RU2467486C1 - Method of embedding compressed message into digital image - Google Patents

Abstract

FIELD: physics, communications.

SUBSTANCE: invention relates to steganography and specifically to methods of embedding a compressed message into a digital image and can be used to realise concealed storage and transmit confidential information over open communication channels, as well as digital watermarks. The technical result is achieved due to that the method of embedding a message into a digital image includes a procedure for compressing the embedded message compared to the digital image itself. The embedded message is compressed by searching for matching of fragments of the message and part of the digital image which does not contain the least significant bits via a sliding window technique, replacing the matching fragments of the message with address indicators of their position in the digital image, forming an ordered list of addresses and subsequent transmission of an address table which does not carry semantic information and whose volume is considerably less than that of the initial message.

EFFECT: possibility of storing and transmitting large volumes of confidential information with minimum adjustment of statistical characteristics.

5 dwg

Description

The invention relates to the field of steganography, and in particular to methods of embedding messages in digital images, and can be used to organize hidden storage and transmission of confidential information through open communication channels, as well as digital watermarks.

Known methods for steganographic protection of information based on the introduction of additional information in raster images by replacing the least significant bits in bytes of the color representation of the original image with bits of added information (See Bender W., Gruhl D., Morimoto N., Lu A., Techniques for data hiding, IBM system journal, VOL 35 (3 & 4): pp. 313-336, 1996. RF patent No. 2002135272, published July 10, 2004).

Another option for organizing a stegochannel is a method of embedding additional information in digital images, in which to counteract statistical analysis methods, only a part of the least significant bits in bytes of the color representation of the original image are used, and the remaining bits are used for subsequent correction of the most important statistical parameters (See Provos N Defending Against Statistical Steganalysis, Proceeding of the 10 USENIX Security Symposium, 2001, pp. 323-335).

To organize the hidden storage and transmission of confidential information through open channels, a method was proposed for introducing additional information into digital images (See RF Patent No. 2288544, published November 27, 2006, Bull. No. 33), namely, that the original digital image decomposed into bit layers, to record additional information, select one of the obtained bit layers, which are presented in the form of a bit sequence, additional information is recorded using a code, while in the received bit after ovatelnosti bits located at the boundaries of transitions of identical sequences of zeros and ones are replaced in accordance with the recorded additional information bits, the remaining bits of the lower bit planes are used as necessary for correcting the original image or to record other additional information.

The disadvantage of these methods is the need for correction of statistical characteristics, which leads to an increase in image processing time, as well as to a decrease in the amount of embedded information. Moreover, an increase in the amount of embedded information over a certain threshold value leads to a critical change in the original image. Changes introduced by the introduction of a significant amount of additional information lead to distortions of the digital image and can be detected using statistical analysis methods or even be noticeable during visual analysis of the resulting image.

The closest in technical essence to the claimed invention (prototype) is a method of embedding a message in a digital image (See RF Patent No. 2407216, publ. 12/20/2010), which consists in replacing the least significant bit in bytes of the original digital image, while the least significant the bit in bytes of the original digital image is assigned a flag value of “one” when the part of the bits of the byte of the digital image signal and the bits of the message signal coincide, or the flag value is “zero” when the mismatch The statistics of the distribution of the least significant bits are produced for the remaining part that is not used as flag values.

The disadvantage of this method is the limited amount of embedded information, as well as the ability to access the semantics of the embedded message with the known embedding method.

The objective of the invention is to provide a method of embedding a compressed message in a digital image, which allows to store and transmit large amounts of confidential information and complicate the procedure for third parties to access the semantics of the embedded message.

This problem is solved by the fact that in the method of embedding a compressed message in a digital image, which consists in setting the flag value of the least significant bit to “one” when part of the bits of the byte of the digital image signal and the bits of the message signal coincide, or the flag value is “zero” - if there is a mismatch , in addition, before embedding the message, it is compressed relative to the digital image itself by searching for matches of message fragments and the part of the digital image that does not contain the least significant bi s, the method of "sliding window", substitution of matched fragments posts pointers to the addresses of their position in the digital image, forming an ordered address list and subsequent embedding a table of addresses which does not bear the semantic information and the volume of which is significantly less than the volume of the original message.

Thanks to the new combination of essential features in the method, it is possible to embed large volumes of confidential information, and it is also difficult to gain access to the semantics of the embedded message by converting the embedded message into a match address table.

The analysis of the prior art made it possible to establish that analogues that are characterized by a combination 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 fame of the distinctive essential features that determine the same technical result, which is achieved in the present method. Therefore, the claimed invention meets the condition of patentability "inventive step".

The claimed method is illustrated by drawings, which show:

figure 1 - memory structure of the compressor;

figure 2 is a block diagram of the compressor in the compression process;

figure 3 - link structure in the memory of the dictionary containing a fragment of an embedded message;

4 is a link structure after reordering;

5 is a block diagram of the operation of the decompressor in the process of extracting messages.

Figure 1 shows the structure of the compression memory 101 used by the compressor, which includes an embedded message 103, an annular buffer 105 and a dictionary 107. 103 buffers the embedded message. The ring buffer is used to store data of the part of the digital image corresponding to the size of the embedded message and to find matches of the sequences of bits stored in the embedded message 103. Dictionary 107 is used as a “history window” of compression, which stores data that can be referenced by the decompressor to restore the original messages. Dictionary 107 consists of parts 109, 111, 113, 115. The first part 109 of the dictionary contains a match bit number that determines the position of the compressible fragment of the embedded message 103. The size of part 109 depends on the size of the embedded message in bits. Part 109 is used only in the algorithm for reordering links in the dictionary memory (Fig.2, 213) and is not used when retrieving the message. The second part 111 of the dictionary contains a flag bit, the value of which determines the values of the third and fourth parts of the dictionary. When the flag bit value is zero, the third part 113 of the dictionary contains the number of the bit that determines the position of the found match of the compressible fragment of the embedded message in the ring buffer 105, since the size of the digital image is 3 megabytes on average, the size of the third part is determined to be 25 bits. The fourth part 115 of the dictionary in this case contains the bit size of the found match in the embedded message part and the ring buffer part. It was experimentally established that the size in bits of coincidence of the part of the embedded message and part of the ring buffer does not exceed 256. Based on this, the size of the fourth part 115 of the dictionary is determined to be 8 bits. Thus, the size of the parts 111, 113 and 115 of the dictionary that are used when embedding and extracting the message is 34 bits (1 + 25 + 8).

The compressor determines the matches in the embedded message and in the digital image that are currently stored in the ring buffer. The search for matches returns the position and size of all found matches at each shift step in the ring buffer and writes their values in the form of links (addresses and sizes of matches) to the dictionary.

Figure 2 shows a block diagram of the operation of the compressor used to implement the present invention.

At the first step 201, an embedded message to be compressed is written to the memory.

At step 203, the compressor loads into the ring buffer a portion of the digital image corresponding to the size of the embedded message.

At step 205, a search is made in the ring buffer for matches of bit sequences of the embedded message and part of the digital image.

At step 207, a link of the following form is generated for each match: the number of the match bit of the embedded message, the flag bit, the number of the match bit in the digital image, the size of the match (Figs. 1, 109, 111, 113, 115). This output signal is used to form part of the compressed data message, which is then embedded in a digital image.

At step 209, the bit sequence of the digital image portion in the circular buffer is shifted.

At step 211, the number of shifts of the bit sequence in the ring buffer is counted. If the number of shifts is not larger than the size in bits of the digital image, then the data processing steps are repeated from step 203. Otherwise, the search in the ring buffer for matching bits of the embedded message is stopped and the dictionary formation is completed.

At step 213, the links are reordered in the dictionary memory according to the following algorithm:

Step 1. Links are written in ascending order of the values of the bits of matches of the embedded message.

Step 2. A pairwise comparison of links is performed as follows.

a) With the same value of the numbers of bits of matches of the embedded message and the size of the matches of two links, one of the links is arbitrarily deleted.

b) With the same value of the numbers of coincidence bits of the embedded message and different values of the sizes of matches, two links delete the link with a smaller value of the size of the match.

c) If the value of the match bit number of the embedded message of one link is greater than the value of the match bit number of the embedded message of the second link and the amount (match bit number of the embedded message + match size) of the second link is greater than the same value of the first, then the link with a lower value of the specified amount is deleted.

d) If the value of the match bit number of the embedded message of one link is greater than the same value of the second link and the value of the sum (the bit number of the embedded message + match size) of the second link is less than the same value of the first, then the value of the match bit number of the embedded message and the match bit number in the digital image of the first link is assigned a new value equal to the sum (match bit number of the embedded message of the second link + match size of the second link), and the value of pa measure matches the first reference is assigned a new value of the expression (old value matches the size of the first reference minus (-) the new value of the bit number matches the embedded messages minus the old value of the bit number matches the embedded message).

e) If the value of the match bit number of the embeddable message of one link is greater than the same value of the second link and the value of the sum (the number of the match bit of the embeddable message + match size) of the second link is less than the value of the match bit number of the embeddable message of the first link, then the first and second links a new link, in the first part of which (FIGS. 3, 301) the number of the first mismatched bit of the embedded message is recorded, the second part of the link — the flag bit 303 is set to “one”, when The third part of the link 305 with a size of 5 bits determines the size of the mismatched fragment of the embedded message with a digital image, and the fourth part of the link 307, the length of which is 28 bits, is filled with mismatched fragments of the embedded message. Moreover, if the size of the mismatched fragment of the embedded message is more than 28 bits, several links with the flag bit set to one are used. If the size of the mismatched fragment of the embedded message is less than 28 bits, the rest is filled with zeros. Thus, the total size of the parts 303, 305, and 307 of the link in this case is also 34 bits (1 + 5 + 28).

Step 3. The operation of the algorithm ends when the sum (the bit number of the embedded message matches + the size of the match) reaches the value equal to the number of bits of the embedded message. Thus, match table links are sorted by embed message. In this case, the first parts of all links are deleted.

At step 215, embed dictionary data in the form of links (Fig. 4) in the transmitted digital image. In this case, embedding consists in setting the flag value of the least significant bit to “one” when the part of the bits of the byte of the digital image signal and the bits of the message signal coincide, or the flag value to “zero” - if there is a mismatch.

Figure 5 shows a block diagram of the operation of the decompressor in the process of decompression.

At the first step 501, the digital image is received and the built-in match table is extracted from it according to the method described in RF patent No. 2407216 publ. December 20, 2010

At step 503, the decompressor loads dictionary links into the memory buffer (match table).

At step 505, match table link processing is performed. With the flag bit of the link set to “zero”, a fragment is searched in the memory buffer storing the original digital image using the dictionary link containing the position / length parameters. With the flag bit of the link set to "one", the message is read from the link itself.

At step 507, the read sequence is written to the memory either from a digital image or directly from a link. This output signal is used to form part of the data message after decompression.

At step 509, the processing of all matches in the match table is checked. If all links are processed, work stops. Otherwise, the data processing steps are repeated from step 505 until all dictionary references have been processed.

At step 511, a data message is generated after decompression.

It should be noted that embodiments of the present invention may be controlled by hardware, software, or any combination of hardware and software.

It should also be noted that along with the above description illustrating embodiments of the present invention, there are other options and modifications of the disclosed solutions that can be made within the scope of the invention defined in the claims.

To assess the effectiveness of the proposed method of embedding a compressed message in a digital image, tests were carried out, the essence of which was as follows. First, a sample was formed, consisting of 100 photorealistic digital images, the size of which ranged from one to three megabytes. Then, a 300 kilobyte message was embedded in digital images using the method described above. The relative volume of embedded data ranged from 10 to 30% of the volume of the original digital image, while maintaining the visual quality of the latter.

Claims (1)

A method of embedding a compressed message in a digital image, which consists in setting the flag value of the least significant bit to “one” when the bit part of the digital image signal byte and the message signal bits coincide, or the flag value to “zero” - if there is a mismatch, characterized in that before embedding messages compress it relative to the digital image itself by searching for matches of message fragments and the part of the digital image that does not contain the least significant bits using the “rolling” method windows ”, replacing the matching message fragments with pointers to the addresses of their position in the digital image, forming an ordered list of addresses and then embedding the address table, which does not carry semantic information and the volume of which is much less than the volume of the original message.

Images