CN113709323A

CN113709323A - JPEG image processing method, robust steganography method, system and storage medium

Info

Publication number: CN113709323A
Application number: CN202110902976.8A
Authority: CN
Inventors: 田辉; 曾凯; 陈可江
Original assignee: Hefei High Dimensional Data Technology Co ltd
Current assignee: Hefei High Dimensional Data Technology Co ltd
Priority date: 2021-08-06
Filing date: 2021-08-06
Publication date: 2021-11-26

Abstract

The JPEG image processing method comprises the following steps of S11, sequentially carrying out quantization recovery and IDCT conversion processing on first quantization DCT coefficients corresponding to a pre-carrier JPEG image to obtain a corresponding airspace image; s12, carrying out overflow detection on each empty domain block in the empty domain image; s13, performing overflow removal processing on the empty domain blocks with overflow, and not processing the empty domain blocks without overflow; and S14, sequentially carrying out DCT transformation and quantization processing on the spatial domain image after overflow removal to obtain a second quantized DCT coefficient, coding the second quantized DCT coefficient and storing the second quantized DCT coefficient as a JPEG image, namely a carrier image. The steganography method can resist JPEG compression, meanwhile, whether the carrier image is modified or not is difficult to distinguish by the existing steganography analyzer, safety is obviously improved when the steganography method resists detection of various existing steganography analyzers, and the possibility of detecting steganography behavior can be effectively reduced.

Description

JPEG image processing method, robust steganography method, system and storage medium

Technical Field

The invention relates to the technical field of steganography, in particular to a JPEG image processing method, a robust steganography system and a storage medium.

Background

Steganography is a covert communication technology, secret information is embedded by slightly modifying elements on a carrier, the carrier embedded with the secret information is sent to a receiving party, the receiving party can finish correct extraction of the secret information, and one of the most key problems in the process is that the secret information cannot be found by a third party, namely steganography safety; the steganographic analyzer can detect whether the carrier is embedded with the message to a certain degree, so the security of steganographic is generally measured by the capability of the steganographic analyzer to resist steganographic detection, that is, the more difficult the carrier and the carrier after the embedded message are distinguished by the steganographic analyzer, the higher the security of steganographic is.

With the development of the internet, people generally share pictures on a social network, and the steganographic picture transmission in the social network can hide both communication parties and communication behaviors, so that behavior safety is realized, and the social network is a good platform for steganography. But in order to save storage space, the social network usually performs JPEG compression operation on pictures uploaded by users, so the steganographic technology applicable to the social platform also needs to be robust against JPEG compression.

Steganographic techniques that are resistant to JPEG compression are generally less secure, i.e., easily detected by the steganographic analyzer. Therefore, the invention aims to improve the safety of the robust steganography in the actual scene.

First, a general concept or technical term will be introduced.

Pre-carrier: the carrier image is typically generated from an original image, referred to as a pre-carrier, by an image processing operation.

Carrier image: an image for carrying a secret message, the secret message being embedded onto the carrier image in a steganographic process.

Secret message: refers to a message that is pre-encoded using an error correction code.

Carrying out secret image: and embedding the secret message into the carrier image to obtain an image which is the secret image.

Spatial domain images: the spatial domain is also called an image space, and is a space composed of image pixel elements.

JPEG image: jpeg (joint Photographic Experts group), a standard for compression of continuous tone still images, has a file suffix of jpg or jpeg, which is the most commonly used format for image files. The JPEG image stores quantized DCT coefficients obtained by performing block DCT transformation and quantization on spatial domain pixels.

JPEG compression: and saving the spatial domain image as a JPEG image. The method mainly adopts a combined coding mode of predictive coding (DPCM), Discrete Cosine Transform (DCT) and entropy coding to remove redundant images and color data, belongs to a lossy compression format, can compress the images in a small storage space, and can cause damage to the image data to a certain extent.

JPEG recompression: a process of recompressing a JPEG image. Firstly, a space domain image is obtained from a JPEG image through Inverse Discrete Cosine Transform (IDCT) and a corresponding decoding process in a JPEG compression process, and then the space domain image is saved as the JPEG image through the JPEG compression process.

And (3) spatial domain truncation: in the JPEG recompression process, a spatial domain image needs to be obtained from a JPEG image through a corresponding decoding process in JPEG compression and Inverse Discrete Cosine Transform (IDCT), but in the actual calculation process, because rounding errors exist when the JPEG image stores quantized DCT coefficients, the spatial domain value obtained from the DCT coefficients through the IDCT exceeds a storage interval, and cannot be directly stored as a pixel value through rounding operation, and the situation is called spatial domain overflow; at this time, the value less than-128 is set to-128, the value greater than 127 is set to 127, the operation is called spatial truncation, and then other spatial domains are rounded and stored as spatial domain images. The spatial truncation usually causes a large change to the spatial value, so that the DCT coefficient obtained after DCT transformation again changes compared with the original JPEG image.

Distortion function: by quantifying the influence of each pixel point on the carrier image being modified, modification distortion is assigned to each pixel point, which guides the preference of the pixel modification position in the embedding process. In additive steganography conditions, the sum of the effects of all modifications represents the overall distortion between the carrier image and the secret image. The additive steganography refers to that the influence generated by modification of each pixel point is considered independently, and if the influence generated by joint modification between adjacent pixel points is considered, the additive steganography is called as non-additive steganography. Many existing distortion function algorithms for spatial domain images are based on additive steganography conditions, such as HUGO, WOW, S-UNIWARD, HILL, MVGG. These distortion functions give equal distortion to addition, subtraction and modification when a pixel is embedded in a secret message, and are called symmetric distortion functions. On the contrary, the asymmetric distortion function gives the pixel points plus or minus a distortion with unequal modification.

STC coding: the method comprises the steps of syndrome lattice coding, wherein a minimized distortion steganography frame is a mainstream frame for realizing safety steganography, and STC coding can be close to the theoretical lower bound of overall distortion under a given embedding rate aiming at any additive distortion function.

Embedding rate: in JPEG images, the embedding rate is the ratio of the bit length of the embedded information to the number of non-0 DCT coefficients.

Steganalysis device: steganographic security is detected by distinguishing carrier images from secret-bearing images, and existing commonly used steganographic analyzers are divided into two categories: the method comprises the following steps of firstly, extracting a steganalyser such as an SRM (sequence-related language) based on manual features; the other is a network steganalyser based on deep learning, such as SRNet, YeNet and the like.

The general algorithm in the prior art is a robust steganography algorithm based on generalized jitter modulation, wherein the robust steganography algorithm based on generalized jitter modulation is obtained by improving a steganography algorithm based on jitter modulation, and the basic idea is that 'robust region selection + error correction code + STC coding', and the embedding process is as follows:

(1) selecting carrier elements: and selecting the intermediate frequency quantized DCT coefficient in the block DCT transform as a carrier element.

(2) And (3) calculating distortion: and calculating distortion corresponding to carrier elements in the carrier image by using the existing JPEG image distortion algorithm.

(3) And (3) error correcting code encoding: the secret message is encoded using an error correction code.

(4) Embedding: and combining the carrier elements obtained by the STC, and modifying the carrier elements by the distorted and error-correction-coded secret message to realize embedding.

Compared with the traditional steganography algorithm, the scheme has robustness of resisting JPEG compression. However, because the quantized DCT coefficients of the intermediate frequency are selected as carriers during steganography, and these coefficients usually have a large quantization step size, their modification can cause a large modification to the image content, thereby greatly reducing the security, and being difficult to use in practice.

Disclosure of Invention

The invention provides a JPEG image processing method, a robust steganography system and a storage medium, which can solve the technical problems.

In order to achieve the purpose, the invention adopts the following technical scheme:

a JPEG image processing method comprises the following steps,

s11, sequentially carrying out quantization recovery and IDCT (inverse discrete cosine transform) conversion processing on the first quantized DCT coefficients corresponding to the JPEG image of the pre-carrier to obtain corresponding spatial domain images;

s12, carrying out overflow detection on each empty domain block in the empty domain image;

s13, performing overflow removal processing on the empty domain blocks with overflow, and not processing the empty domain blocks without overflow;

and S14, sequentially carrying out DCT transformation and quantization processing on the spatial domain image after overflow removal to obtain a second quantized DCT coefficient, coding the second quantized DCT coefficient and storing the second quantized DCT coefficient as a JPEG image, namely a carrier image.

Further, the S12 performs overflow detection on the space domain block, including:

and detecting whether the spatial block overflows or not, namely detecting whether the spatial value in the spatial block is larger than or equal to 127 or smaller than or equal to-128 or not, if so, the block overflows, and otherwise, no overflow exists.

Further, in S13, the specific step of performing the overflow removal processing on the space block is as follows:

all the null values are multiplied by a coefficient α, 0< α <1, so that all the null values are equal to or greater than-128 and equal to or less than 127.

if the airspace value in the airspace block is larger than 127, setting the value of the airspace block to be 127-beta;

if the airspace value in the airspace block is less than-128, setting the value of the airspace block to-128 + beta;

otherwise, the original numerical value is maintained unchanged;

wherein beta is a preset constant, beta is an integer, and beta is more than or equal to 5 and less than or equal to 10.

On the other hand, the invention also discloses a robust steganography method based on spatial domain de-overflow preprocessing, which comprises the following steps,

s100, carrying out overflow removal preprocessing on the pre-carrier JPEG image according to the steps S11-S14 to obtain a carrier image;

s200, selecting any second quantized DCT coefficient corresponding to the carrier image as a carrier element;

s300, modifying carrier elements by using a steganographic algorithm to realize the embedding of the secret message;

and S400, extracting the secret message by using an extraction algorithm corresponding to the steganography algorithm in the step S300.

Further, in step S300, embedding the secret message by using STC coding in combination with the carrier element, the distortion and the secret message modification carrier element after error correction coding in step S200, and coding the quantized DCT coefficient after embedding the secret message to obtain the secret-carrying image.

Further, the step S400 includes the following steps:

s401, decoding the compressed secret-carrying image at the extraction end to obtain a third quantized DCT coefficient;

s402, sequentially carrying out quantization recovery and quantization processing on the third quantized DCT coefficient to obtain a fourth quantized DCT coefficient, wherein the quantization recovery uses the same quantization table as that used in the compression of the secret-loaded image, and the quantization processing in the step S14 use the same quantization table;

and S403, selecting carrier elements located at the same positions in the embedding process and decoding the carrier elements by using STC and error correction codes in sequence according to the fourth quantized DCT coefficient to obtain the secret message.

In a third aspect, the present invention further discloses a robust steganography system, comprising:

the preprocessing unit is used for carrying out overflow removal preprocessing on the pre-carrier JPEG image according to the steps S11-S14 to obtain a carrier image;

the information embedding unit is used for selecting any second quantized DCT coefficient corresponding to the carrier image as a carrier element and modifying the carrier element by using a steganographic algorithm to realize the embedding of the secret information to obtain a secret image;

and the message extraction unit is used for extracting the secret message from the compressed secret-carrying image.

Finally, the invention also discloses a computer-readable storage medium, in which a computer program is stored, which, when being executed by a processor, causes the processor to carry out the steps of the method as described above.

According to the technical scheme, the robust steganography method based on spatial domain de-overflow preprocessing comprises the steps of firstly performing de-overflow preprocessing on a pre-carrier to obtain a carrier image, wherein the carrier image subjected to the de-overflow preprocessing has strong robustness, so that the carrier element selection process in the robust steganography algorithm based on generalized jitter modulation introduced before is not needed, and all second quantized DCT coefficients corresponding to the carrier image can be used as steganography carriers. This has two benefits: firstly, carrier elements are more selected, and higher embedding rate can be realized; secondly, a coefficient with a small quantization step size can be selected as a steganographic carrier during actual processing, so that the image content is prevented from being greatly modified during steganographic modification, and the safety of steganographic is greatly improved. Therefore, the robust steganography method of the overflow removal preprocessing can resist JPEG compression, meanwhile, an existing steganography analyzer is difficult to distinguish whether a carrier image is modified or not, safety is remarkably improved, and the possibility of detecting steganography behavior can be effectively reduced.

Drawings

FIG. 1 is a flow diagram of the present invention for a desaturation process;

fig. 2 is a flow chart of the method of the present invention.

Detailed Description

In order to make the objects, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to fig. 1 and 2, and it is obvious that the described embodiments are a part of the embodiments of the present invention, but not all of the embodiments.

As shown in fig. 1, the embodiment of the present invention first discloses a JPEG image processing method, including the following steps,

By carrying out the overflow removal processing on the empty domain block with overflow, the defects caused by adopting an empty domain truncation method in the prior art can be avoided, and the large change of the empty domain value is avoided.

It should be noted that, in the JPEG processing, the image is divided into 8 × 8 blocks, and the preprocessing operation is to process each block separately, so in step S11, the first quantized DCT coefficients of the blocks are processed sequentially to obtain spatial domain blocks, and all the spatial domain blocks form a spatial domain image together; in step S14, the overflow-removed spatial domain image is processed, and the overflow-removed spatial domain block is processed to obtain the second quantized DCT coefficient of the block. In order to avoid redundant description, the distinction between blocks and the whole in the above steps is not obvious.

Wherein the S12 carries out overflow detection on the space domain block, including:

There are many ways to perform the de-overflow processing on the empty domain block, and two preferred embodiments are provided in the present invention for reference.

The first embodiment is as follows: the specific steps of performing the overflow removal processing on the space domain block in the step S13 are as follows: all the null values are multiplied by a coefficient α, 0< α <1, so that all the null values are equal to or greater than-128 and equal to or less than 127. And reducing the overflow airspace value to be within a normal value range by geometric reduction.

Example two: in S13, the specific steps of performing the overflow removal processing on the space block are as follows: if the airspace value in the airspace block is larger than 127, setting the value of the airspace block to be 127-beta; if the airspace value in the airspace block is less than-128, setting the value of the airspace block to-128 + beta; otherwise, the original numerical value is maintained unchanged; in the formula, β is a predetermined constant, β is an integer, and preferably 5. ltoreq. β.ltoreq.10. Partial truncation is used to avoid overflow of spatial domain values.

Secondly, as shown in fig. 2, the invention also discloses a robust steganography method based on spatial domain de-overflow preprocessing, which comprises the following steps,

In step S300, embedding the secret message is implemented by using STC coding in combination with the carrier element, distortion and the secret message modification carrier element after error correction coding in step S200, and the quantized DCT coefficient after embedding the secret message is coded to obtain a secret-loaded image.

The step S400 includes the following steps: s401, decoding the compressed secret-carrying image at the extraction end to obtain a third quantized DCT coefficient; s402, sequentially carrying out quantization recovery and quantization processing on the third quantized DCT coefficient to obtain a fourth quantized DCT coefficient, wherein the quantization recovery uses the same quantization table as that used in the compression of the secret-loaded image, and the quantization processing in the step S14 use the same quantization table; this operation is to make the quantization step size in embedding and extraction the same. And S403, selecting carrier elements located at the same positions in the embedding process and decoding the carrier elements by using STC and error correction codes in sequence according to the fourth quantized DCT coefficient to obtain the secret message.

It is specifically stated here that, in the prior art, only intermediate-frequency quantized DCT coefficients can be selected as carrier elements, but after the de-overflow preprocessing, the present invention can select any second quantized DCT coefficients as carrier elements, so that carrier elements with small quantization step size can be selected, and a higher embedding rate can be achieved.

The following detailed description of "deglue preprocessing" is provided:

1. obtaining a corresponding spatial domain block by quantizing the DCT coefficients: firstly, the block quantization DCT coefficient obtained by decoding each image is multiplied by a quantization table and IDCT (DCT inverse) conversion is carried out to obtain a corresponding spatial domain block.

2. And (3) overflow detection: since 128 is subtracted from all pixel values during the JPEG process, the spatial domain values obtained after IDCT have a-128 difference with respect to the pixel values. And detecting whether the block overflows or not, namely detecting whether the spatial domain value in the block is larger than 127 or smaller than-128 or not, if so, the block overflows, and otherwise, the block does not overflow.

3. And (3) spatial domain overflow removal: no processing is required for blocks without overflow. For an overflowing block, two methods for removing the overflow are designed.

Overall scaling: all the null values are multiplied by a coefficient α, 0< α <1, so that all the null values are equal to or greater than-128 and equal to or less than 127.

Partial truncation: the part position in the empty domain (the position where the overflow is usually selected) is cut off, that is, if the original value of the position is greater than 127, the value is 127- β, and if the original value of the position is less than-128, the value is-128 + β, where β > is 0, which can be selected according to actual needs, and most preferably, 5 ≦ β ≦ 10.

4. Calculating corresponding block quantization DCT coefficients: the preprocessed space domain blocks do not overflow, and the corresponding block quantized DCT coefficients after overflow removal can be obtained only by performing DCT (discrete cosine transform) transformation and then dividing the transformed space domain blocks by the corresponding quantization tables.

All the blocks are processed according to the method and then stored to obtain the carrier image in the embodiment of the invention.

In summary, the carrier image is obtained through the overflow removal preprocessing, and all positions can become carrier elements when steganography is performed on the image, so that the capacity and the safety of steganography can be obviously improved, and the method is convenient to use in practice.

On the other hand, the invention also discloses a robust steganography system which comprises a preprocessing unit and a processing unit, wherein the preprocessing unit is used for carrying out overflow removal preprocessing on a pre-carrier JPEG image according to the steps S11-S14 to obtain a carrier image; the information embedding unit is used for selecting any second quantized DCT coefficient corresponding to the carrier image as a carrier element and modifying the carrier element by using a steganographic algorithm to realize the embedding of the secret information to obtain a secret image; and the message extraction unit is used for extracting the secret message from the compressed secret-carrying image.

Meanwhile, the invention also discloses a computer readable storage medium which stores a computer program, and the computer program is used for causing a processor to execute the steps of the method when being executed by the processor.

It is understood that the system provided by the embodiment of the present invention corresponds to the method provided by the embodiment of the present invention, and the explanation, the example and the beneficial effects of the related contents can refer to the corresponding parts in the method.

As will be appreciated by one skilled in the art, embodiments of the present application may be provided as a method, system, or computer program product. Accordingly, the present application may take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, the present application may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, and the like) having computer-usable program code embodied therein.

The present application is described with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the application. It will be understood that each flow and/or block of the flow diagrams and/or block diagrams, and combinations of flows and/or blocks in the flow diagrams and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general purpose computer, special purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.

These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instruction means which implement the function specified in the flowchart flow or flows and/or block diagram block or blocks.

These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.

The above examples are only intended to illustrate the technical solution of the present invention, but not to limit it; although the present invention has been described in detail with reference to the foregoing embodiments, it will be understood by those of ordinary skill in the art that: the technical solutions described in the foregoing embodiments may still be modified, or some technical features may be equivalently replaced; and such modifications or substitutions do not depart from the spirit and scope of the corresponding technical solutions of the embodiments of the present invention.

Claims

1. A JPEG image processing method is characterized in that: comprises the following steps of (a) carrying out,

2. The JPEG image processing method according to claim 1, characterized in that: the S12 carries out overflow detection on the space domain block, including:

3. The JPEG image processing method according to claim 2, characterized in that: the specific steps of performing the overflow removal processing on the space domain block in the step S13 are as follows:

4. The JPEG image processing method according to claim 2, characterized in that: in S13, the specific steps of performing the overflow removal processing on the space block are as follows:

otherwise, the original numerical value is maintained unchanged;

wherein beta is a preset constant, beta is an integer and beta is more than or equal to 5 and less than or equal to 10.

5. A robust steganography method based on spatial domain de-overflow preprocessing is characterized in that: comprises the following steps of (a) carrying out,

6. The robust steganography method based on spatial domain de-overflow preprocessing as claimed in claim 5 wherein: in the step S300, embedding the secret message is implemented by using STC coding in combination with the carrier element, distortion and the secret message modification carrier element after error correction coding in the step S200, and the quantized DCT coefficient after embedding the secret message is coded to obtain the secret-carrying image.

7. The robust steganography method based on spatial domain de-overflow preprocessing as claimed in claim 6 wherein: the step S400 includes the following steps:

8. A robust steganographic system, characterized by: the method comprises the following steps:

9. A computer-readable storage medium, storing a computer program which, when executed by a processor, causes the processor to carry out the steps of the method according to any one of claims 1 to 7.