31, size is 256 × 256.

Step 7：The initial position co-ordinates x of the zigzag scrambles obtained using step 1₀And y₀Inverse is applied to matrix P31 Zigzag is converted, and obtains matrix P1, is then implemented discrete wavelet to matrix P1 and is operated against sparse transformation, it is possible to after being decrypted Plaintext image P, size be 256 × 256.

Concrete steps are referred to described in example one.

In Fig. 8：A () is original image used by example two, (b) be the ciphertext graph picture after compression encryption, (c) is carrier figure Picture, is (d) the ciphertext graph picture of visual security, (e) is correct key decrypted image, (f) is solution when being decrypted using wrong key Close image.As shown in Figure 8, after being encrypted to plaintext image (a) according to encipherment scheme of the present invention, ciphertext graph picture (d) for obtaining It is the significant ciphertext graph picture of a width vision, when it is in transmission over networks and storage, visually with carrier image (c) phase Together, found and cracked by attacker so as to avoid, increased transmission and the safety for storing.Simultaneously according to decryption side of the present invention Case, using correct key, the decrypted image (e) for obtaining is essentially the same with plaintext image (a), can correctly decrypt, and uses wrong Miss key (y '₀=0.7531+10^-14, other keys are constant), decrypted image (f) class for obtaining is noise-like, and this illustrates this The encryption method that bright scheme is proposed is extremely sensitive to key, and algorithm has very high safety.

From above-described embodiment, the AES that the present invention is provided can carry out visual security and data peace to gray level image Full double-encryption, has broad application prospects in field of information encryption.

Above-mentioned specific embodiment is the invention is not limited in, those skilled in the art can also accordingly make various changes, But it is any all to cover within the scope of the claims with equivalent of the invention or similar change.

Claims (10)

1. a kind of image encryption method of the visual security and data safety based on compressed sensing, it is characterised in that：Comprising as follows Step：

Plaintext image P of the step 1. using the hash functions of SHA 256 to size for m × n is calculated, and obtains image key key, And calculate the initial value of one-dimensional skew tent chaotic maps and zigzag scrambles；

Step 2. carries out sparse process using wavelet transform to plaintext image P, obtains coefficient matrix P1；To coefficient matrix P1 Enter line shuffle according to zigzag paths, obtain the matrix P2 after scramble；By predetermined threshold value TS, matrix P2 is modified, is obtained To revised matrix P3；

Step 3. utilizes one-dimensional skew tent chaotic maps to generate the calculation matrix Φ for compressed sensing, and matrix P3 is entered Row measurement, obtains measuring value matrix P4, wherein, the compression ratio of plaintext image P is expressed as CR, the size of calculation matrix Φ be M × The size of N, then M=CR × m, N=m, matrix P4 is M × n；

Step 4. quantifies to the element of matrix P4, obtains matrix P5, as the ciphertext graph picture after compression encryption；

It is the visual pattern of m × n as carrier image CI that step 5. selects size, and implements wavelet transform to it, Obtain four matrixes：Low frequency part decomposition coefficient LL, horizontal direction decomposition coefficient LH, vertical direction decomposition coefficient HL and diagonal Directional Decomposition coefficient HH, is embedded matrix P5, and discrete wavelet inverse transformation is carried out again, obtains final vision significant Ciphertext graph picture.

2. the image encryption method of the visual security and data safety based on compressed sensing according to claim 1, it is special Levy and be：The initial value of one-dimensional skew tent chaotic maps and zigzag scrambles is calculated in the step 1, comprising：By image Key key is scaled 32 decimal number k₁, k₂..., k₃₂, and calculate the parameter of one-dimensional skew tent chaotic maps and initial Value, and the initial position co-ordinates of zigzag scrambles.

3. the image encryption method of the visual security and data safety based on compressed sensing according to claim 2, it is special Levy and be：One-dimensional skew tent chaotic maps are expressed as：

$z(k+1)=T\&lsqb;z\left(k\right);r\&rsqb;=\left\{\begin{array}{c}\frac{z\left(k\right)}{r},0<z\left(k\right)<r\\ \frac{1-z\left(k\right)}{1-r},r\&le;z\left(k\right)<1\end{array}\right.$ ,

Wherein, r ∈ (0, be 1) systematic parameter, z ∈ (0,1) be chaos system state variable；And according to formula：

$t_i=\&lsqb;l_i+\left(\frac{k_{4i-2}+\left(4i-2\right)}{k_{3i}+3i}+\frac{k_{5i-3}+\left(5i-3\right)}{k_{4i}+4i}\right)\&times;\underset{j=1}{\overset{32}{\mathrm{\&Sigma;}}}\frac{k_j\&times;2^{8\&times;\left(j-1\right)}}{2^{256}}\&rsqb;\mathrm{mod}1,i=1,2,3,4$

To calculate 4 intermediate parameters t_i(i=1,2,3,4), wherein, l_i(i=1,2,3,4) are preliminary setting parameter, represent respectively Parameter x '₀, y '₀, r ', z '₀, calculated t₃And t₄Parameter r and initial value of one-dimensional skew tent chaotic maps are represented respectively z₀；The initial position co-ordinates x of zigzag scrambles₀And y₀By formula：

It is calculated, wherein, abs (x) represents the absolute value for seeking x, and mod is modulo operation.

4. the image encryption method of the visual security and data safety based on compressed sensing according to claim 1, it is special Levy and be：Matrix P2 is modified in the step 2, is specifically included：By predetermined threshold value TS, the element in P2 is repaiied Just, the element less than or equal to TS is changed into 0.

5. the image encryption method of the visual security and data safety based on compressed sensing according to claim 1, it is special Levy and be：The one-dimensional skew tent chaotic maps that utilize in step 3 generate the calculation matrix Φ for compressed sensing, comprising such as Lower content：

Step 3.1, orderWherein, d represents the sampling interval, r,z₀The parameter mapped for one-dimensional skew tent and initial value, l represents sampling length, and n represents Sampling starting point, step 1 will be counted Parameter r and initial value z of the one-dimensional skew tent chaotic maps for obtaining₀It is raw in substituting into one-dimensional skew tent chaotic maps Into chaos sequence；

Step 3.2, according to formulaLine translation is entered to the numerical value of chaos sequence；

Step 3.3, generateArranged by row, obtained calculation matrix, size is M × N,

In formula,For normalized.

6. the image encryption method of the visual security and data safety based on compressed sensing according to claim 5, it is special Levy and be：Following content is also included in the step 3.1：The front n-1 numerical value of chaos sequence is abandoned, is opened from n-th numerical value Begin, at interval of d numbers, take out one, obtain new sequence of the length for MN, as the new chaos sequence replaced.

7. the image encryption method of the visual security and data safety based on compressed sensing according to claim 1, it is special Levy and be：Element in the step 4 to size for the matrix P4 of M × n quantifies, specifically comprising following content：By matrix Element in P4 is converted into a vector according to order from left to right, from top to bottom, according to formula：

p_5i=floor (255 × (p_4i-min)/(max-min))

Vector element is quantified one by one, wherein, min represents the minima of element in matrix P4, and max represents unit in matrix P4 The maximum of element, floor (x) is represented and is taken the no more than maximum integer of x, p_4iMatrix P4 is represented according to from left to right, from top to bottom The vector that obtains of order in i-th element (1≤i≤Mn), by p_5iThe vector of composition is converted into the matrix that size is M × n P5。

8. the image encryption method of the visual security and data safety based on compressed sensing according to claim 1, it is special Levy and be：The step 5 specifically includes following content：

Step 5.1, according to formula：

D1 (i, j)=P5 (i, j) mod10, D2 (i, j)=floor (P5 (i, j)/10)

Element in matrix P5 is processed one by one, matrix D 1 and D2 is obtained, wherein, D1 (i, j), D2 (i, j) and P5 (i, j) Represent respectively in matrix D 1, D2 and P5 positioned at the element at (i, j) place, 1≤i≤M, 1≤j≤n；

Step 5.2, according to compression ratio CR of plaintext image P, the size of matrix D 1, D2 and P5 is set；

Step 5.3, wavelet transform is carried out to carrier image CI, obtain four matrixes, be expressed as：Low frequency part is decomposed Coefficient LL, horizontal direction decomposition coefficient LH, vertical direction decomposition coefficient HL and diagonal decomposition coefficient HH, their size Size is identical with matrix D 1, D2 sizes；

Step 5.4, one size of definition and matrix D 1, D2 size identical empty matrix IT, in calculating LL matrixes Meansigma methodss MV of all elements, according to rule：

Rule 1：If LL (i, j) >=MV, then IT (i, j)=1, LH (i, j)=D1 (i, j), HL (i, j)=D2 (i, j),

Rule 2：If LL (i, j)<MV, then IT (i, j)=0, LH (i, j)=D2 (i, j), HL (i, j)=D1 (i, j), D1, In D2 embeded matrix LH and HL, wherein, 1≤i≤m/2,1≤j≤n/2；

Step 5.5, to be embedded in after four matrix Ls L, LH, HL and HH carry out discrete wavelet inverse transformation, obtain vision significant Ciphertext graph picture, ciphering process terminates.

9. a kind of image decryption method of the visual security and data safety based on compressed sensing, it is characterised in that：Specifically include Step is as follows：

Step a：Obtain image key key of plaintext image P, and parameter x '₀、y′₀、r′、z′₀, min and max；

Step b：Image key key is scaled into 32 decimal number k₁, k₂..., k₃₂, and calculate one-dimensional skew tent chaos and reflect The parameter penetrated and initial value, and the initial position co-ordinates of zigzag scrambles；One-dimensional skewtent chaotic maps are expressed as：

$z(k+1)=T\&lsqb;z\left(k\right);r\&rsqb;=\left\{\begin{array}{c}\frac{z\left(k\right)}{r},0<z\left(k\right)<r\\ \frac{1-z\left(k\right)}{1-r},r\&le;z\left(k\right)<1\end{array}\right.$ ,

Wherein, r ∈ (0, be 1) systematic parameter, z ∈ (0,1) be chaos system state variable；And according to formula：

$t_i=\&lsqb;l_i+\left(\frac{k_{4i-2}+\left(4i-2\right)}{k_{3i}+3i}+\frac{k_{5i-3}+\left(5i-3\right)}{k_{4i}+4i}\right)\&times;\underset{j=1}{\overset{32}{\mathrm{\&Sigma;}}}\frac{k_j\&times;2^{8\&times;\left(j-1\right)}}{2^{256}}\&rsqb;\mathrm{mod}1,i=1,2,3,4$

To calculate 4 intermediate parameters t_i(i=1,2,3,4), wherein, l_i(i=1,2,3,4) are preliminary setting parameter, represent respectively Parameter x '₀, y '₀, r ', z '₀, calculated t₃And t₄Parameter r and initial value of one-dimensional skew tent chaotic maps are represented respectively z₀；The initial position co-ordinates x of zigzag scrambles₀And y₀By formula：

It is calculated, wherein, abs (x) represents the absolute value for seeking x, and mod is modulo operation；

Step c：To size to be decrypted for m × n ciphertext graph as C, carry out wavelet transform, obtain 4 sizes for (m/2) The matrix of × (n/2), respectively：Low frequency part decomposition coefficient LL1, horizontal direction decomposition coefficient LH1, vertical direction decomposition coefficient HL1 and diagonal decomposition coefficient HH1；

Step d：Define size for (m/2) × (n/2) empty matrix IT1, meansigma methodss MV1 of all elements of calculating matrix LL1, According to rule：

Regular A：If LL1 (i, j) >=MV1, then IT1 (i, j)=1, D11 (i, j)=LH1 (i, j), D21 (i, j)=HL1 (i, J),

Regular B：If LL1 (i, j)<MV1, then IT1 (i, j)=0, D21 (i, j)=LH1 (i, j), D11 (i, j)=HL1 (i, J), cleartext information is extracted from LH1 and HL1, obtains matrix D 11 and D21, wherein, 1≤i≤m/2,1≤j≤n/2；

Step e：Matrix D 11, D21 is transformed to into the matrix that size is (m/4) × n, and by formula：P51 (i, j)=D21 (i, J) × 10+D11 (i, j), obtains matrix P51, wherein, D11 (i, j), D21 (i, j) and P51 (i, j) represent respectively matrix D 11, D21 and P51 is located at the element at (i, j) place, 1≤i≤m/4,1≤j≤n；

Step f：Matrix P51 is converted into the vector that length is 1 × (mn/4), p according to order from left to right, from top to bottom_51i (1≤i≤mn/4) is i-th element of vector, according to formula：

Inverse quantization operation is carried out one by one to vector element, p is obtained_41i(1≤i≤mn/ 4), vector is changed into the matrix P41 that size is (m/4) × n；

Step g：The calculation matrix Φ 1 for compressed sensing is generated using one-dimensional skew tent chaotic maps, using compressed sensing Algorithm for reconstructing is reconstructed to P41, obtains matrix P31 of the size for m × n；

Step h：Using the calculated x of step b₀And y₀Implement inverse zigzag conversion to matrix P31, matrix P11 is obtained, to square Battle array P11 carries out discrete wavelet and operates against sparse transformation, the plaintext image P after being decrypted.

10. the image decryption method of the visual security and data safety based on compressed sensing according to claim 9, it is special Levy and be：The one-dimensional skew tent chaotic maps that utilize in step g generate the calculation matrix Φ 1 for compressed sensing, bag Containing following content：

Step g1, orderWherein, d represents the sampling interval, r,z₀The parameter mapped for one-dimensional skew tent and initial value, l represents sampling length, and n represents Sampling starting point, and step b is calculated Parameter r and initial value z of the one-dimensional skew tent chaotic maps for obtaining₀In substituting into one-dimensional skew tent chaotic maps, generate Sequence, the front n-1 numerical value of sequence is abandoned, and from the beginning of n-th numerical value, at interval of d numbers, takes out one, and obtaining length is The new sequence of MN, as chaos sequence；

Step g2, according to formulaLine translation is entered to the numerical value of chaos sequence；

Step g3, generateArranged by row, obtained calculation matrix, size is M × N,

In formula,For normalized.