CN107845055B

CN107845055B - Image encryption method for resisting Facebook compression

Info

Publication number: CN107845055B
Application number: CN201710579130.9A
Authority: CN
Inventors: 殷赵霞; 刘晓庆
Original assignee: Anhui University
Current assignee: Anhui University
Priority date: 2017-07-17
Filing date: 2017-07-17
Publication date: 2020-10-16
Anticipated expiration: 2037-07-17
Also published as: CN107845055A

Abstract

The invention discloses an image encryption method for resisting Facebook compression, which sequentially comprises the following steps: 8 multiplied by 8 blocking is carried out on the image; performing discrete cosine transform and quantization on each block; encrypting the image DC coefficient and the AC coefficient; encrypting the image by taking the image block as a unit; performing border crossing processing; and (5) carrying out secondary encryption. The method combines the characteristics of the Facebook social network platform on image processing, is simple to realize, has high safety, can effectively resist various attacks, and has wide application prospect on image encryption protection based on the Facebook social network platform.

Description

Image encryption method for resisting Facebook compression

Technical Field

The invention relates to an image compression and encryption technology, in particular to an image encryption method for resisting Facebook compression.

Background

In the 21 st century informatization age, with the development of the internet and the popularization of mobile devices, social networks have become an inseparable part of our lives, and the mainstream social network platforms in the world now include Google +, Facebook, WeChat and Twitter, etc. However, as social networks are widely used, people are concerned about personal privacy security more and more, and pictures are taken as a main information communication medium, hundreds of millions of pictures are uploaded every day on mainstream social network platforms such as Facebook, and the pictures contain a large amount of privacy information including the appearance, geographical position and even business secret information of people. Privacy protection in social networks is therefore of increasing interest. With the development of image encryption technology, image encryption analysis technology has also advanced greatly. General encryption technology can easily obtain a part of image information under external attack, so that at present, a safer and more effective image encryption method is adopted, and in reality, users of a social network platform have great demands on privacy protection. The global user volume of Facebook in 12 months of 2016 exceeded 23 hundred million, becoming the largest social networking site worldwide.

Facebook has a unique way to handle uploaded images and thus also presents a significant difficulty for image encryption research based on Facebook social networking platform. When images are uploaded to Facebook, for oversizeJPEG (Joint Photographic Experts group) is a committee which sets forth the compression standard for still images under the international organization for standardization (ISO) leader, JPEG (Joint Photographic Experts group) is a committee which has succeeded in short years, and is widely used in the fields of Internet and digital cameras, JPEG (Joint Photographic Experts group) being a compression standard for JPEG that 80% of images on websites employ the JPEG compression standard which only describes how to convert an image into a data stream of bytes, but does not describe how the bytes are sealed on any specific storage medium, JPEG compression standard being a lossy compression format which can compress images in a small storage space, JPEG compression standard reading the pixel values of an image first, quantizing the image into small blocks of 398 blocks, and quantizing each block of the image into small blocks of discrete cosine (D) after each discrete cosine transform (8 × 8) of each block of DCT block, and then quantizing each block of the DCT block into a discrete cosine (D) block (each block of DCT)_i,j0 < i, j < 7, wherein D_0,0Referred to as DC coefficients, the remainder are AC coefficients. Differential encoding is used for the DC coefficients, and run-length encoding and entropy encoding are used for the AC coefficients.

According to the experimental analysis of images in 1038 UCID-v2 image databases, Facebook selects the quality factor of image compression between 71 and 92, because the correlation of the encrypted images is very low, namely the activity of the images is very high, and according to the characteristics of Facebook processing, the images with high activity adopt low compression factor, so that the quantization table with the quality factor of 71 is adopted to compress and quantize the images.

For the image encryption technology based on the Facebook social network platform, the image encryption technology can only ensure that the image encryption technology can play a role in keeping secret when not being attacked by the outside, and can protect information in the image, but can not play a good role when receiving some outside attacks, and can not protect the information in the image from being leaked. Sun et al in 2016 proposed a Facebook social networking platform based image encryption technique that enabled correct decryption of images after Facebook compression. The quality of the image can be well guaranteed when the image is decrypted by the method, but the encryption security is not high, and the basic outline of the image can be exposed under some attack means.

In the existing method, for image encryption of a Facebook social network platform, the encryption effect of image information is relatively high, but the attack resistance is low, because Facebook reads from a pixel domain side when processing an uploaded image, and then JPEG compression is performed, in the current encryption scheme, only water washing encryption and AC coefficient exclusive or encryption are respectively performed on an image DC coefficient, because after image DCT transformation and quantization, exclusive or encryption is performed according to an equivalent binary data stream, the positions of a non-zero AC coefficient and a last non-zero AC coefficient do not change, and the size of a small AC coefficient does not change too much, so that a basic outline of the image can be obtained through attack on the AC coefficient.

Disclosure of Invention

The purpose of the invention is as follows: the invention aims to solve the defects in the prior art and provide an image encryption method for resisting Facebook compression.

The technical scheme is as follows: the invention relates to an image encryption method for resisting Facebook compression, which comprises two steps of encryption and decryption;

(1) the specific process of encryption is as follows:

(1.1) partitioning an image I to be processed, which is given a size of H × W, into blocks of 8 × 8 size, and totally partitioning the blocks

Each image block, each block being DCT transformed and a quantization with a quality factor of 71 being selectedThe table quantizes each block, and in each block, the quantized coefficient D ═ D _i,j0 < i, j < 7, wherein D_0,0Called the DC coefficient, and the rest are AC coefficients;

(1.2) different encryption modes controlled by a secret key K are respectively adopted for the DC coefficient and the AC coefficient: the DC coefficient is changed by adopting an encryption mode of circularly shifting controlled by a key K, the DC coefficient of each block is changed, and a stream key generated by the key K is adopted for the AC coefficient, and a non-zero AC coefficient and a block stream password are used for carrying out bitwise XOR encryption;

(1.3) for pixel blocks having pixel values outside the range of the original pixel values, transmitting them to the receiver using bitmap marking compression, and multiplying the AC coefficients of the out-of-range image blocks by a contraction coefficient

The shrinkage treatment is carried out to the mixture,

(1.4) taking the image block as a basic unit, and carrying out scrambling encryption operation on the whole image according to the key K;

(2) the specific process of decryption is as follows:

(2.1) restoring the image block to the correct block position according to the secret key K;

(2.2) multiplying the AC coefficient of the out-of-bounds block against the bitmap

And (2.3) decrypting the DC coefficient and the AC coefficient of each block according to different decryption modes: the DC coefficient employs an inverse cyclic shift operation determined by the key K, and a stream cipher is generated from the key K, and the image data can be decrypted by bitwise xoring the packets with the encrypted non-zero AC coefficient.

Further, the specific method of the step (1.2) is as follows:

(1.2.1) count all DC coefficients to DCoe ═ DC in the progressive scanning order₀,DC₁,......,DC_n-1One-dimensional matrixPerforming the following steps;

(1.2.2) is provided with

Converting a one-dimensional matrix DCoe containing n DC coefficients into

A two-dimensional matrix DCoe' of rows and 4 columns;

(1.2.3) performing cyclic shift operation on the DC coefficient matrix DCoe 'according to the secret key K, wherein the DCoe' after cyclic shift is used as the DC coefficient of each block in sequence;

(1.2.4) generating a binary data stream KA by using a key K, grouping the binary stream keys KA into groups, recording the groups as KA (0) and KA (1),... said.. each group of stream keys has a length of 12, sequentially carrying out bitwise XOR encryption on non-zero AC coefficients, setting a binary system of each non-zero AC coefficient to be l bits, and carrying out encryption on the encrypted AC coefficients

Since the maximum length of the non-zero AC coefficient in the image is 10, the stream key 12 bits are grouped and far exceed the binary length of the AC coefficient, and when the AC coefficient is attacked to change the AC length, the data robustness can be ensured to the maximum extent.

Further, the method for scrambling and encrypting the whole image according to the key K by using the image block as a basic unit in the step (1.4) is as follows:

(1.4.1) image I to be processed with size H × W, the image is partitioned into blocks according to 8 × 8 size, and the blocks are totally divided into

Blocks, denoted as n blocks;

(1.4.2) taking out t blocks at equal intervals in sequence each time: the image I has n blocks in total, every other time for the first time

Taking out one block, taking out t blocks, and remaining n-t blocks in the image according to the first method at intervals

Taking out one block, and the same way is carried out until all blocks are taken out;

(1.4.3) finally arranging the blocks according to the extraction sequence as the positions of the encrypted blocks.

Further, a specific method for restoring the image block to the correct block position in step (2.1) is as follows:

(2.1.1) setting the position marker matrix Loc to {1, 2.... multidot.n } for an image containing n 8 × 8 blocks;

(2.1.2) taking out K values at equal intervals in the Loc according to the key K and the sequence every time, and sequentially putting the K values in the Loc' matrix until all elements are taken out;

(2.1.3) restoring the position of the ith block in the image to the position of the Loc '(i) block according to the Loc' position matrix, thereby realizing the restoration of the position of the image block.

Further, the specific method for decrypting the DC and AC coefficients of the image in the step (2.3) in different manners is as follows:

(2.3.1) counting all DC coefficients in order to DCoe ═ DC₀,DC₁,......,DC_n-1In the one-dimensional matrix;

(2.3.2) converting the one-dimensional matrix DCoe containing n DC coefficients into

A two-dimensional matrix DCoe' of rows and 4 columns;

(2.3.3) performing reverse cyclic shift operation on the DC coefficient matrix DCoe according to the secret key K, wherein the DCoe after cyclic shift is used as the DC coefficient after each block is encrypted in sequence;

(2.3.4) generating a binary data stream KA by the key K, grouping the binary stream keys, recording the binary stream keys as KA (0) and KA (1), and carrying out bitwise XOR operation on the nonzero AC coefficients in sequence when the length of each group of keys is 12, and setting the binary length of the nonzero AC coefficients as l, then decrypting the decrypted AC coefficients

According toThis way decryption can be done separately for the DC coefficients and the AC coefficients.

Further, in the step (1.3), the encrypted border-crossing pixel block data is encoded according to a JPEG encoding mode and transmitted to the receiving party.

Has the advantages that: according to the method, the images are encrypted according to the Facebook processing process, in order to avoid the loss of unknown and uncontrollable image information, the size of the images is not required to be too large before uploading, then DCT (discrete cosine transformation) and quantization are carried out on the images to obtain the DC coefficient and the AC coefficient of the images, and the DC coefficient and the AC coefficient of the images are respectively encrypted, so that the images have better safety when resisting external purposeful attacks.

In conclusion, the implementation method is very simple and convenient, is beneficial to encrypting the image transmitted based on the Facebook social network platform, has good adaptability to a real-time communication system, a mobile terminal with limited computing resources and the like due to uncomplicated encryption calculation, and has good application prospect.

Drawings

FIG. 1 is a diagram of an effect of a standard image after encryption in an embodiment;

FIG. 2 is a diagram illustrating the effect of decryption by an error key according to an embodiment;

FIG. 3 is a diagram illustrating the effect of external attacks in an embodiment;

FIG. 4 is a diagram illustrating an embodiment of correctly decrypting a decrypted image;

FIG. 5 is an example AC encryption process data matrix;

FIG. 6 is an embodiment of a DC encryption process data matrix;

FIG. 7 is a diagram illustrating image block scrambling in an embodiment;

FIG. 8 is a schematic overall flow chart of the present invention.

Wherein, fig. 1(a), fig. 1(b) and fig. 1(c) are respectively Lena, Baboon and Elaine encryption effect diagrams; fig. 2(a), fig. 2(b), fig. 2(c) and fig. 2(d) are diagrams of the effect of decrypting 1(a) using 21,22, 24 and 25 error keys, respectively; FIG. 3(a), FIG. 3(b), FIG. 3(c) and FIG. 3(d) are the effect graphs of FIG. 1(a) after encryption of an attack on the image using DCM, NCC, EAC and PLZ, respectively; fig. 4(a), fig. 4(b) and fig. 4(c) are schematic diagrams after decryption of 1(a), 1(b) and 1(c), respectively, with correct keys; FIGS. 5(a), 5(b) and 5(c) are an 8 × 8 image block and a DCT transformed and quantized coefficient matrix and an encrypted AC coefficient matrix, respectively; fig. 6(a) and 6(b) are a matrix containing 16 DC coefficients and a matrix of DC coefficients encrypted with water washing, respectively.

Detailed Description

The technical solution of the present invention is described in detail below, but the scope of the present invention is not limited to the embodiments.

As shown in fig. 8, the image encryption method for resisting Facebook compression of the present invention sequentially includes the following steps: firstly, encrypting, namely performing 8 multiplied by 8 blocking on an image; performing Discrete Cosine Transform (DCT) on each block and quantizing; encrypting the image DC coefficient and the AC coefficient; encrypting the image by taking the image block as a unit; performing border crossing processing; and (5) carrying out secondary encryption. And II, decrypting.

Example 1:

the specific steps of this example are as follows:

firstly, an image encryption process:

an 8 × 8 image block as shown in FIG. 5(a) is subjected to a first encryption operation, first DCT transformed, and then quantized with the selection of a standard quantization table with quality factor 71, resulting in FIG. 5 (b). The total can be divided into H × W sized image partitions

Blocks, each of which is then DCT transformed and quantized in this manner, share

A DC coefficient, and a number of AC non-zero coefficients.

And encrypting the DC coefficients, setting an encryption key K to be 3, arranging all the DC coefficients into a two-dimensional matrix according to the original sequence, circularly moving the DC coefficients downwards for K times, and circularly moving the DC coefficients rightwards for K times to obtain a new coefficient matrix. And taking the new coefficient matrix as the image DC coefficient of each block in sequence. Let fig. 6(a) be a DC coefficient matrix of a certain size 32 × 32, which has 16 DC coefficients in total, and place the 16 coefficient matrices in a two-dimensional DC coefficient matrix of 4 × 4 in sequence, resulting in a new coefficient matrix as in fig. 6 (b). And finally, the DC coefficients of the blocks are sequentially used.

Encrypting the AC coefficients, generating a plurality of binary key streams KA by a key K, grouping the binary key streams KA into groups, recording the binary key streams KA (0) and KA (1), and carrying out bitwise XOR encryption on the non-zero AC coefficients in sequence, setting the binary length of the non-zero AC coefficients as l, and setting the decrypted AC coefficients as

The non-zero AC coefficients as shown in fig. 5(b) are { -2,3,2,1,2,1, -1,1,1, -1, -1} in order according to the Zigzag scanning direction, and after bitwise xor encryption of the packet stream key, the non-zero AC coefficients as shown in fig. 5(c) are { -3, -2, -2, -1,2,1, -1, -1, -1,1}, respectively.

The image blocks are scrambled and encrypted, as shown in fig. 7, with 8 × 8 blocks as the basic unit, and the image blocks are scrambled and rearranged, for example, 256 blocks in fig. 7, every other time

Block, take out one block, take out K blocks, the picture still remains 256-K blocks, according to the first method, every other

And taking out one block, and the same way is carried out until all blocks are taken out. The method can break the original block layout of the image so as to break the outline of the image, and further improve the anti-attack capability of the image.

Example 2:

to detect the effect of the present embodiment, experiments and analyses are performed from three aspects of the encryption effect, the common attack effect, and the image quality.

Taking a common standard test image of 512 × 512 pixels as an example, using the key K ═ 23, the effects of different image encryption, Facebook upload, and download viewing are shown in fig. 1.

If the correct key is not available, fig. 1(a) is decrypted with the wrong keys 21,22 and 24,25 respectively to verify the authenticity of the encryption, and the effect is shown in fig. 2.

The encrypted image shown in fig. 1(a) is subjected to dcm (dc category mapping), NCC (non-zero AC coeffient count), eac (energy of AC coeffients block) and PLZ (position of last non-zero AC coeffients) without a key, and the effect is shown in fig. 3.

If the correct key is available, the effect of decrypting the image is shown in fig. 4.

After the image is encrypted by the method, the original image is invisible, and has good resistance when resisting some attacks, the contour information and the like of the image are not exposed, and the method uses one-time pad and has high safety.

To further illustrate that the image visual effect is not affected after the image information is decrypted by the present invention, table 1 lists the peak signal-to-noise ratio PSNR between the standard test image decryption and the original image. The peak signal-to-noise ratio is an evaluation index of the visual quality of the disguised image, and when the peak signal-to-noise ratio between two gray-scale images is larger than 30dB, human eyes cannot distinguish the two images.

TABLE 1 Standard test image quality after decryption

Claims

1. An image encryption method for resisting Facebook compression is characterized in that: comprises two steps of encryption and decryption;

(1) the specific process of encryption is as follows:

An image block, each block is DCT-transformed, and each block is quantized by selecting a quantization table with a quality factor of 71, and in each block, quantized coefficients D ═ D { (D)_i,j},0≤i, j is less than or equal to 7, wherein D_0,0Called the DC coefficient, and the rest are AC coefficients;

The shrinkage treatment is carried out to the mixture,

(2) the specific process of decryption is as follows:

And (2.3) decrypting the DC coefficient and the AC coefficient of each block according to different decryption modes: the DC coefficient adopts reverse cyclic shift operation determined by a key K, a stream cipher is generated by the key K, and the image data can be decrypted by performing bitwise XOR on the packets and the encrypted non-zero AC coefficient;

the specific method of the step (1.2) comprises the following steps:

(1.2.1) count all DC coefficients to DCoe ═ DC in the progressive scanning order₀,DC₁,......,DC_n-1In the one-dimensional matrix;

(1.2.2) is provided with

Converting a one-dimensional matrix DCoe containing n DC coefficients into

A two-dimensional matrix DCoe' of rows and 4 columns;

(1.2.4) generating a binary data stream KA from a key K, grouping the binary stream keys KA, and recording the binary stream keys KA as KA (0) and KA (1), wherein the length of each stream key is 12, performing bitwise xor encryption on non-zero AC coefficients in sequence, setting a binary sequence of each non-zero AC coefficient to be l bits, and setting an encrypted AC coefficient AC' ═ AC ⊕ KA_1→l(i) Because the maximum length of the non-zero AC coefficient in the image is 10, 12 bits of the stream key are a group which is far beyond the binary length of the AC coefficient, and when the AC coefficient is attacked and the AC length is changed, the robustness of the data can be ensured to the maximum extent;

the specific method for decrypting the image DC and AC coefficients in the step (2.3) in different ways is as follows:

A two-dimensional matrix DCoe' of rows and 4 columns;

(2.3.4) generating a binary data stream KA by the key K, grouping the binary stream keys, recording the binary stream keys as KA (0) and KA (1), and carrying out bitwise XOR operation on the nonzero AC coefficients in sequence when the length of each group of keys is 12, and setting the binary length of the nonzero AC coefficients as l, then decrypting the binary stream keys and carrying out XOR operation on the nonzero AC coefficientsAC coefficient AC ═ AC ⊕ KA_1→l(i) (ii) a In this manner, decryption can be done separately for the DC coefficients and the AC coefficients.

2. The Facebook compression resistant image encryption method of claim 1, wherein: in the step (1.4), the image block is used as a basic unit, and the method for scrambling and encrypting the whole image according to the key K is as follows:

Blocks, denoted as n blocks;

3. The Facebook compression resistant image encryption method of claim 1, wherein: the specific method for restoring the image block to the correct block position in the step (2.1) is as follows:

4. The Facebook compression resistant image encryption method of claim 1, wherein: and (1.3) coding the encrypted border-crossing pixel block data according to a JPEG coding mode and transmitting the coded border-crossing pixel block data to a receiving party.