RU2602670C1 - Method of embedding information in graphic file compressed by fractal coding - Google Patents

Abstract

FIELD: data processing.

SUBSTANCE: invention relates to steganography and specifically to methods of embedding a message into a digital image. Proposed is a method of embedding information into a graphic file. Image file is compressed by fractal coding. Method includes the stages of formation of the vector of image compression parameters, input of hidden information, separation of domains and rank areas, correlation of rank areas and domains, forming of final archive. Herewith during image compression the hidden information bits are embedded in the initial image file by conversion of real domain coordinates into imaginary ones implemented due to introduction of correcting coefficients, values of which are calculated depending upon the position of input of additional information in the code structure created on the basis of virtual images.

EFFECT: technical result is the possibility of hidden transfer of confidential data using the container represented in the form of a fractal compressed image.

1 cl, 3 dwg

Description

The invention relates to the field of steganography, and in particular to methods of embedding information in a graphic file, and can be used to organize hidden storage and data transfer through open communication channels.

The well-known "Method of embedding messages in a digital image" (patent RU 2407216 C1, published: December 20, 2010), in which embedding is carried out by replacing the least significant bit in bytes of the original digital image, while the least significant bit in bytes of the original digital image assign a flag value of "one" when the coincidence of a part of the bits of the byte of the digital image signal and the bits of the message signal, or the flag value of "zero" when the mismatch.

The disadvantage of this method is the impossibility of its use for a graphic file compressed by the fractal method.

The well-known "Method for transmitting additional information in fractal coding of an image" (patent RU2292662C2, IPC H04N 7/08, published: 01/27/2011), in which embedding is performed in the lower digits of domain indices.

The disadvantage of this method is the distortion of the original image as a result of the reorientation of domains associated with irreversible transformations of the coordinates of the domains.

The well-known "Method for transmitting additional information when sharing vector quantization and fractal coding of images, taking into account the classification of domains and blocks from the codebook" (patent RU 2327301 C2, IPC H04N 7/08, G06T 9/00, published: 27.12.2007), in which in the index vector of a domain or block from the codebook, consisting of n bits, m bits of additional information are entered, instead of the least significant bits of this vector.

The disadvantage of this method is the distortion of the original image due to the reorientation of the domains and their displacement relative to the initial position caused by the transformation of the least significant bits of the domain coordinates.

The closest in technical essence and functions performed is the "Method of embedding information in an image compressed by the fractal method, based on the generated domain library" (patent RU 2530339 C1, IPC H04N 19/46, H04N 19/90, H04L 9/00, published: 10.10.2014), in which the domain library is divided into several equal parts for encoding embedded information.

The disadvantage of this method is the low efficiency of embedding information, expressed in a small ratio of the volume of embedded information to the size of the source file.

The objective of the invention is to provide a method of embedding information in a graphic file compressed by the fractal method, which enables the hidden transfer of a larger amount of data compared to the prototype using a container presented in the form of a fractally compressed image.

This problem is solved in that in the method of embedding information in a graphic file compressed by the fractal method, which includes the steps of forming a vector of image compression parameters, entering hidden information, highlighting domains and ranking areas, correlating ranking areas and domains, forming a final archive, in the process of image compression bits of hidden information are embedded in the original graphic file by converting the real coordinates of the domain into imaginary, implemented by introducing correction factors, Niya which is calculated depending on the position of the block input additional information in the code construction that is based on virtual images.

The analysis of the prior art made it possible to establish that analogues that are characterized by a set of features identical to all the features of the claimed technical solution are absent, which indicates the compliance of the claimed method with the condition of patentability “novelty”.

Search results for known solutions in this and related fields of technology in order to identify features that match the distinctive features of the claimed object from the prototype showed that they do not follow explicitly from the prior art. The prior art also did not reveal the popularity of the impact provided for by the essential features of the claimed invention, the transformations to achieve the specified technical result. Therefore, the claimed method meets the condition of patentability "inventive step".

"Industrial applicability" of the invention is due to the fact that a device that implements the proposed method can be implemented using a modern elemental base, with achievement of the destination specified in the invention.

The claimed method of embedding information in a graphic file compressed by the fractal method is illustrated by drawings, which show:

FIG. 1 - a method of forming imaginary domains responsible for the value of embedded information, where i, j are the block indices in the code structure;

FIG. 2 - an algorithm that summarizes the proposed method of embedding information;

FIG. 3 is a functional diagram of the proposed method of embedding information.

In accordance with the functional diagram (Fig. 3), the process of generating a fractally compressed graphic file with parallel embedding of information is based on five blocks. The first block implements the input of compression parameters along with embedded information. The second and third blocks are designed to search in the structure of the source graphic file of domains and ranking areas, respectively. The steps 1 through 3 are described in (Y. Fisher, Fractal Image Compression - Theory and Application: New York: Springer, 1994). The fourth block, which is presented in the source (A.E. Jacquin, “Fractal image coding based on a theory of iterated contractive image transformations” in Proceedings of SPIE Visual Communications and Image Processing ′90, performs the stage of correlating domains and ranking areas with information embedding) Vol. 1360, 1990). The final stage of forming the archive file is carried out by the fifth block. The functional features of the described blocks are indicated in (Y. Fisher, Fractal Image Compression - Theory and Application: New York: Springer, 1994).

The implementation of the proposed method consists in converting the compression process of a graphic file, expressed in parallel inclusion of the procedure for embedding information in the stage of correlation of domains and ranking areas of the image.

The first block, in accordance with the general fractal compression algorithm, is designed to enter compression parameters, including the minimum domain size, minimum domain pitch, root-mean-square deviation (RMS) threshold value, quadro tree depth and embedded information, which are discussed in (T. Laurencot, A.E. Jacquin, Hybrid image block coders incorporating fractal coding and vector quantization, with a robust classification scheme, AT&T Tech. Memo., February 1992).

The entered data goes to the input of the second block, the second input of which is loaded with the original image intended for compression. On the basis of the second block, the image is divided into domains, the minimum size of which is specified in the compression parameters. Groups of larger domains are also distinguished, and for a more efficient calculation, each subsequent domain group is twice as large as the previous ones. The maximum possible domain size is selected in accordance with the size of the original image. After the domains are allocated, they are classified based on the values of the mathematical expectation of the pixel intensity of the “rings” of the segments, which is presented in (Patent RU 2541203 C2, IPC H04N 19/90, G06T 9/00, published: 02/10/2015).

Next, the downloaded image enters the third block, which selects the ranking areas in the graphic file. Ranked areas completely fill the entire image; unlike domains, they do not overlap with each other and two times less than the width of domains, as indicated in (A.E. Jacquin, “Fractal image coding based on a theory of iterated contractive image transformations” in Proceedings of SPIE Visual Communications and Image Processing ′90, vol. 1360, 1990).

After the formation of the domain library and the determination of rank areas, the data from the third block is input to the fourth block, which carries out the main stage of correlation of rank areas and domains with the parallel embedding of information. This procedure is implemented by selecting the appropriate domain rank domain under consideration. Classification can reduce the complexity of the task by several times, since the correlation is carried out for segments belonging to the same class.

The similarity of domains and ranks is determined by the least squares method (OLS). The domain under study is subjected to affinity transformations to ensure maximum similarity with rank, after which, according to the value obtained by the least-squares method, a decision is made to continue the search for correspondence. Affine transformations mean operations on image segments such as mirror image, rotation at angles of 90, 180, 270 degrees and scaling (Kronover PM Fractals and chaos in dynamic systems. Fundamentals of the theory. M., 2000. - 352 p.) . In the compression parameters, the minimum SDE threshold obtained in the calculation of the value using OLS is indicated in advance. In the case of determining the threshold value, it is necessary to find a compromise between the duration of the compression of the graphic file and the quality of its compression, since these parameters have a reciprocal relationship. If a suitable domain is not found for the rank domain, the quad tree method is used, in which the rank domain is divided into four equal parts and they perform similar actions to find the corresponding domains, as shown in (Y. Fisher, Fractal Image Compression - Theory and Application: New York: Springer, 1994).

After the stage of correlation of rank areas and domains, the data from the fourth block is sent to the fifth block. This block forms an archive consisting of a header, which indicates the parameters of the generated compressed file, and a data field containing the coordinates of the corresponding domains, affine transformation coefficients, and brightness and contrast values.

, иначе по формуле

, где n≥2. Разбитую по блокам встраиваемую информацию на каждой итерации при поиске соответствия ранг - домен соотносят с составленной кодовой конструкцией, после чего рассчитывают значения мнимых координат домена, с учетом расположения этого блока в кодовой конструкции. Расчет значений мнимых координат происходит по формулам

и

, где fx, fy - корректирующие коэффициенты, соответствующие порядковому номеру расположения мнимого изображения в кодовой конструкции по осям x и y соответственно; h, v - ширина и высота изображения, x, y - координаты реального домена, соответствующего конкретной ранговой области (фиг. 2).The developed method proposes the use of imaginary domains, which are formed on the basis of the original image. In the process of correlating domains and ranking areas, the corresponding ranking area is assigned the coordinates from the imaginary area, depending on the position in the code structure of the information block that needs to be embedded in the image file. The code structure is formed on the basis of the original image by duplicating it along the x and y axes, where each imaginary image (marked with a dashed line), and accordingly the domain of the imaginary image (shaded areas), is responsible for various values of the embedded information presented in binary form (Fig. . one). The size of the code structure depends on the dimension of blocks n specified when the compression parameters are entered, into which the embedded information is divided. If the block length is an even number, then the dimension of the code construction s is calculated by the formula

otherwise by the formula

where n≥2. The embedded information broken down into blocks at each iteration when searching for a rank-domain correspondence is correlated with the compiled code structure, after which the imaginary domain coordinates are calculated taking into account the location of this block in the code structure. The imaginary coordinates are calculated using the formulas

and

where fx, fy are correction factors corresponding to the ordinal number of the location of the imaginary image in the code structure along the x and y axes, respectively; h, v is the width and height of the image, x, y are the coordinates of the real domain corresponding to a specific rank area (Fig. 2).

The maximum size of the code construction depends on the bit dimension of the field allocated for the domain coordinate parameter in the compressed file. In the used algorithm for compressing graphic files, the size of the coordinate area, floating with an extreme value of 12 bits (from 0 to 4095 pixels). Depending on the source image, it is possible to create a code structure with a dimension of about 10 by 10, that is, about 6 bits of embedded information per rank area. To increase the efficiency of the steganographic algorithm, it is possible to expand the coordinate range, depending on the goals of the developer.

The inverse operation of extracting embedded information is based on obtaining the quotient of dividing the specified coordinates by the width and height of the graphic file, respectively. The obtained integer values indicate a block of embedded information in the code structure and the corresponding embedded data. The decompression process itself is based on recalculating the imaginary coordinates of the domains in a real way by taking the modules from the imaginary coordinates in width and height, respectively. Further decompression is implemented according to the standard algorithm.

The effectiveness of the proposed method compared to the prototype is to increase the volume of embedding additional information in the absence of structural changes in the generated file.

The implementation of the proposed method of embedding information has confirmed the effectiveness of its application. The most popular software products designed for steganographic analysis do not reveal the fact of embedding information when using this method, since the statistics of the distribution of bits does not change. The potential amount of information transmitted as a percentage is, depending on the source file, 6% or more of the total volume of the container file, unlike the prototype, where the same value reached 1%. The resulting percentage is calculated based on the ratio of the product of the number of ranking areas in the image with the number of bits described by one imaginary domain to the total file size. During the experiment, an image of 617 kB in size was used, in which about 50,000 ranking regions were formed, with 6 bits per domain. Accordingly, the amount of embedded information is 300,000 bits (37500 bytes), which is 37500/617000 = 0.06 (6%) of the total size of the source file.

Claims (1)

A method of embedding information in a graphic file compressed by the fractal method, which includes the steps of generating a vector of image compression parameters, entering hidden information, highlighting domains and ranking areas, correlating ranking areas and domains, forming a final archive, characterized in that the hidden information bits are compressed during image compression embed in the source image file by converting the real coordinates of the domain into imaginary, implemented by introducing correction factors, the values of which s is calculated in dependence on the position of the block input additional information in the code construction that is based on virtual images.

Images