RU2434356C2 - Creating watermarks in coded information content - Google Patents

Abstract

FIELD: information technology.

SUBSTANCE: coded information content is received, at least one pre-processed set of watermarks is received and a certain number of bits is directly replaced with an alternative value, starting with a certain position in the coded information content. Said alternative value contains embedded watermark signals, wherein said alternative value is selected from a plurality of alternative values determined by at least one of said set of watermarks. The invention also describes a device and a method of creating a set of watermarks, which involves selecting positions, where the value in the coded information content must be replaced by an alternative value, and calculating a plurality of alternative values for each selected position.

EFFECT: creating watermarks in coded information content by embedding watermarks into selected positions in the coded information content without deterioration of quality.

18 cl, 4 dwg

Description

FIELD OF THE INVENTION

The present invention relates to the creation of digital watermarks, and more particularly to the implementation, deletion / replacement and detection of digital watermarks in encoded content (content).

BACKGROUND

Coded content requires a “decoding” process to use the content. Some examples of encoded content include the formats MPEG-1 (expert group on moving images), MPEG-2, H264 / AVC, WMA (Windows sound files), MPEG4, JPEG2000 (a joint group of experts on computer-aided processing of photographic images), MP3 (MPEG-1 Audio Layer 3), PDF (portable document format), Windows Word, Postscript (page description language), etc. and their encrypted versions.

One of the methods of the prior art for creating watermarks in encoded content is to embed watermark signals in the structure and syntax elements. US Patent No. 6,687,384 is an example of embedding data into syntax elements in an encoded bitstream such as MPEG-1 and MPEG-2. Such watermarks, however, cannot withstand a format change or digital-to-analog conversion.

Another way to create watermarks in a prior art encoded content is to embed a watermark by adding distortion to DCT (dynamic cosine transform) coefficients for MPEG encoded content. An example of this approach is described in F. Hartung and B. Girod, “Digital Watermarking of MPEG-2 Coded Video in the Bit Stream Domain,” Proc. IEEE ICASSP, p. 2621-4, April 1997. The Hartung and Girod method does not use perceptual techniques.

Another prior art method for creating watermarks in encoded content is to directly change the encoded content. US Pat. No. 6,373,960, hereinafter “Conover,” describes a method for creating watermarks within a compressed MPEG video bitstream by changing some DCT coefficients so that the length of the statistical (entropy) encoding of these coefficients remains unchanged after this change. No specific watermark embedding methods are defined in Conover. Conover watermark locations are limited to coefficients with 0 “series” of zeros. Conover coefficients are in the high frequency region. The Conover method does not include preprocessing phases, and there is no “alternative value” for actually inserting a watermark in the later phase described. The concept of generating and using “watermark kits” is not disclosed anywhere and is not taught at Conover.

Prior art methods for creating watermarks in encoded content do not have a preprocessing unit to create “alternative values” for some parts of the encoded content. Prior art methods for creating watermarks in encoded content exclude encoded content that is further masked, darkened, scrambled, or encrypted (collectively, “encrypted”).

SUMMARY OF THE INVENTION

In the present invention, watermarks in coded content are created by embedding watermarks in selected locations in the coded content. This is achieved by replacing the original values in the encoded content with their alternative values. Each source value may have one or more alternative values, and each alternative value contains a watermark signal. In addition, watermarks can be removed by replacing the alternative values with the original values or by replacing one alternative value with another alternative value.

By choosing one of the alternative values to replace the original value in the corresponding locations in the encoded content, the following goals are achieved:

- Alternative values have a size identical to the original value, and replacing any one of the alternative values creates encoded content that is compatible with a particular format and does not introduce perceptual distortions into the encoded content.

- Alternative values may improve the quality of the encoded content. For example, the original value may or may not have a valid value. In other words, without replacing one of the alternative values, the encoded content may not have a valid format. In another case, when the original value is a valid value, without replacing the alternative value, the original value can lead to a decrease in the quality of the encoded content.

- Each alternate value contains watermark signals. These watermark signals, either together with other watermark signals at other positions within the encoded content, or themselves can have one or more embedded information elements. An information element consists of one or more bits.

- The information volume of alternative values (which is determined using the amount of data in the alternative value and the number of alternative values) should be small compared with the data volume of the encoded content.

One important goal of directly creating watermarks in encrypted content is “limited encryption”. Limited encryption takes into account the correspondence between plaintext and ciphertext. For example, a plaintext block includes components c1, c2 ... cn. After limited encryption, the ciphertext consists of c1 ', c2', ..., cn ', where ci' is the encrypted version of ci (1≤i≤n). A simple example of limited encryption is splitting content into parts and encrypting each part separately. For example, in a pinch, each coefficient or group of coefficients is encrypted separately. One of the categories of encryption that makes limited encryption possible is selective encryption or partial encryption. Instead of processing the content (audio or video) as a stream of binary data (also called “natural encryption”), selective encryption methods “understand” the syntactic structure of the content (for example, the MPEG-2 structure) and selectively encrypt only some parts of the content. Selective encryption occurs after compression. Encrypted content should not be of commercial value, although some of the content may be visible. Some selective encryption schemes may preserve the format, bit rate, and size of the encoded content that is not encrypted using selective encryption. Other selective encryption schemes may increase the bit rate or require a special decoder.

Taking the MPEG-2 content as an example, a simple selective algorithm encrypts only I-frames. The selectively encrypted MPEG stream may be a valid MPEG stream. Although P- and B-frames in MPEG-2 are useless without knowing the corresponding I-frames, most of the MPEG video data is still visible due to interframe correlation and mainly because of unencrypted I-blocks in P- and B-frames. Other selective encryption schemes include encryption of MPEG-2 headers and / or encryption of DCT coefficients. DCT coefficients are divided into “DC coefficients” and “AC coefficients”. DC coefficients are coefficients with zero frequency in both dimensions, and AC coefficients are other coefficients with nonzero frequencies. All DC coefficients or part of the AC coefficient values of all I-blocks can be encrypted.

A system and method for embedding a watermark in a coded content are described, which includes receiving a coded content, receiving a pre-processed set of a watermark that is an output of an external pre-processing unit, receiving a sequence of bits, the sequence of bits containing useful watermark information, and performing a direct replacement of a certain number of bits of encoded content, starting from the position with al an alternative value, and the watermark signal embedded in it has an alternative value, and the alternative value is one of the many alternative values defined in the watermark set, and the only alternative value defined in the watermark set, and the position is selected in the watermark set so that when the alternative value is the only alternative value, then no replacement is performed, and when the alternative value is selected Xia of a plurality of alternative values specified in the kit watermark bit sequence determines which alternative is selected.

Additionally described are a system and method for embedding a watermark in a coded content, which includes receiving a coded content, receiving at least one set of a watermark, receiving a sequence of bits, this bit sequence containing useful watermark information, and performing one of the steps : perform a direct replacement of a certain number of bits, starting at a certain position, with an alternative value or leave some unchanged to lichestvo bits starting from a position based on the values of bits in the bit sequence useful watermark information and wherein the alternative value has embedded therein watermark signals. A system and method for detecting a watermark is also described, which includes receiving at least one set of watermark, receiving the watermarked content, extracting a plurality of coefficient values from the watermarked content, and extracting the watermark bit values from the values coefficients.

BRIEF DESCRIPTION OF THE DRAWINGS

The present invention can best be understood from the following detailed description when read along with the accompanying drawings. The drawings include the following figures, briefly described below.

1 is a block diagram that shows a pre-processing flow for embedding a watermark in accordance with the principles of the present invention.

Figure 2 shows a set of watermark (KVZ) of the present invention.

Figure 3 is a flowchart of a watermark detection method in accordance with the principles of the present invention.

4 is a block diagram of a reproducing apparatus that receives encoded content having watermark sets embedded therein.

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

The process of creating watermarks in coded content can be divided into three separate stages.

• The choice of location, which is the choice of the position where the value in the encoded content can be replaced with alternative values containing watermark signals.

• Calculation of alternative values, which is the definition of alternative values so that the alternative values have the same number of bits as the value that the alternative values should replace in the encoded content, and such a replacement will not cause a change in the perception of the content. In addition, these alternative values contain watermark signals.

• Embedding / inserting a watermark, which is the actual replacement of the value in the encoded content with one of the alternative values.

The first two stages can be pre-processed. As a result of pre-processing, a set of watermark sets (KVZ) is created. KVZ contain all the information for the actual implementation / insertion of a watermark. The watermark preprocessing unit uses the encoded content as input and the watermark key and outputs a sequence of watermark sets. In the case where only one alternative value is generated for each SC, the useful information of the watermark can become additional input information for the pre-processing unit. These watermark sets are integrated into the final content as metadata of the encoded content, as a separate channel multiplexed with the content, as steganographic data hidden in syntax elements or in the content, or as a separate file stored on physical media (optical disk, tape, hard drive, etc.) or transmitted over the network (via TCP / IP, satellite communications, etc.).

Useful watermark information is embedded using a selection of alternative values in the LSC. KVZ has at least one alternative value. When each KVZ has only one alternative value, there are two ways to embed / insert watermark information (useful information). The first method is the introduction of one fixed useful information of the watermark by replacing all the initial values in the water storage facility with their alternative values. The second method provides the ability to embed various useful watermark information by switching between replacing or not replacing for the KVZ. For example, a replacement indicates a positive bit value, and a lack of replacement indicates a negative bit value. To implement the useful information of the watermark “00101001”, the initial values in the first and second KVZ are not replaced, the initial value in the third KVZ is replaced with its alternative value. For a watermark with more than one alternative value, each alternative value contains a watermark signal, which may represent another element of useful watermark information. An alternative replacement value is selected based on the useful information of the watermark. For example, for the introduction of one bit in the KBZ, only two alternative values V1 and V2 are required. To embed the bit value “0”, V1 is selected to replace the original value in the encoded content, and to embed the bit value “1”, V2 is selected. Using the two values V1 and V2, a bit equal to “0” and “1” can be expressed. If there are four alternative values (V1, V2, V3 and V4), then two-bit information can be embedded (ie, “00”, “01”, “10” and “11”). By using more alternative values in one position, more bits can be embedded, which allows very efficient implementation of information.

Useful watermark information is usually taken in the third step, “watermark insertion”. It can be saved or calculated using a component external to the watermarking system. Useful watermark information is usually an identifier that uniquely identifies the recipient, the playback device (manufacturer, model number and / or serial number) or the date and time the content was played. To potentially reduce the size of the SCL in the SCL, the difference between the initial value of V and its alternative values can be maintained. KVZ can be additionally compressed.

It is important that the watermark is protected from unauthorized access or modification, because with this information the watermarking system is not only vulnerable to various attacks, but it is also easy for a hacker to insert fake watermarks, or modify or delete existing watermarks. If the KBZ is stored and transmitted as steganographic (watermarked) data, then access can be controlled using the watermarking key. If the KVZ are stored and transmitted through some other channel, then the encryption of the KVZ is required.

As indicated earlier, the first and second steps can be performed as pre-processing. Thus, prior to the third step, the encoded content is not watermarked. At the third stage, the implementation of the watermark is performed by replacing some of the values in the encoded content with the alternative values defined in the KBZ. These values may be encoded syntax elements, such as packet identifier, headers, quantization table, Huffman table for statistical encoding, encoded coefficients, or encoded motion vectors. KVZ determine where the watermark signals are embedded (in alternative values) and what possible watermark signals can be in these positions by selecting the appropriate alternative values.

Figure 1 is a structural diagram that shows the sequence of operations of the preliminary processing of the encoded content to generate the CVC. Figure 2 shows a set of watermark in accordance with the principles of the present invention. Each CLC is described using a vector (P, L, C, {V}, V1, V2 ... Vn), where P is the position of the original value of V, which can be replaced in the future by an alternative value in the encoded content, L is the number of bits occupied by the initial value of V in the encoded payload, starting from P, C is a set of global positions of the coefficients (for example, quantized using DCT or wavelet transform of the coefficients), which are encoded in L bits (in the encoded payload) using the statistical method coding. Statistical coding, such as Huffman coding, is usually applied to the last coding stage to create coded content. V1, V2 ... Vn are valid alternative V values, and each of these values contains a watermark signal. Each position of the coefficients is represented by (ch, f, b, co), where ch is the channel index, f is the frame index, b is the block index and co is the coefficient index within the block of the video element. An alternative value is acceptable if, in the case when the current value is replaced with this value, the formats are supported and there are no effects on the perception of the content. In addition, V1, V2 ... Vn occupy L bits in the encoded payload, just as V. {V} indicates that the original value is optional in the embed / insert process. An initial value may be required in watermark removal processes.

The initial value of V may contain one or more coefficients in coded form. When the CEC is generated, if the input is coded content such as MPEG-2 or MPEG-4, statistical coding, i.e. variable length coding (COP), first must be “deleted” to access the coefficients and then the corresponding position and alternative values are found and the original coefficient and these alternative values are stored. An alternative value of Vi may correspond to the same coefficients as V corresponds to, but in some cases Vi may correspond to more or less coefficients than that corresponding to V.

A typical block of 8x8 quantized DCT coefficients in MPEG-2 is shown below. Most higher order coefficients are quantized to 0.

After a zigzag scan, the sequence of DCT coefficients that will be transmitted is as follows:

The first DCT coefficient (12) is sent through a separate Huffman table. After the “series-level” analysis, the remaining coefficients and the corresponding “series” of zeros:

Using the zero table of DCT coefficients, given below and defined in the MPEG-2 standard, these coefficients are encoded into 6 variable length codes (Efficiency) as follows (in the representation of binary bits):

Zero table of coefficients DCT MPEG-2 (part)

Variable Length Code (Efficiency) "Series" "Level" 10 End of block

... ... ... 0001 10 s one 2 0001 11 s 5 one 0001 01 s 6 one 0001 00 s 7 one ... ... ... 0000 100 s 2 2 ... ... ... 0000 01 s Control sequence ... ... ... 0000 0000 0111 11 s 0 16 ... ... ... (note: the last bit of 's' in each efficiency above indicates the sign of a "level": '0' for a positive and '1' for a negative value)

MPEG2 encoding of “series” and “level” after escape sequence code

Fixed Length Code "Series" Fixed Length Code "Level" with a sign 0000 00 0 1000 0000 0000 Reserved 0000 01 one 1000 0000 0001 -2047 0000 10 2 1000 0000 0010 -2046 ... ... ... ... 1111 1111 1111 -one ... ... 0000 0000 0000 Prohibited ... ... 0000 0000 0001 +1 ... ... ... ... 1111 11 63 0111 1111 1111 +2047

For ease of reading, the above efficiencies (also called statistical codes) are separated by “|”. The first and second DCT coefficients use a fixed-length code to encode the "series" and "level" according to the table "MPEG2 Encoding of the" series "and" level "after the control sequence code", and the remaining DCT coefficients are encoded according to the "zero table of DCT coefficients MPEG-2 ". The last efficiency is a special code "end of the block", indicating that the remaining coefficients in the block are 0.

An alternative value may be any part of the above encoded bits. For example, one alternative value may consist of part of the efficiency, one or more consecutive efficiency, or one efficiency plus part of the next efficiency.

In the following KVZ example, the alternative value consists of one efficiency, which is the 5th efficiency:

(P0 + 71, 7, C, 0001010, 0001110, 0001100, 0001000)

Each element in this CVC is explained as follows:

• P0 is the starting position of this block in the encoded payload. “63” is the relative position (in bits) from the beginning of this block, 7 indicates the length L, C consists of 7 coefficients as follows:

s (f, ch, b, 7), s (f, ch, b, 8) ... s (f, ch, b, 13), where f is the index of the current frame, ch is the index of the current channel and b is index of the current block in the example.

• the size of the alternative value is 7 bits, i.e. L = 7.

• 0001010 is the initial value (V), there are three alternative values: 0001110 (VI), 0001100 (V2) and 0001000 (V3). These alternative values are encoded as the following values of "series" and "level" (referring to the zero table of DCT coefficients MPEG2), respectively:

0001110: “series” = 5, “level” = 1

0001100: “series” = 1, “level” = 2

0001000: “series” = 7, “level” = 1

The “series” (R) and “level” (L) pair represents a sequence of numbers that starts with R zeros followed by L. For example, the “series-level” pair (5, 3), i.e. “Series” = 5 and “level” = 3, represents 000003. Thus, the above alternative values represent the following coefficients before statistical coding:

0 0 0 0 0 1 (“series” = 5, “level” = 1)

0 2 (“series” = 1, “level” = 2)

0 0 0 0 0 0 0 1 (“series” = 7, “level” = 1)

Since the 5th efficiency is the last efficiency encoding nonzero coefficients in this block, alternative values do not necessarily correspond to the same coefficients encoded with V. In the case of V2 and V3, instead of encoding seven coefficients in V1, V2 encodes only 2 coefficients and V3 encodes 8 odds.

If the first alternative value in the KVZ example is selected by inserting a watermark in step 3, then this block will be replaced by the following quantized DCT coefficients:

Another example of encoded content is content encoded using H.264 (MPEG4 profile 10). There are several major differences in statistical coding between MPEG-2 and H264, which include:

• H.264 supports both CAKPD (context adaptive variable-length coding) and CABAC (context adaptive binary arithmetic coding).

• Instead of the 8x8 DCT block in MPEG2 in H.264, you can also use the 4x4 DCT block.

• Unlike ACPD (adaptive variable length coding) in MPEG-2, CACPD uses smaller efficiency tables by encoding “levels” and “series” separately.

• Instead of fixed efficiency tables in MPEG-2, efficiency tables can be switched according to contextual information.

Due to context-based statistical coding, it is harder in H.264 than in MPEG-2 to predict the coded value after changing the coefficient. However, you can force the H.264 encoder to reduce such context-based coding during coding. For example, an encoder may use a fixed NumTrail table (for example, a NumTrail table 0) instead of a NumTrail table based on certain properties of neighboring blocks. Another example is the use of a fixed-length code xxxxyyy to encode the number of non-zeros (NumCoeff) and the number of last units (T1) for a quantized 4x4 DCT block, where xxxx is intended to encode the number of non-zeros (0-16), and yy to encode the number of non-zeros units (0 ... 3). You can also force the encoder to use only 4x4 blocks. In the following example, some of the above rules are used to encode a 4x4 block.

1. 4x4 quantized DCT coefficients:

After a zigzag scan, the coefficients in this block are: 0 3 0 1 -1 -1 -1 0 1 0 ... 0

2. CAKPD coding consists of the following five stages:

a. Encode the number of non-zeros (NumCoeff) and the number of last units (T1): (5.3). The maximum number of T1s is limited to 3. These two numbers (NumCoef and T1) are encoded into “0001011” bits using the Huffman table 17x4 (see below) (NumCoef: 0-16, T1: 0-3).

Huffman Table 17x4 NumTrail

b. Encode the signs of the last units in the reverse order, if they exist (0 for positive and 1 for negative): 0,1,1.

c. Encode the remaining non-zero coefficients in the reverse order: 1.3. Non-zero coefficients are encoded as “1” using the efficiency level “0” table, and “0010” using the efficiency level “1” table, respectively.

Efficiency "level" table 0 The table of "level" efficiency 1

d. Encode the total number of zeros (TotalZeros) from the beginning to the last nonzero coefficient. The maximum total number of zeros is 16 - NumCoeff. If NumCoeff is 16, then TotalZeros should be 0. If NumCoeff is 0, then no additional code is required, i.e. complete the coding for this block. For the remaining 15 cases, in each case, use the Huffman table to encode TotalZeros (see table TotalZero). In this case, TotalZeros is 3, which is encoded as “1110” using the TotZeros table [NumCoeff = 5].

TotZeros table

e. Encode the positions of all zeros from the beginning to the last nonzero coefficients. In this case, the positions of all zeros are encoded in “001101”.

The final coded bits for this 4x4 block contain 25 bits: 0001011 | 011 | 10010 | 1110 | 001101. One example of an alternative value is the encoded bits of an entire encoded 4x4 block. Another example of an alternative value may be bits encoding the characters of the last units, as described in step c) above. In this case, when changing the alternative value, only the sign of the last units is changed and therefore the entire bit length of the encoded block is not changed.

Another example of encoded content is a JPEG2000 image. Instead of converting DCT to JPEG2000, a discrete wavelet transform (DWT) is used. The fiberboard of an image consisting of pixels is calculated by sequentially applying vertically and horizontally low-pass and high-pass filters to image pixels, where the resulting values are called “wavelet coefficients”. A wavelet is an oscillating wave that lasts only for one or more cycles. At each iteration, the wavelet coefficients of the previous iteration filtered only at a low frequency are reduced by a factor of 10, then they pass through the low-pass and high-pass filters vertically, and the results of this process are transmitted through the low-pass and high-pass filters horizontally. The resulting set of coefficients is grouped into four “subbands”, namely the subbands LL, LH, HL and HH. Each iteration corresponds to a specific “layer,” or “level,” of coefficients. The first level of coefficients corresponds to the highest level of image resolution, while the last level corresponds to the lowest level of resolution. One example of an alternative value for a JPEG2000 encoded image is the LL encoded subband at a certain level (for example, at the lowest resolution).

A simple way to create a KVZ is to use the “trial and error method” by performing the following steps for each KVZ:

a) For each statistical code (e.g., efficiency in MPEG2), a value is randomly selected as a candidate for an alternative value.

b) Determine whether this candidate has a valid value by checking the format conformity after replacing the candidate with the existing value. If not, then return to step a).

c) Check the change in perception after replacing the existing value with this candidate. If the change in perception is acceptable, then write down this candidate as an alternative value for the CVC. Otherwise, they return to step a).

The above steps are repeated until all necessary KVZ are created.

In one embodiment, the implementation of the HHC can be created using the watermarking system, which is no longer used, which works in the field of conversion. First, using the selected watermarking system, two copies with watermarks (A and B, both in coded form) of the original encoded content are created with two different copies of useful information: the first copy contains useful watermark information in which each bit is “1 ", And another copy contains useful watermark information in which each bit is" 0 ". Steps 1 and 2 (see Fig. 1) can be performed as follows to select and create an SCC using scan A and B until the following conditions are met:

• (Location selection) P1a and P2a are the starting position and ending position of the set of consecutive coded coefficients (their positions before the statistical coding will be marked as C) in the coded payload A, P1b and P2b are the starting position and ending position of the same coded coefficients in encoded payload B, and (P2b-P2a) and (P1b-P1a) must be equal (and equal to the length L), and

• (Calculation of alternative values) It is believed that at least one bit of useful watermark information needs to be embedded in the coded coefficients (in both A and B). Record one set of watermark (P, L, C, V, V1, V2) as follows:

P = P1a, L = P2a-P1a, V is not available in this case, V1 is the value from position P1a to P2a in A, and V2 is the value from position P1b to P2b in B.

• To select N watermark sets, repeat the second step above.

In an alternative embodiment, the watermark payload bit is encoded as an interdependence between two selected quantized coefficients at a medium frequency. By scanning the encoded content, a watermark set (P, L, C, V, V1, V2 ... Vn) can be described, and its location can be determined using the following conditions:

• V contains many consecutive coded coefficients and there is at least one interdependence (the type / category of interdependencies is determined previously) between and / or among these coefficients. For example, if C1, C2 are two successive coded coefficients and there is an interdependence C1> C2 + T, where T is a threshold value that can be used to adjust the degree of interdependence, which in turn determines the reliability of the watermark.

• According to the perception model of the watermarking system, the coefficients can be changed to establish a different interdependence without changing the perception of the content. For example, C1 and C2 can be changed to C1 'and C2', respectively, so that an inverse relationship is formed, i.e. C1 '<C2'. A new value is written containing C1 'and C2' as an alternative value of Vi (1 <= i <= n). This step is repeated until all alternative values are found.

• Repeat the above steps to select N watermark sets. Ideally, the first N watermark sets containing most of the coefficients are selected so that most of the interdependencies can potentially be set within each watermark set. In a reference watermarking system, one interdependence can represent one bit of useful watermark information, so a set of watermarks with the largest interdependencies may have the highest ability to embed more bits of useful watermark information.

In yet another example of a watermarking system, bits of useful watermark information are encoded as a relationship between property values of two sets of pixels, or DCT / wavelet coefficients. Typical examples of properties include the average value of the luminance signal, the average value of the histogram of the distribution of colors, or the energy in a particular frequency range.

One of the principles of the present invention is the preprocessing of the encoded content and the generation of alternative values containing watermarked signals. These alternative values will later be used to replace explicitly defined portions of the encoded content. Although existing watermarking systems can be applied to preprocessing (i.e., steps 1 and 2) and to the actual insert (step 3), new watermarking algorithms can be developed specifically for this scheme.

A simple and universal way of detecting a watermark is to correlate the current coefficient in the content with watermarks with alternative values of the water shortage to find an alternative value that has a better match with the current coefficient. To perform this correlation, alternative values must be decoded into coefficient values using the appropriate statistical decoder. Each KVZ alternative value may contain a watermark signal that corresponds to the useful information of the watermark. By finding the appropriate alternative values in all of the CLCs, useful watermark information can be extracted. Watermark detection can be performed at multiple levels. At the first level, KVZ elements are required: P, L, and C. It is assumed that the number of KVZ is N. For each KVZ, coefficient values are obtained according to the global index C of the coefficients. If the input content is a band of unmodulated frequencies or is encoded in a different format than the format used when embedding the watermark, then the content is converted first to the same coded form as when embedding the watermark. The raw bits of the useful watermark information are extracted by determining the relationship between and / or among the coefficients or between the values of the pixel properties or the values of the coefficients. 3 is a flowchart of a watermark detection process of the present invention. At step 305, the index is set to an initial value. At step 310, “N” watermark sets are received / received, and at step 315, suspicious content is received. At 320, an index is checked to determine if all watermarks are processed / detected. If the index is greater than the number of watermarks "N", then the process ends. If the index is not greater than "N", then the process proceeds to step 325, where the coefficients are obtained. At 330, bit values are extracted according to the interdependencies between / among the coefficients. At step 335, the index is increased, and the process proceeds to step 320.

If very weak interdependencies are extracted from the KVZ, it is useful to refer to other elements of the KVZ: V and V1, V2 ... Vn, to provide additional information for determining the interdependence. The interdependence can be better determined by correlating the current coefficients with the coefficients encoded in V, V1, V2 ... and Vn. The highest correlation value is the correlation, which indicates the closest match between the current coefficient and one of V, V1, V2 ... Vn. Thus, the highest correlation between current coefficients and coefficients encoded in Vi (1 <i <n) indicates compliance. Suspicious content with serious attacks may require initial content for additional analysis and correlation.

Another way to use the KVZ to deliver content that is protected by copyright is to use the KVZ to recover damaged content before playback. In this case, the encoded content that is delivered to the playback / execution device has either a clearly damaged format / structure (invalid format) or damaged content (poor quality content). The decoder will not be able to decode content with a damaged format. If distortion of the content is applied, then this content can be decoded and reproduced, but damaged visual parts of the content may have various visual image defects, such as noise, signals, and random patterns.

To create content with a damaged format using this watermark set (P, L, V, V1, V2 ... Vn), you can simply replace the L bits starting with P in the delivered encoded content with a random value. To further ensure that a random value does damage the format, you can simply exclude random values that are valid statistical codes. To introduce visual image defects in the content or to reduce the quality of the content, a valid alternative value is selected that is far from the original value V and from all other alternative values V1, V2 ... Vn, to replace the current value in the encoded content. One example of using a damaged format or degraded quality is to counteract a particular device or one category of devices that have been compromised or used to illegally use content. These alternative values, which cause format corruption or poor content, are selected to replace only certain useful watermark information that identifies a device or device category.

A watermark can be removed if the watermarks are known, by simply restoring the content to its original value V in each watermarks. To overwrite an existing watermark in a coded payload, you can simply select the appropriate value from V1, V2 ... Vn in each CIC according to useful information to replace the value containing the existing watermark in the coded payload. Using the watermark with the ability to delete is useful in some applications to support multiple watermarking. An example of multiple creation of watermarks is the introduction of a judicial mark at each stage of the final stage of creating information content. In order to avoid the accumulation of perceptual degradation that may be introduced by a plurality of watermarks, some may want to delete some or all of the previous watermarks before introducing a new watermark. 4 shows an example of a watermark embedding means in a reproducing apparatus. The watermark embedder embeds the watermark by replacing some of the values / coefficients in the encoded content received by the devices with alternative values defined in the KBZ. A replaceable watermark can also be useful for switching copy protection states for copyrighted content. For example, for a content item with watermark information indicating “copy once”, after one copy is made, you might want to change the watermark information from “copy once” to “do not copy”. There are several approaches in order to avoid any potential conflicts of multiple watermarks (i.e., the mutual influence between multiple watermarks). One approach is to use a unique subband to generate each watermark. Another approach is to select a watermark with positions that differ from previous watermarks.

It should be understood that the present invention can be embodied in various types of hardware, software, firmware, special processors, or a combination thereof. Preferably, the present invention is embodied as a combination of hardware and software. In addition, the software is preferably implemented as an application program explicitly embodied on a program memory. An application program can be downloaded to a computer containing any suitable architecture and run with it. Preferably, the computer is embodied on a computer platform having hardware such as one or more central processing units (CPUs), random access memory (RAM), and input / output (I / O) interface (s). The computer platform also includes firmware and an operating system. The various processes and functions described may be either part of the firmware code, or part of the application program (or a combination thereof) that are executed through the operating system. In addition, various other peripheral devices can be connected to a computer platform, such as an additional data storage device and a printing device.

You must also understand that due to the fact that some of the components of the system and the steps of the method shown in the accompanying drawings are preferably carried out in software, the actual connections between system components (or process steps) may differ depending on the method that is programmed the present invention. In view of the foregoing, those skilled in the art may consider these and similar implementations or configurations of the present invention.

Claims (28)

1. The method of embedding a watermark in coded information content, the method comprises the steps of: accepting coded information content; accept a pre-processed set of watermark, which is the output of the external pre-processing unit; receiving a sequence of bits that contains useful watermark information; and they directly replace a certain number of bits of coded information content, starting from a position with an alternative value, moreover, the watermark signal embedded in it has an alternative value, the alternative value being one selected from the set of alternative values defined in the watermark set and the only alternative value defined in the watermark set, the position being selected in the watermark set so that when alternate vnoe value is the only alternative value, then Nezam said watermark is optional, and when an alternative value selected from a plurality of alternative values specified in the kit watermark, said bit sequence determines which alternative is selected. 2. The method according to claim 1, in which the specified number of bits is determined in the specified set of watermark. 3. The method according to claim 1, in which the specified position is determined in the specified set of watermark. 4. A system for embedding a watermark in coded information content, comprising: means for receiving coded information content; means for receiving a pre-processed set of watermark, which is the output of the external pre-processing unit; means for receiving a sequence of bits that contains useful watermark information; and means for directly replacing a certain number of bits of coded information content, starting from a position with an alternative value, moreover, the watermark signal embedded in it has an alternative value, the alternative value being one selected from the set of alternative values defined in the watermark set and the only alternative the value defined in the watermark set, and the position is selected in the watermark set so that when If the optional value is the only alternative value, then the replacement of the specified watermark is optional, and when the alternative value is selected from the set of alternative values defined in the watermark set, the specified bit sequence determines which alternative value is selected. 5. The system according to claim 4, in which the specified number of bits is determined in the specified set of watermark. 6. The system according to claim 4, in which the specified position is determined in the specified set of watermark. 7. The system according to claim 4, in which the system is a playback device, the specified playback device receives the encoded information content and the watermark set, wherein the encoded information content and the watermark set are stored either as part of the encoded information content, or as metadata, or in a catalog . 8. A method of embedding a watermark in coded information content, the method comprises the steps of: receiving coded information content; receive pre-processed watermark data; receiving a sequence of bits that contains useful watermark information; and perform one of the steps: they directly replace a certain number of bits, starting from the position, with an alternative value, or leave the specified number of bits unchanged, starting from the specified position, based on the bit values of the bit sequence of the useful information of the watermark, and the alternative value has signals embedded in it a watermark, wherein said alternative value is selected from a plurality of alternative values defined in the watermark data. 9. A system for embedding a watermark in coded information content, comprising: means for receiving coded information content; means for receiving pre-processed watermark data; means for receiving a sequence of bits that contains useful watermark information; and means for performing one of the steps: directly replacing a certain number of bits, starting from a position with an alternative value, or leaving the specified number of bits unchanged, starting from a specified position, based on the bit values of the specified bit sequence of the specified useful information of the watermark, and the specified alternative the value has watermark signals embedded in it, wherein said alternative value is selected from a plurality of alternative values, nnyh in said watermark data. 10. A method of replacing a watermark in encoded information content, the method comprises the steps of: receiving encoded information content; receive pre-processed watermark data and replace a certain number of bits, starting from the position in the specified coded information content, with the value that is determined in the watermark data, and also the specified position is determined in the watermark data, and the specified value is an alternative value selected from a plurality of alternative values defined in the watermark data, wherein the selected alternative values form a watermark different from that watermark, which replace. 11. A system for replacing a watermark in encoded information content, comprising: means for receiving encoded information content; means for receiving pre-processed watermark data and means for replacing a certain number of bits, starting from a position in the specified coded information content, with a value that is determined in the watermark data, the specified number of bits is determined in the watermark data, and also the specified position is determined in the data watermark, and the specified value is an alternative value selected from a variety of alternative values defined in the watermark data, and selected a alternative values form a watermark that is different from the watermark that is being replaced. 12. A method for detecting a watermark, the method comprises the steps of: receiving a pre-processed set of a watermark, which is the output of an external pre-processing unit; accept informational content marked with watermarks; extracting a plurality of coefficient values from the information content marked with watermarks based on the position information in said pre-processed watermark set; and extracting bit values of said watermark from said coefficient values, wherein bit values are extracted by correlating the current coefficient with a plurality of coefficient values encoded in alternative values of said watermark set to determine an alternative value having the best match with the current coefficient. 13. The method according to item 12, in which the bit values are extracted from the set of coefficient values according to the interdependencies between / among the specified values of the coefficients. 14. The method according to item 13, in which the interdependencies are weak and the current coefficient correlate to determine compliance with the values of the coefficients from the set of coefficient values encoded in the specified pre-processed set of watermark. 15. A system for detecting a watermark, comprising: means for receiving a pre-processed set of watermarks, which is the output of an external pre-processing unit; means for receiving information content marked with watermarks; means for extracting a plurality of coefficient values from the information content marked with watermarks based on the position information in said pre-processed watermark set; and means for extracting the watermark bit values from the coefficient values, the bit values being extracted by correlating the current coefficient with a plurality of coefficient values encoded in alternative values of the specified watermark set to determine an alternative value having the best match with the current coefficient. 16. The system of claim 15, wherein the bit values are extracted from the plurality of coefficient values according to the interdependencies between / among the coefficient values. 17. The system of claim 16, wherein the interdependencies are weak and said current coefficient is correlated to determine compliance with coefficient values from a plurality of coefficient values encoded in said watermark set. 18. A method of creating a watermark set, the method comprises the steps of: randomly selecting a value as a candidate for an alternative value for each statistical code; determining whether this selected candidate has a valid value by checking the format to match after replacing the selected candidate with an existing value; check changes in perception after replacing the existing value with this selected candidate, and if the change in perception is acceptable, then write this selected candidate as an alternative value; choose a position in which the value in the encoded information content must be replaced with an alternative value; and calculating a plurality of alternative values for each selected position, wherein each of the alternative values includes a watermark signal, each of the alternative values having the same number of bits as the indicated value, which is replaced by an alternative value selected from the plurality of alternative values, wherein the replacement does not lead to unacceptable changes in the perception of the specified encoded information content, while the replacement does not lead to an incompatible format of the specified encoded of information content and wherein the method is performed by the pretreatment unit. 19. The method of claim 18, wherein the selected alternative value is a candidate that causes the smallest change in perception. 20. A device for creating sets of watermark, comprising: means for randomly selecting a value as a candidate for an alternative value for each statistical code; means for determining whether this selected candidate has a valid value by checking the format matching after replacing the selected candidate with an existing value; means for checking the change in perception after replacing the existing value with this selected candidate and, if the change in perception is acceptable, recording this selected candidate as an alternative value; means for selecting a position in which the value in the encoded information content must be replaced with an alternative value; and means for calculating a plurality of alternative values for each selected position, each of the alternative values including a watermark signal, each of the alternative values having the same number of bits as the indicated value, which is replaced with an alternative value selected from the set of alternative values while the replacement does not lead to unacceptable changes in the perception of the specified encoded information content, while the replacement does not lead to an incompatible format specified th coded information content and wherein said device is a pretreatment unit. 21. The device according to claim 20, in which the selected alternative value is a candidate that causes the smallest change in perception. 22. A device for storing programs, comprising: means for storing encoded information content; means for storing sets of watermarks and means for storing an application program executed by a processor connected to said program storage device, said application program implementing a method for embedding a watermark in encoded information content, comprising the steps of: receiving encoded information content; accept a pre-processed set of watermark, which is the output of the external pre-processing unit; receiving a sequence of bits that contains useful watermark information; and they directly replace a certain number of bits of coded information content, starting from a position with an alternative value, moreover, the watermark signal embedded in it has an alternative value, the alternative value being one selected from the set of alternative values defined in the watermark set and the only alternative value defined in the watermark set, the position being selected in the watermark set in such a way that when alternate vnoe value is the only alternative value, then Nezam said watermark is optional, and when an alternative value selected from a plurality of alternative values specified in the kit watermark, said bit sequence determines which alternative is selected. 23. The device according to item 22, in which the sets of watermarks are encrypted. 24. The device according to item 22, in which sets of watermarks are metadata of the specified encrypted information content. 25. The device according to item 22, in which sets of watermarks are steganographic data hidden in one of the syntactic elements or in encoded information content. 26. The device according A.25, in which access to the sets of watermarks is controlled using a key. 27. The device according to item 22, in which sets of watermarks are stored as a separate file. 28. The device according to item 22, which is one of the optical disk, tape and hard drive.

Images