KR20030078517A - Watermark embedding and detction method for authentication and detection of manipulated position in digital video - Google Patents

Abstract

PURPOSE: A watermark insertion/detection method for recognizing digital video authorizing and manipulating positions is provided to recognize authorization procedure and manipulation position by extracting authorization data inserted for video data in the raw format or MPEG-2 bit stream format. CONSTITUTION: A watermark insertion/detection method for recognizing digital video authorizing and manipulating positions includes determining whether a video data in the raw format is inputted(S100), converting the inputted video data to frequency domains proper for corresponding application(S101), and inserting authorization data by using a quantized coefficient value(S102). If a video data in the MPEG-2 bit stream is inputted(S103), a DCT coefficient value quantized by variable length decoding is recovered from the bit stream(S104). A hash function output value is obtained with a 128 bit length by using the quantized DCT coefficient value as an input value of the hash function(S105). Am LSB is extracted from the coefficient selected for video data authorization for generating a bit stream formed of the LSB(S106). LSB bit streams for the video data authorization is compressed without loss(S107). In a space reduced by the lossless compression is inserted information required for an output value of the hash function(S108). Change information of a DC coefficient is inserted into the LSB(S110). The LSB of the coefficient selected in the step S106 is converted to the LSB modified in the steps S108,S110(S111). Insertion of authorization data for all frames are determined by variable length decoding(S112,113). If so, video data in the compressed video bit streams format inserted with the authorization data is generated(S114).

Description

Watermark embedding and detction method for authentication and detection of manipulated position in digital video}

The present invention relates to a method for certifying digital video for authentication and operation location, and more particularly, to invisibly insert information for authentication into video data in a raw format or video data in an MPEG-2 bit stream. And a method of extracting and identifying the authentication and manipulation position of the digital video.

Recently, with the rapid growth of the Internet and the development of information and communication technologies, the convergence of the Internet, broadcasting and communication services has been required to provide interactive information and communication services through digital broadcasting networks. In order to perform such an interactive information and communication service, authentication processing for video data is an issue. Generally, when authentication data is inserted into video data in the form of MPEG-2 bit stream, the same video data as the original video data is obtained. The reversible characteristics can be deteriorated, and there is a problem in that recompression often occurs during video data transmission and artificial data manipulation cannot be distinguished.

Accordingly, the present invention is to solve such a problem, the present invention is to invisibly insert the data necessary for authentication in the video data and to extract it again to determine the authentication and operation position for the video data Its purpose is to provide a method for locating authentication and operation.

Another object of the present invention is to insert and extract the authentication data to determine the authentication and the operation position, but the video data before inserting the data for authentication after the extraction process of the authentication data is finished in the absence of manipulation on the video data The present invention provides a method for certifying and manipulating digital video that obtains the same data as.

It is another object of the present invention to distinguish between recompression frequently occurring during the transmission of video data and artificial data manipulation, and an authentication and manipulation position identification method of digital video capable of detecting an operation position in block units when artificial manipulation is applied. To provide.

1 is a block diagram of an apparatus for implementing a watermark embedding and detection method for authentication and operation position identification of digital video according to the present invention;

2 is a view illustrating an embodiment of a watermark embedding method for identifying an authentication and manipulation position of a digital video according to the present invention;

3 is a view for explaining a watermark embedding and detection method for authentication and operation position grasp of digital video according to the present invention;

4 is a view for explaining a watermark embedding and detection method for authentication and operation position grasp of digital video according to the present invention;

5 is a diagram showing a sequence for implementing a watermark embedding and detection method for authentication and manipulation position identification of digital video according to the present invention;

FIG. 6 is an embodiment of a watermark detection method for authentication and manipulation position identification of digital video according to the present invention; FIG.

7 is a view for explaining a watermark embedding and detection method for authentication and operation position identification of digital video according to the present invention;

8 is a view for implementing a watermark embedding and detection method for authentication and manipulation position identification of digital video according to the present invention;

9 is a view for implementing a watermark embedding and detection method for authentication and operation position identification of digital video according to the present invention;

10 is a diagram for implementing a watermark embedding and detection method for digital video authentication and operation location identification according to the present invention;

<Explanation of symbols for the main parts of the drawings>

10: digital input means 20: analog input means

30: D / A converter 40: MPEG-2 encoder

50: digital recording means 60: user computer

A feature of the present invention for achieving the above object is a first method for obtaining an output value of a hash function by using a quantized predetermined coefficient of input video data as a input of a hash function in units of pictures to insert data for authentication. step; Bitstream consisting of LSB only by dividing each picture of the input video into predetermined blocks, selecting two coefficients for authentication and operation position of the video data from each predetermined block, and extracting LSBs of coefficients selected for authentication Generating a multiplier by multiplying each value of the selected coefficients by 2 to obtain an operating position so that the LSBs of the selected coefficients are all zero; Performing a lossless compression on an LSB bitstream of coefficients selected for authentication of the video data and inserting a hash function value and other information in a space reduced by the compression; A fourth step of changing the LSB of the DC coefficient of the predetermined block according to a random number generated by using a secret key of the user and inserting the changed information into a coefficient selected for grasping the manipulation position of the video data; And a fifth step of replacing the LSB and the coefficient value of the coefficient selected in the second step with the LSB and the coefficient changed in the third and fourth steps and performing variable length decoding to generate the video data into which the authentication data is inserted. It is included.

Preferably, the predetermined coefficient is a DCT coefficient.

Preferably, the predetermined block is characterized in that the 8x8 DCT block.

Preferably, the other information is characterized in that the information of the video or the owner.

In another aspect of the present invention, in the process of detecting data for embedded authentication and operation position grasp, the LSB bitstream and operation position grasp for authentication of video data are selected by selecting quantized DCT coefficients at the same position as the insertion process. Generating a bitstream of the coefficients selected for; The LSB bitstream for authentication of the video data among the bitstreams generated in the first step is de-losslessly decompressed so that the LSBs of the original picture with the coefficients selected for authentication are hashed with the inserted original picture. A second step of obtaining function values and other information; The LSBs obtained by decompression are exchanged with the LSBs of the coefficients selected for authentication, and the coefficients selected for locating the operation are divided by two, the coefficient value is quotient, and the LSB of the DC coefficient of a given block is changed according to the rest. Three steps; A fourth step of obtaining an output of a hash function value using the quantized coefficients of the picture changed by the second step and the third step as inputs; A fifth step in which an authentication and operation location determination process is terminated when the hash function value obtained in the second step and the hash function value in the fourth step are compared to be the same; And a sixth step of performing lossy compression of the input video data, artificial manipulation, and operation position determination when the two hash values are not the same in the fifth step.

Preferably, the predetermined coefficient is a DCT coefficient.

Preferably, the other information is characterized in that the information of the video or the owner.

According to still another aspect of the present invention, a first method of obtaining an output value of a hash function by using a quantized DCT coefficient of each pixel on a picture-by-picture basis in input raw format video data or video data in a bitstream format is used as an input of a hash function. step; After dividing the picture into NxN-sized DCT blocks, except for the DC coefficient, the user selects an AC coefficient to be used for authentication at an arbitrary position and an AC coefficient to be used for identifying an operation position, and LSB is selected from the coefficient selected for authentication. A second step of extracting and generating an LSB bitstream composed of LSBs, multiplying a coefficient selected for operation position by 2 to make the LSB of the coefficient become zero; The LSB bitstream for authentication of video data includes: a third step of performing lossless compression and inserting a hash function value and other necessary information in the first step into a space reduced by compression; And a fourth step of changing the LSB of the DC coefficients in the NxN DCT block in the picture according to the random number generated by the secret key and inserting the changed information into the coefficients selected to grasp the manipulation position of the video data in the second step. It is to include.

Preferably, each coefficient value of the position selected in the second step is changed to a coefficient value changed for authentication of the video data in the third step and a coefficient value changed for grasp of the operation position of the video data in the fourth step. The method may further include generating video data in which authentication data is inserted by performing variable length encoding.

Another feature of the present invention is to extract video data by extracting LSBs from two coefficients into which the authentication data is inserted in an NxN DCT block, in order to detect the authentication data inserted into the input video data and to identify the authentication and operation positions. A first bitstream that generates a bitstream of LSB bitstreams composed of lossless compressed LSBs to be used for authentication, a hash function value, other information, and DC coefficient change information to be used for determining an operation position of video data. step; Lossless compression is performed on the LSB bitstream generated for authentication of the video data to restore the LSBs of the coefficients, and separate the hash function value and other information, and to the LSB of the DC coefficients to determine the operation position of the video data. After changing the LSB of the DC coefficient by using the bitstream of the coefficients containing the change information for the information, and restore the selected coefficient value to the original value to restore the same video data as the original video data if there is no change to the video data Second step; The output value of the hash function is obtained by using the hash function value in the second step and the changed quantized coefficient value of the picture as an input of the hash function, and the two hash function values are compared to authenticate each other if the input video data are the same. A third step of doing; And when the values of the two hash functions in the third step are not the same, generate the same random number, and input the video data when the LSBs of the DC coefficients in the NxN DCT block in the first step are changed according to the generated random number. Determining that lossy compression has been performed on the controller, and if there is a block that is different from the random number in which the LSB of the DC coefficient is generated, it is determined that the input video data is artificially manipulated, and determines that the differently changed block is the manipulated position. It includes four steps.

Hereinafter, exemplary embodiments of the present invention will be described in detail with reference to the accompanying drawings.

1 is a block diagram of an apparatus for implementing a method for identifying an authentication and manipulation position of a digital video according to the present invention.

In Fig. 1, an apparatus for implementing the method of the present invention includes a digital input means 10, an analog input means 20, a D / A converter 30, an MPEG-2 encoder 40, and a digital recording means 50. And a user computer 60.

The digital video obtained through the input device consisting of the digital input means 10, the analog input means 20, and the D / A converter 30 is transferred to the user computer through the MPEG-2 encoder 40 and the digital recording means 50. Entered at 60, the MPEG-2 encoder 40 may be replaced by a hardware encoder module composed of one module.

The video data generated from the MPEG-2 video data is recorded in the digital recording means 50. The video data recorded in the digital recording means 50 uses the MPEG-2 decoder program and the user computer 60 for authentication data. Generate embedded video data.

2 is a view illustrating an embodiment of a watermark embedding method for identifying an authentication and manipulation position of a digital video according to the present invention.

In FIG. 2, video data or MPEG-2 bitstream in raw format form as an input through an input device composed of a digital input means 10, an analog input means 20, and a D / A converter 30. As a flow chart for inserting authentication data when video data in the form is input, when video data in Raw format is input (step S100), the input video data is converted into a frequency domain suitable for each application (step S101). ), Authentication data is inserted using the quantized coefficient values (step S102). On the other hand, when video data in the form of MPEG-2 bitstream is received (step S103), variable length decoding is performed to restore the quantized DCT coefficient value from the bitstream (step S104).

In step S100 (2), if the input data is in the low format, 8x8 DCT is performed (S101). In step S102, the DCT coefficient values are quantized (S102). A quantized DCT coefficient value is obtained for each pixel in the quantized coefficient for the input data in the raw format. The quantized coefficient is expressed by Equation 1 below.

In Equation 1, (i, j) represents the position coordinates of the pixels in each block, and [x] represents the largest number not exceeding x, Denotes the DCT coefficients of the k-th DCT block in the picture. Also, Is The quantization step of.

Using a quantized DCT coefficient value of each picture represented by [Equation 1] as an input value of a hash function, a hash function output value of 128 bits is obtained (step S105), and the hash function output value obtained in step S105 is It is expressed by [Equation 2].

The size of the image is defined as NxN, the hash function is represented by H, and the result is represented as a function of the bitstream. As shown in Equation 2, the output value of the hash function is obtained by using the quantized coefficient value of the picture as input.

FIG. 3 is a view for explaining a method for locating digital video for authentication and manipulation according to an embodiment of the present invention, wherein two coefficients are selected for each 8x8 DCT block in each picture divided into 8x8 DCT blocks. The coefficient selected for grasping the operating position is multiplied by the coefficient value by 2 so that the LSB of the coefficient is zero to extract the LSB. Therefore, the LSB bitstreams generated for identifying the operation position are all configured with zeros. On the other hand, the coefficient selected for authentication of the video data is extracted as it is without changing the LSB to generate the LSB bitstream to be used for authentication (step S106), the process is represented by the equation (3).

In [Equation 3] Represents the set of LSB bitstreams generated for authentication, Denotes a set of LSB bitstreams generated for operation location determination.

4 is a view for explaining a method for identifying digital video authentication and operation location according to the present invention. Lossless compression is performed on an LSB bitstream generated for video data authentication among bitstreams generated in step S106 as shown in FIG. 4. (Step S107), a lossless compressed LSB bitstream is generated. When the step S107 is expressed by Equation 4, it is as follows.

A lossless compressed bitstream is created as shown in Equation 4, and other necessary information such as video or owner's information is inserted into the output of the hash function in the space reduced by lossless compression as shown in FIG. At this time, in order to change the LSB of each 8x8 DC coefficient in the picture shown in FIG. 4 to be the same as the random number generated using the secret key, the DC coefficient value is changed according to the generated sequence (step S109), and the DC coefficient change information. Is inserted into the LSB (step S110). The process of changing the DC coefficient value according to the sequence in step S109 is shown in detail in the figure showing a sequence for implementing a method for identifying the authentication and operation position of the digital video according to the present invention.

On the other hand, the LSB bitstream generated by multiplying the selected coefficients by 2 for the manipulation position of the video data is changed to be the same as the LSB change information of the DC coefficient, which is doubled to the LSB of the DC coefficient after doubling the selected coefficient. The change is to have an invertible characteristic that can restore the original video data after the end of the authentication process if no manipulation is applied to the video data. In addition, if the quantized DCT coefficients of the high frequency region of the 8x8 DCT block are selected, most of the selected coefficients are 0 or 1, so that the image quality deterioration cannot be visually recognized even when the LSB is doubled and the LSB is changed. The LSBs of the coefficient selected in the step S06 are changed to the LSBs changed in the steps S108 and S110 (step S111), variable length coding is performed (step S112), and it is determined whether authentication data has been inserted into all frames (step S113). If it is determined in step S113 that the authentication data has been inserted into all frames, video data in the form of a compressed video bitstream into which the authentication data has been inserted is generated (step S114), and at the same time, a data insertion process for identifying the authentication and operation positions is completed. .

FIG. 6 is a diagram illustrating an example of a detection process of an authentication and manipulation position determining method of digital video according to an embodiment of the present invention. Shows the process of grasping.

In Fig. 6, video data in the form of MPEG-2 bitstream is input (step S200), and variable length decoding is performed (step S201) to obtain quantized coefficient values for each pixel of the picture. Each picture is divided into 8x8 size DCT blocks identical to the authentication data insertion process, and the LSB is extracted from the coefficient at the same position as the position selected during the authentication data insertion process to contain the lossless compressed LSB bitstream, hash function value, and other information. Two bitstreams are created, which are composed of LSB bitstreams to be used for authentication of video data and coefficients containing changed information of DC coefficients to be used for identifying operation values (step S202).

As shown in FIG. 7, lossless compression is released using an LSB bitstream to be used for authentication of video data among two bitstreams generated in step S202 (step S203), and an LSB containing information about whether a DC coefficient is changed is generated. (Step S204). In step S201, a variable length decoding is used to construct a picture using quantized DCT coefficient values (step S205), and in step S203, lossless compression is released to generate a decompressed LSB (step S206), and authentication is performed using the obtained LSB bitstream. The LSBs of the selected coefficients are exchanged (step S207). In step S203, lossless compression is performed to obtain a hash function value inserted in the authentication data insertion process (step S208). When the coefficient value changed in the insertion process and the LSB are exchanged in step S207, the hash function value is obtained by inputting the coefficient value of the picture in which the LSB is exchanged (step S209).

FIG. 8 is a diagram for implementing a method for certifying a digital video for authentication and operation position according to the present invention, wherein the DC coefficient is based on information of coefficients containing change information about LSBs of DC coefficients to be used for determining an operation position of video data. It shows the process of changing the LSB. As shown in FIG. 8, the coefficients selected for the manipulation position of the video data were changed LSB twice after the authentication data insertion process, and thus the selected coefficients were subtracted from the LSB of each coefficient from the coefficient value and divided by 2 to return to the original coefficient value. do.

It is determined whether the hash function values obtained in steps S208 and S209 are the same (step S210). If the hash function values are the same, the input video data authentication is completed (step S211), and the original image is restored (step S212). ). If each hash function value is not the same in step S210, random numbers are generated with the same key used in the authentication data insertion process as shown in FIG. 9, and the generated random numbers are compared with LSBs of DC coefficients of the respective N × N DCT blocks in step S205. It is determined whether they are the same (step S213). If the sequence is the same as the LSB of the DC coefficient in step S213 (step S214), and if the sequence is not the same, it is determined that the artificial manipulation has been performed (step S215), and the operation unit of the block unit is determined (step S216).

The reason why the lossy compression and the artificial manipulation can be discriminated in the above-described steps is as follows.

In the MPEG-2 bitstream, each picture is divided into three types: I picture, P picture, and B picture. The coefficients of each 8x8 DCT block in the I picture are assigned to each coefficient except the DC coefficient value as shown in FIG. Coefficients quantized and quantized by each value in the corresponding quantization table are converted into bitstreams through variable length coding (VLC). It is made into a bitstream through encoding. Therefore, even if recompression occurs, the value corresponding to the DC coefficient of the I picture in the MPEG-2 video bitstream does not change. Therefore, if the values of the two hash functions are not the same, it compares the random numbers generated by the LSB and the secret key of the DC coefficients in the I picture and determines that they are lossy compressed if they are the same, and the LSB and the random numbers of the DC coefficients are the same. If not, it can be determined that the manipulation has been artificially performed, and if the LSB of the DC coefficient is not the same as the random number, it can be determined that the manipulation has been artificially performed.

As described above, in the detailed description of the present invention, specific embodiments have been described, but various modifications are possible without departing from the scope of the present invention. Therefore, the scope of the present invention should not be limited to the described embodiments, but should be defined by the claims below and equivalents thereof.

As a result, according to the watermark embedding and detection method for digital video authentication and operation position grasp according to the present invention the following advantages occur.

That is, authentication data that is not detected by the human eye is inserted into video data in the raw format or video data in the MPEG-2 bitstream format, and then embedded authentication is performed on the video data requiring authentication and operation position identification. Data can be extracted to determine the authentication procedure and operation location. In addition, when no change is made to the video data, after the detection process of the inserted authentication data is completed, the same data as the original video data before the insertion of the data for authentication may be restored, and the digital video data transmission process may be performed. It is possible to distinguish between video data change and winy video data change due to lossy compression, which are frequently generated in the PC, so that an artificially manipulated position can be identified in 8x8 DCT block units.

Claims (10)

A first step of obtaining a hash function value by using a predetermined coefficient quantized in units of pictures of input video data to insert data for authentication; Select two coefficients for authentication and operation position of the video data in a predetermined block, extract LSB from coefficients selected for authentication, and create LSB bitstream consisting of LSB only. Multiplying to make all LSBs zero; Performing a lossless compression on an LSB bitstream of coefficients selected for authentication of the video data and inserting a hash function value and other information in a space reduced by the compression; A fourth step of changing the LSB of the DC coefficient of the predetermined block according to a random number generated by using a secret key of the user and inserting the changed information into a coefficient selected for grasping the manipulation position of the video data; And And a fifth step of replacing the LSB and the coefficient values of the two coefficients selected in the second step with the LSB bitstream and the coefficient value into which the authentication data is inserted and variable length decoding to generate the video data into which the authentication data is inserted. Characterized in the digital video authentication and operation positioning method. The method of claim 1, wherein the predetermined coefficient A method for locating authentication and manipulation of digital video, characterized in that it is a DCT coefficient. The method of claim 1, wherein the predetermined block is A method of locating authentication and manipulation of digital video, characterized by an 8x8 DCT block. The method of claim 1, wherein the other information is A method of identifying and manipulating digital video, characterized by the information of the video or owner. A first step of generating a bit stream of LSB bitstreams for authentication of video data and a bitstream of coefficients for identifying an operation position by selecting a quantized predetermined coefficient at the same position as the insertion of the authentication data to detect the inserted authentication data. ; Lossless compression is performed on the LSB bitstream for the authentication of the video data among the bitstreams generated in the first step to obtain hash function values and other information about the LSBs and the original picture of the original picture of the selected coefficient. Second step; Exchanging LSBs of coefficients selected for authentication with LSBs of quantized coefficients obtained by decompression, and changing LSBs of DC coefficients by using a bitstream of coefficients for determining an operation position of the video data; A fourth step of obtaining a hash function value using the quantized coefficients of the picture obtained in the third step; A fifth step in which authentication is terminated if the hash function value obtained in the second step and the hash function value in the fourth step are compared and equal; And And a sixth step of determining whether the input video data is lost compression or not, if the two hash values of the fifth step are not the same, and determining an artificial manipulation and a manipulation position. Location method. The method of claim 5, wherein the predetermined coefficient A method for locating authentication and manipulation of digital video, characterized in that it is a DCT coefficient. The method of claim 5, wherein the other information A method of locating authentication and manipulation of digital video, characterized by the information of the video or owner. A first step of obtaining an output value of the hash function by using the quantized DCT coefficients of each pixel of the picture in the input raw format video data or the bitstream video data as an input of the hash function; After dividing the picture into NxN-sized DCT blocks, the user selects two coefficients to be used for authentication and manipulation location at arbitrary locations except DC coefficients, extracts LSBs from the coefficients selected for authentication, and then uses only LSBs. A second step of creating a constructed LSB bitstream and multiplying the coefficient selected for operation position by 2 to make all LSBs zero; A third step of performing a lossless compression and inserting a hash function value and other necessary information in the first step into a space reduced by the compression; And A fourth step of changing the LSB of the DC coefficients in the NxN DCT block in the picture according to the random number generated by the secret key and inserting the changed information into the coefficients selected to grasp the operation position of the video data generated in the second step; Authentication and operation location identification method of the digital video comprising a. The method of claim 8, Change each coefficient value of the position selected in the second step into a changed coefficient value for authentication of the video data in the third step and a changed coefficient value for grasping the operation position of the video data in the fourth step And performing variable length encoding to generate video data into which the authentication data is inserted. Lossless compressed LSBs to be used for authentication of video data by extracting two coefficients into which the authentication data is inserted in an N × N DCT block to detect authentication data inserted for input video data A first step of generating a bitstream of coefficients containing LSB bitstream composed of a hash function value, other information, and DC coefficient change information to be used for determining a manipulation position of video data; Lossless compression is performed on the LSB bitstream generated for authentication of the video data to restore the LSBs of the coefficients, and separate the hash function value and other information, and to the LSB of the DC coefficients to determine the operation position of the video data. After changing the LSB of the DC coefficient by using the bitstream of the coefficients containing the change information for the information, and restore the selected coefficient value to the original value to restore the same video data as the original video data if there is no change to the video data Second step; The output value of the hash function is obtained by using the hash function value in the second step and the changed quantized coefficient value of the picture as an input of the hash function, and the two hash function values are compared to authenticate each other if the input video data are the same. A third step of doing; And If the values of the two hash functions in the third step are not the same, the same random number is generated to determine that lossy compression is performed on the input video data when the LSBs of the DC coefficients in the NxN DCT block are changed according to the generated random number. And a fourth step of determining that the input video data is artificially manipulated when the LSB of the DC coefficient is different from the random number generated by the DC coefficient, and determining that the differently changed block is the manipulated position. Digital video authentication and operation location.