KR100327210B1 - Copy control apparatus of video data and method thereof - Google Patents

Abstract

PURPOSE: An apparatus and a method for controlling copy of video data are provided to minimize deterioration in picture quality of a motion picture compression and decompression system employing MPEG motion picture coding algorithm and prevent illegal copy of data. CONSTITUTION: A copy controlling apparatus using a watermark for protection

Description

Copy control apparatus and method thereof

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an apparatus and method for copy control of video data, and more particularly, to an apparatus and method for copy control for protecting a copyright used in a moving picture expert group (MPEG) moving picture compression / restore device.

Recently, as the spread of digital transmission and storage media including the Internet, digital satellite broadcasting, and DVD (Digital Versatile Disc) has rapidly increased, the issue of copyright protection for digital multimedia data such as audio, video, images, etc. has emerged. have.

In particular, anyone can easily copy or edit digital works using a personal computer (PC), and these copies can be commercially distributed without altering the quality and contents of the original data due to the inherent characteristics of digital methods. It raises new social problems surrounding copyright. Therefore, digital watermark technology is in the spotlight as a countermeasure against this.

The following two methods are widely used to prevent illegal copying of digital A / V data. It is a copy control method using encryption and scramble, and a digital watermark that embeds an invisible signal into multimedia data for the purpose of preventing illegal copying of digital information.

The first method is a technology that prohibits the duplication of digital A / V data by informing the descramble information and the decryption information that can be operated only for the A / V works purchased normally, and the second method is the digital method. Digital water, such as identifiers (abbreviated as "IDs"), logos, or copy control information sent to digital recorders in A / V content data for the purpose of prohibiting illegal commercial use of the information. It is a technology that allows the user to refrain from illegal copying by inserting mark information in the form of noise.

However, the duplication control method using encryption and scramble can be duplicated bit by bit and can be easily duplicated when the encrypted data is leaked, and the duplication control method of inserting a watermark into the digital data is the A / V content data. Coding efficiency can be reduced by inserting ID, logo, or copy control information transmitted to a digital recorder in the form of noise, and since the original image is needed to detect the inserted watermark information, as shown in FIG. There is a problem that is very difficult to apply to a real-time video compression / restoration device.

In the conventional digital watermark detection apparatus, as illustrated in FIG. 1, the first frequency converter 11 converts a test image D ″ (i, j) having an embedded watermark into data in a frequency domain. The two frequency converter 12 converts the original image D (i, j) into data in the frequency domain, and the subtractor 13 outputs the output of the second frequency converter 12 from the output of the first frequency converter 11. The watermark extractor 14 extracts the watermark from the subtraction result of the subtractor 13. The extracted watermark signal X ^* (k) is applied.The comparator 15 extracts the extracted watermark signal. (X ^* ) is compared with the original watermark signal X (k) and the extracted watermark signal is detected as the actual watermark according to the comparison result.

As mentioned above, the watermark detection apparatus shown in FIG. 1 can be used in a device that does not require real-time processing such as still images, but requires a real-time processing and uses an original image in a moving picture compression / restore device that uses a high compression coding system. It is expensive because of the amount of extra hardware to store the data, and the real-time processing was difficult.

In order to solve the above problems, an object of the present invention is to provide a copy control device capable of real-time processing, and does not require the original image.

Another object of the present invention is to provide a copy control device for use in a moving picture compression / restore device using block based motion prediction / compensation.

Another object of the present invention is to provide a copy control method capable of real-time processing and does not require the original image.

In order to achieve the above object and other objects, the copy control apparatus according to the present invention is a copy control apparatus using a watermark for copyright protection: spreading watermark information on a spectrum in a predetermined sub-picture unit, And a modulation unit generating modulated watermark information by using a noise string and an insertion unit generating watermarked data by inserting watermark information modulated by an invisible signal into an input image.

In addition, the copy control apparatus of the present invention demodulates watermark information from watermarked data using a pseudo noise string to generate demodulated watermark information and converts the demodulated watermark information into sub-picture units having a predetermined size. In addition, the apparatus further includes a detection unit that detects the watermark information.

In order to achieve the above another object, the copy control method according to the present invention is a method for encoding and decoding video data by a block-based compression encoding scheme, the method comprising the steps of: assigning the number of information bits and patterns of watermark information, Generating the spread watermark information by spreading the allocated watermark information into a chip rate consisting of a predetermined number of blocks in consideration of the degree of correlation, and generating moving image data into which the spread watermark information is inserted. It is done.

The copy control method according to the present invention comprises the steps of applying DCT coefficients by transforming DCT to video data embedded with watermark information, generating IDCT data by converting IDCT to only AC coefficients among DCT coefficients, and using pseudo noise strings. And demodulating IDCT data to generate demodulated data, and adding the demodulated data in a chip rate unit to detect watermark information.

The duplication control method according to the present invention comprises the steps of DCT converting video data having watermark information embedded therein to block classification on a DCT region, extracting an insertion mask using characteristics of the classified block, and using the extracted insertion mask. IDCT transforming and applying IDCT coefficients to DCT coefficients excluding the DC coefficients of the current DCT block and the AC coefficients of the classified block, applying demodulated data using a pseudo noise string from the IDCT coefficients, and demodulated. The method may further include detecting watermark information by adding data in units of chip rates.

1 is a block diagram of a conventional digital watermark detection apparatus.

2 is a block diagram of a general encoding apparatus for moving picture data, to aid in understanding the present invention.

3 is a diagram showing the structure of a video picture for MPEG-2 video encoding.

4 is a block diagram according to an embodiment of an apparatus for encoding moving image data including a digital watermark according to the present invention.

5 is a block diagram according to another embodiment of an apparatus for encoding moving picture data including a digital watermark according to the present invention.

6 is a detailed block diagram of an embodiment of the watermark information inserter shown in FIGS. 4 and 5.

FIG. 7 is a detailed block diagram of another embodiment of the watermark information inserter shown in FIGS. 4 and 5.

8A to 8D are diagrams for describing a classification mask used in the block classifier shown in FIG. 7.

9 is a detailed block diagram of an embodiment of the watermark information remover illustrated in FIG. 4.

FIG. 10 is a detailed block diagram of another embodiment of the watermark information remover illustrated in FIG. 4.

11 is a block diagram of a general decoding apparatus for moving picture data for understanding the present invention.

12 is a block diagram according to an embodiment of an apparatus for decoding moving picture data including a digital watermark according to the present invention.

13 is a block diagram according to another embodiment of an apparatus for decoding moving picture data including a digital watermark according to the present invention.

FIG. 14 is a detailed block diagram of an embodiment of the watermark information detector illustrated in FIGS. 12 and 13.

15 is a detailed block diagram of another embodiment of the watermark information detector illustrated in FIGS. 12 and 13.

Hereinafter, with reference to the accompanying drawings, it will be described a preferred embodiment of the copy control apparatus and method of the video data according to the present invention.

FIG. 2 is a block diagram of a general apparatus for encoding MPEG-2 video data for understanding of the present invention. The video encoding apparatus performs intra-picture compression and inter-picture compression. For example, this is performed according to the compression sequence of I, P, and B pictures.

In general, compression coding of a video signal has an intra (I) picture that reduces information on the spatial redundancy of the video information with only the current picture as shown in FIG. To reduce the correlation between the previous I or P pictures, to reduce the correlation between the pictures through bi-prediction and P-picture compression-coding the difference component with the previous I or predicted (P) picture. A bidirectional predicated (B) picture that compresses and encodes a difference component with subsequent I or P pictures is constructed and encoded.

Meanwhile, the analog-to-digital (A / D) converter 102 shown in FIG. 2 converts the input original video signal into digital video data. Unlike MPEG-1, MPEG-2 uses the frame DCT mode of DCT conversion of one frame as it is, and the field DCT mode of DCT conversion of only one frame as two fields (odd field and even field). In addition, both have two DCT conversion modes. In most cases, field DCT conversion is more efficient for high-motion video. In order to correspond to the DCT conversion mode, the video data supplied from the A / D converter 102 is temporarily stored in the frame / field memory 104 on a frame or field basis.

The pixel data of the I picture is applied to the discrete cosine transform (DCT) converter 108 without being affected by the subtractor 106. The DCT converter 108 converts the input image into 8 pixels. DCT conversion in units of 8 lines of DCT blocks, the quantizer 110 quantizes the DCT coefficients supplied from the DCT converter 108, and the variable length encoder and multiplexer (VLC & MUX: 112) is a quantizer 110. Bit encoded by quantizing the quantized data supplied from the multiplexed code, multiplexing the variable length coded data with additional information such as the motion vector supplied from the motion predictor 124 and the quantization interval supplied from the rate controller 130. Output the stream. The buffer 114 temporarily stores the bitstream to adjust the bitrate of the encoded bitstream.

The compressed I picture supplied from the quantizer 110 is inversely quantized in the inverse quantizer 116, and the Inverse Discrete Cosine Transform (IDCT) converter 118 decompresses an IDCT transform for the dequantized I picture. The picture is applied to the adder 120. The decompressed I picture is transmitted to the frame memory 122 serving as a buffer as it is without being influenced by the adder 120, and then stored in the frame memory 122 to predictively compress the P and B pictures.

The predictive coding of P and B pictures is similar, and P picture compression is described as an example. The image frames stored in the frame memory 122 are applied to the motion predictor 124. The motion predictor 124 calculates a motion vector using the image data currently input and the image data of previous or subsequent reference frames stored in the frame memory 122, and outputs the motion vector to the VLC & MUX 112 and the motion compensator 126. do.

The motion compensator 126 reads the block corresponding to the motion vector predicted by the motion predictor 124 from the frame memory 122 and supplies it to the subtractor 106. The subtractor 106 corresponds to the picture to be decompressed currently. Subtract the predicted block from the frame memory 122 from the block through the motion compensator 126, subtracting pixel by pixel. The difference or residual obtained by subtraction of the subtractor 106 is applied to the DCT performer 108.

Meanwhile, the compressed P picture is decoded by the dequantizer 116 and the IDCT converter 118, and the decoded data is applied to the first input terminal of the adder 120. At the same time, each block of the reference frame stored in the frame memory 122 is accessed to predict the current picture, and the accessed block is applied to the second input terminal of the adder 120 through the motion compensator 126. The adder 120 adds the encoded remainder or difference and the data output from the motion compensator 126 to reconstruct the actual image. The reconstructed P picture from the adder 120 is then stored in frame memory 122 to predictively encode / decode the B picture.

On the other hand, the activity calculator 128 calculates the activity in the spatial region of the input image stored in the frame / field memory 104 to set the initial buffer fullness, thereby quantizing steps based on the activity. Predict The rate controller 130 applies the actual quantization interval to the quantizer 110 and the VLC & MUX 112 based on the predicted quantization interval and the fullness of the buffer 114. In this case, the buffer 114 changes the buffer fullness according to the encoded bit amount, thereby controlling the quantization interval as feedback. The inverse quantizer 116 and IDCT converter 118 shown in FIG. 2 may be referred to as local decoders, including a subtractor 106, an adder 120, a frame memory 122, and a motion predictor 124. And motion compensator 126 may be referred to as a motion predictor and compensator.

FIG. 4 is a block diagram of an apparatus for encoding video data having a digital watermark embedded therein according to an embodiment of the present invention. As compared with the apparatus for encoding video data shown in FIG. The watermark information inserter 206 connected between the subtractor 208 and the watermark information remover 222 connected between the IDCT converter 220 and the adder 224 are different from each other.

That is, the encoding apparatus shown in FIG. 4 inserts watermark information only into an I picture used as a reference picture in MPEG-2 video encoding, and for P and B pictures that are motion predicted and compensated based on the I picture, In order to reduce errors in motion prediction and compensation due to the inserted watermark information, the watermark is configured to be removed from the local decoded I picture used as a reference picture in motion prediction.

FIG. 5 is a block diagram according to another embodiment of an apparatus for encoding video data having a digital watermark embedded therein, according to an embodiment of the present invention. Watermark information is used for simplifying a circuit as compared with the apparatus for encoding video data shown in FIG. 4. Only the eliminator is not configured.

That is, the encoding apparatus shown in FIG. 5 inserts watermark information only for an I picture used as a reference picture in MPEG-2 video encoding by the watermark information inserter 205, and predicts motion based on the I picture. For the P and B pictures to be compensated, the experiment shows that there is no significant effect on the image quality even if the watermark is not removed from the local decoded I picture used as the reference picture in the motion prediction. Mark information eliminator is not configured.

6 is a detailed block diagram of an embodiment of the watermark information inserter 206 shown in FIG. 4 and the watermark information inserter 205 shown in FIG. 5, wherein the watermark information inserter is a first DCT. Converter 238, accumulator and divider 240, variance calculator 242, amplifier 244, replication control information generator 246, spectrum spreader 248, pseudo random noise sequence (PN) Generator 250, modulator 252, second DCT converter 254, multiplier 256, adder 258, and IDCT converter 260.

The operation of the watermark information inserter shown in FIG. 6 will be described in conjunction with FIG. 4 for convenience of explanation, and described as copy control information as an example of watermark information.

In FIG. 6, the first DCT converter 238 performs DCT conversion on the I picture data x _i supplied from the frame / field memory 204 shown in FIG. 4 to generate a DCT coefficient, The calculator 240 accumulates the DCT coefficients for the N DCT blocks, calculates the accumulated DCT coefficients, and calculates an average variance in units of chip rates.

Here, in order to detect a watermark inserted without an original image, it is necessary to reconstruct an image into a region in which watermark information is inserted into a relatively strong region of the image. Is defined as the block multiple of the DCT block to hide 8 It consists of eight. Here, the chip rate unit may be referred to as a sub-picture unit.

The variance calculator 242 is currently eight pixels supplied from the first DCT converter 238. If the variance of the DCT block of eight lines is calculated and the variance (complexity) of the calculated DCT block is greater than the average variance of the corresponding chip rate supplied from the accumulator and divider 240, the amplifier 244 amplifies the amplification coefficient (amplication). factor) increases alpha This amplification coefficient may be referred to as visual sensitivity coefficient or embedding strength.

However, if the variance of the current DCT block is less than the average variance of the chip rate, alpha = n, otherwise alpha = m. n is less than m. That is, if the complexity of the current DCT block is greater than the average complexity of the corresponding chip rate, alpha is increased.

In addition, in the present invention, extended spectrum spreading using pseudo-noise sequence is used to prevent degradation of image quality during MPEG encoding due to copy control information inserted in the form of noise and to give robustness to removal of inserted copy control information. ) Is inserted into the copy control information, and an example of the copy control information a _i generated by the copy control information generator 246 is as follows.

Allow one copy

Copy never

No more copy

- Other

In the embodiment of the present invention, the copy control information has four patterns, which can be represented by two bits.

Spectrum spreader 248 spreads and inserts the copy control information bits into the chip rate. Pseudo-noise sequence generator 250 generates a pseudo-noise sequence _pi for modulation. Modulator 252 modulates using the pseudo noise column (p _i) to be supplied to (b _i) the spread spectrum duplication control information supplied from the spreader 248 from the pseudo noise generator column 250. The second DCT converter 254 performs DCT conversion on the modulated copy control information supplied from the modulator 252.

The multiplier 256 multiplies the visual sensitivity coefficient alpha supplied from the amplifier 244 with the DCT coefficient for the replication control information supplied from the second DCT converter 254 to apply the multiplication result to the adder 258. The adder 258 adds the DCT coefficient for the I picture of the source image supplied from the first DCT converter 238 and the multiplication result of the multiplier 256, and applies the addition result to the IDCT converter 260.

IDCT transformer 260 to the noise form the IDCT transform to replication and control information is output to the inserted picture (x _'i), that is, one is the original image I picture of the copying control information of the four patterns for the addition result Is inserted.

Here, an embodiment of a method of inserting copy control information is as follows.

First step: The number of information bits of the inserted copy control information may be allocated as in Equation 1, and the copy control information (a _{j, k} ) according to the allocated number of information bits and the pattern may be represented as in Equation 2.

Here, j = 1, ..., the number of information bits, k = 1, 2, 3, 4, and represents one copy control information pattern among four copy control information patterns.

Step 2: Spread the bits of copy control information into the chip rate. This is represented by the equation (3).

b_i, k = a_j, k

Where j chiprate = i <(j + 1) chiprate, and i is the pixel.

Third Step: As given in Equation 4, the copy control information is modulated by using a pseudo noise string _pi for bipolar modulation.

Fourth step: Generate moving picture data x ' _i of the I picture in which the copy control information is inserted. This is given by Equation 5 below.

x_i '= IDCT (DCT (x_i) + alpha (DCT (b_i, p_i))

Where alpha is the visual sensitivity factor and DCT and IDCT units are 8 pixels It is a block unit of 8 lines. Therefore, one of four patterns of copy control information is inserted into the original picture I picture in the form of noise as watermark information through Equation 5 above.

FIG. 7 is a detailed block diagram of another embodiment of the watermark information inserter 206 shown in FIG. 4 and the watermark inserter 205 shown in FIG. 5, wherein the watermark information inserter is in the form of noise. Inserting a digital watermark of a spread spectrum technique using a pseudo noise string to prevent deterioration of image quality during MPEG encoding due to embedded copyright information and copy control information and to give robustness to removal of inserted copyright information and copy control information. The first DCT converter 233, the block classifier 235, the insertion strength calculator 237, the insertion mask generator 239, the watermark information generator 241, the spectrum spreader 243, and the pseudo noise string generator 245. ), A modulator 247, a second DCT converter 249, a multiplier 251, an adder 253, and an IDCT converter 255.

A description of the operation of the watermark information inserter shown in FIG. 7 will be described with reference to FIG. 5 for convenience. The watermark information includes not only copy control information but also author information.

In FIG. 7, the first DCT converter 233 has 8 pixels for the data I _mn (i, j) of the intra picture supplied from the frame / field memory 203 shown in FIG. After dividing into 8-line blocks, the DCT transform is performed, and the block classifier 235 classifies the DCT block through the first DCT converter 233 using the classification mask shown in FIG. 8 as follows. That is, each DCT block is classified according to energy distribution using a classification mask.

Flat block

Horizontal edge block

Vertical edge block

Diagonal edge block

-Detailed block

The classification mask used for block classification in the present invention is a vertical edge mask shown in Fig. 8A, a horizontal edge mask shown in Fig. 8B, and a diagonal edge shown in Fig. 8C. Mask, the complex block mask shown in Fig. 8D. Here, the mask may be referred to as a window. In addition, 8 If the block image of size 8 is a vertical edge image, AC coefficients among DCT coefficients generated after DCT conversion on the block image are mainly in the horizontal direction as shown in FIG. Like 8 If the block image having the size of 8 is a horizontal edge image, AC coefficients among DCT coefficients generated after DCT conversion on the block image mainly have AC coefficients in the vertical direction as shown in FIG. The classification of the DCT block performed in the block classifier 235 is as follows.

That is, 8 The class is classified as a flat block when the value B _sum obtained by subtracting the DC coefficient from the absolute value of the sum of the DCT coefficients of the eight blocks is smaller than the first threshold Th1. The absolute sum of the result of multiplying each DCT coefficient of the DCT block by the horizontal edge mask shown in FIG. 8B is called H _sum , and each DCT coefficient of the DCT block is shown in FIG. 8A. the absolute sum of the results by multiplying the vertical edge mask multiplying V _sum La, and the DCT block absolute sum of the diagonal result by multiplying the edge mask multiplication shown in (c) of Figure 8 with each DCT coefficient d _sum La When the value of H _SUM , D _SUM , and V _SUM is larger than the second threshold Th2, the class is classified as an edge block having the maximum value among them. If B _sum is not smaller than the first threshold Th1 and the values of H _sum , D _sum , and V _sum are not greater than the second threshold TH2, the class is classified as a complex block.

The insertion strength calculator 237 calculates an insertion strength alpha corresponding to a local property of each DCT block using Equation 7 when the block classification is performed in the block classifier 235.

Insertion strength alpha is adaptively selected according to the total energy of each block. In the case of a block having a relatively small energy, the insertion strength alpha is small. In the case of a block having a relatively large energy, the insertion strength alpha is also large.

The insertion mask generator 239 generates an insertion mask t _mn (i, j) indicating a position at which watermark information is inserted according to the characteristics of the blocks classified by the block classifier 235 using Equation (8).

That is, when the characteristic of the current DCT block is classified as a flat block in the block classifier 235, and the coefficients of the DCT block have a value larger than the B _sum value multiplied by the third threshold Th3, the value of the insertion mask t ( An insertion mask with i, j)) set to " 0 " is generated, i.e., if it is a flat block, no watermark information is inserted. Otherwise, if it is an edge block or a complex block, the value (t (0,0)) of the insertion mask at the DC coefficient position is set to "0", and the value of the insertion mask (t (i, j)) sets the classification mask at 1. If an insertion mask having a subtracted value is generated, that is, a block excluding the flat block, watermark information is inserted at a position excluding a DC position and a position where the classification mask of the block is "1". The main cause of the watermark extraction error is when watermark information is inserted into a relatively large DCT coefficient, and the present invention prevents the watermark information from being inserted into a relatively large DCT coefficient using an embedding mask. Insertion masks are the opposite of block classification masks.

On the other hand, the watermark information generator 241 generates watermark information w _m . Here, the author information and the copy control information inserted by the digital watermark technique of the present invention can be configured as follows.

-Attribution information: N bits,

Replication control information: 2 bits

Therefore, the amount of watermark information to be inserted is M = N + 2 bits. Among them, the replication control information is composed of the following four patterns.

Allow one copy

Copy never

No more copy

- Other

The spreader 243 spreads and inserts the watermark information bits into the chip rate spectrally. Pseudo-noise sequence generator 245 generates a pseudo-noise sequence p _mn (i, j) for modulation. The modulator 247 modulates the spread watermark information w _m supplied from the spectrum spreader 243 using the pseudo noise string p _mn (i, j) supplied from the pseudo noise string generator 245. . That is, the author information and the copy control information spread as described above are modulated by a pseudo noise string known only to the author, and then DCT-converted by the second DCT converter 249.

The multiplier 251 is provided from the insertion intensity alpha output from the insertion strength calculator 237, the insertion mask value t _mn (i, j) supplied from the insertion mask generator 239, and the second DCT converter 249. The DCT coefficient for the supplied watermark information is multiplied to apply the multiplication result to the adder 253.

The adder 253 adds the DCT coefficient for the I picture of the source image supplied from the first DCT converter 233 through the block classifier 235 and the multiplication result of the multiplier 256, and adds the addition result to the IDCT converter 255. ) Is applied. Here, the addition result of the adder 253 is represented as follows.

The IDCT converter 255 performs IDCT conversion on the addition result of the adder 253 and transmits the picture _I'mn (i, j) into which the watermark information is inserted, to the MPEG encoder.

That is, the watermark embedding process according to the present invention concentrates the watermark on a relatively small DCT coefficient, thereby degrading the image quality due to the watermark embedding and watermark using only the watermarked MPEG encoded bitstream without using the original image. This is to reduce the detection error during detection.

Here, the method of embedding watermark information may be performed as follows.

First step: DCT transform the input intra picture data and perform block classification on the DCT domain using Equation 6 above.

Second step; After the block classification, the insertion strength corresponding to the local property of each DCT block is obtained using Equation 7 above, and an insertion mask indicating an insertion position corresponding to the characteristics of the classified block is expressed using Equation 8 above. Occurs.

Third step: The watermark information to be inserted is constructed to spread the watermark information into the chip rate represented by the following expression (10).

Where horizontal_pixel_size is the number of pixels per line, vertical_line_size is the number of lines per frame, and block_size is 8 (pixels) 8 (lines) and M is the number of bits of copyright information and control information to be inserted.

A fourth step; The spread watermark information is modulated using a pseudo noise string and then DCT-converted.

Step 5: The watermarked image is generated by inserting the modulated watermark information corresponding to the characteristics of the block from the insertion strength and the insertion mask on the DCT region.

On the other hand, when digital watermark information in the form of noise is inserted into an I picture of an original image and encoded, and the encoded I picture is locally decoded and used as a reference for predicting motion of P and B pictures, the watermark information is eventually mixed. Errors may occur when predicting motion of P and B pictures.

In order to solve this problem, as shown in FIG. 9, the digital watermark information inserted into the I picture is removed at the local decoding to reduce errors in motion prediction and compensation in the time domain for P and B pictures. .

9 is a detailed block diagram of an embodiment of the watermark information remover 222 shown in FIG. 4, wherein the watermark information remover includes a first DCT converter 262, an accumulator and a divider 264, and distribution. It consists of a calculator 266, an amplifier 268, a spectrum spreader 270, a modulator 272, a second DCT converter 274, a multiplier 276, a subtractor 278, and an IDCT converter 280.

In comparison with the configuration of the watermark information inserter shown in Fig. 9 and the configuration of the watermark information inserter shown in Fig. 6, the watermark information remover instead of the adder 258 of the watermark information inserter is a watermarked I. it is different to that by subtracting the DCT coefficients for the copying control information from the DCT coefficients for the picture (x _'i) a subtracter for extracting a DCT coefficient to the original I-picture (x _i) (278) configuration. For simplicity of the circuit, the duplication control information generator for generating the duplication control information a _i applied to the spectrum spreader 270 is not configured separately, but is shared with the duplication control information generator 246 shown in FIG. The pseudo noise stream generator for generating a pseudo noise stream _pi applied to the modulator 272 is also shared with the pseudo noise train generator 250 shown in FIG.

Although the configuration of the watermark information remover shown in FIG. 9 may be shared with the same blocks as the configuration of the copy control information inserter shown in FIG. 6, it is separately configured as shown in FIG. 9 for the reliability of the system. desirable. Because the watermarked I picture (x ' _i ) input to the first DCT converter 262 shown in FIG. 9 is the DCT converter 210, the quantizer 212, and the inverse quantizer 218 shown in FIG. 4. This is because data loss is unavoidable while passing through the IDCT converter 220, which is not the same as the watermarked I picture output from the watermark information inserter 206 shown in FIG.

FIG. 10 is a detailed block diagram of another embodiment of the watermark information remover 222 shown in FIG. 4, which includes a first DCT converter 257, a block classifier 259, and an insertion strength regulator (FIG. 261, an insertion mask generator 263, a spectrum spreader 265, a modulator 267, a second DCT converter 269, a multiplier 271, a subtractor 273, and an IDCT converter 275.

Similarly, in comparison with the configuration of the watermark information remover shown in FIG. 10 and the configuration of the watermark information inserter shown in FIG. 7, the watermark information remover instead of the adder 253 of the watermark information inserter 205. It is by subtracting the DCT coefficient of the watermark information in the DCT coefficient to the watermark I picture _{(I 'mn (i, j} )) extracts the DCT coefficients for the original I-picture _{(I mn (i, j)} ) The subtractor 273 is different. The watermark generator for generating watermark information (w _m ) applied to the spectrum spreader 265 for simplicity of the circuit is shared with the watermark information generator 241 shown in FIG. A pseudo noise stream generator for generating a pseudo noise stream p _mn (i, j) applied to 267) is also shared with the pseudo noise stream generator 245 shown in FIG.

Fig. 11 is a block diagram of a general decoding apparatus for MPEG-2 moving picture data for better understanding of the present invention. Decoding is performed in the reverse order of the encoding order. In Fig. 11, the buffer 302 temporarily stores an MPEG-2 encoded bitstream. The variable length decoding and demultiplexer (VLD & DEMUX) 304 variable length decodes an encoded bitstream to detect a quantization interval and a motion vector from the variable length decoded data, and the quantization interval and the variable length decoded data are dequantized. 306, and the motion vector is applied to the motion compensator 312.

The inverse quantizer 306 dequantizes the variable-length decoded data supplied from the VLD & DEMUX 304 according to the quantization interval, and the IDCT converter 308 applies the dequantized data supplied from the inverse quantizer 306. For IDCT conversion.

The adder 310 outputs through the buffer 314 as it is, if the decoded image supplied from the IDCT converter 308 is an I picture, and at the same time, decodes the P or B picture based on the I picture. If it is a P or B picture, the motion-compensated signal and the currently decoded P or B picture are added by using the motion vector output from the motion compensator 312 and output through the buffer 314.

That is, when decoding a video, a P picture can be completely reconstructed through motion compensation using a motion vector only when there is a decoded previous I picture or P picture, and a B picture is not only a previous I or P picture but also a future. The motion compensator 312 stores a previous I and P picture and a subsequent I and P picture because the I or P picture can be decoded through motion compensation using a motion vector.

12 is a block diagram of an apparatus for decoding a moving image data having a digital watermark inserted therein according to the present invention, wherein a watermark information detector for detecting watermark information inserted together with moving image data on an encoded bitstream without an original image ( 410 and a watermark information remover 412 for removing the watermark from the decoded I picture used as the reference picture during motion compensation to prevent errors in motion compensation of the P and B pictures due to the watermark information. This point is distinguished from the general MPEG-2 moving picture decoding apparatus shown in FIG.

That is, the watermark information detector 410 detects watermark information from the decoded I picture supplied from the IDCT converter 408 and applies it to the digital recorder to control illegal duplication. The watermark information remover 412 removes the watermark information from the decoded I picture used as the reference picture output from the IDCT converter 408 to prevent errors in motion compensation of the P and B pictures due to the watermark information. To the adder 414. Here, the watermark information remover 412 may be configured similarly to the configuration shown in FIGS. 9 and 10.

That is, if the watermark information remover 412 shown in FIG. 12 has the configuration of the watermark information remover shown in FIG. 9, the watermarked I picture input to the first DCT converter 262 is the IDCT of FIG. 12. It becomes a watermarked I picture output from the converter 408, and the copy control information a _i input to the spectrum spreader 270 is different from the copy control information detected by the watermark information detector 410, 10, the watermarked I picture input to the first DCT converter 257 becomes a watermarked I picture output from the IDCT converter 408 shown in FIG. The watermark information w _m input to the spectrum spreader 265 is different from the watermark information detected by the watermark information detector 410.

Accordingly, since the encoding apparatus encodes by removing the influence of the watermark when encoding the P or B picture, the decoding apparatus also removes the watermark detected during decoding of the I picture to compensate for motion in the time domain with respect to the P and B pictures. Reduce errors

FIG. 13 is a block diagram according to another embodiment of a decoding apparatus for moving picture data having a digital watermark embedded therein, which is distinguished from the general decoding apparatus for MPEG-2 moving picture data shown in FIG. The watermark information detector 415 is further configured to be connected to the rear stage and detect watermark information from the decoded video data.

FIG. 14 is a detailed block diagram of the watermark information detector 412 shown in FIG. 12 and the watermark information detector 415 shown in FIG. 13. The watermark information detector includes a DCT converter 420; A DC coefficient remover 422, an IDCT converter 424, a pseudo noise string generator 426, a demodulator 428, a summation circuit 430, a copy control information pattern generator 432, and a comparator 434.

The operation of the watermark information detector shown in FIG. 14 will be described in conjunction with FIG. 12 for convenience of explanation, and as copy control information as an example of watermark information. In Figure 14, DCT converter 420 is the IDCT converter for watermarked I-picture data (x _'i) supplied from the (408) DCT transform, the DC coefficient remover 422 shown in Figure 12 is a DCT converter DC coefficients among DCT coefficients supplied from 420 are removed, and IDCT transformed by the IDCT converter 424 for the AC coefficients. This is because the watermark information is inserted in the form of noise so that the watermark information can be detected from the AC coefficient among the DCT coefficients.

The demodulator 428 is duplicate combined from the pseudonoise heat generator 426, pseudonoise column (p _i), the data (b _i) a conference to 430 demodulates the chip rate units to demodulate the IDCT data using the generated from Control information is detected. Namely, if the sum result is positive, the copy control information is "1", and if the result is negative, the copy control information is "-1". This pseudo noise stream generator 426 can be shared with the pseudo noise train generator used in the encoder.

The comparator 434 compares the output of the circuit 430 with the copy control information pattern generated by the copy control information pattern generator 432 and if the comparison result is smaller than the threshold value TH, the digital copy recorder detects the detected copy control information. Otherwise, copy control information indicating no copy control information is outputted.

Examples of the detected copy control information are as follows.

Allow one copy

Copy never

No more copy

- Other

No watermark

In the present invention, in order to detect a watermark inserted without an original image, the chip rate of the copy control information in the encoding apparatus is reconstructed into a region where the image has a relatively high correlation. 8 The duplication control information inserted in units of 8 chip rates is detected.

An embodiment of a method of detecting copy control information is as follows.

First step: DCT transform the decoded I picture in units of DCT blocks.

Second step: IDCT data is generated by using only the AC coefficient among the DCT coefficients.

Step 3: Demodulate IDCT data using the same pseudo noise string p _i used in the encoding apparatus to generate demodulated data b _i .

Fourth Step: The demodulated data b _i is summed into units of chip rate. This is represented by the equation (11).

Step 5: The inserted copy control information is restored by using the sum result of the fourth step, which is represented by Equation 12.

Step 6: If the watermark information is copy control information, the inserted copy control information is detected by comparing the restored copy control information bit and the copy control information pattern or information indicating that there is no watermark information is generated. This is given by Equation 13, and the detected copy control information is transmitted to the digital recorder to control illegal copying.

Here, pattern_a _{j, k} is a replication control information pattern. However, if simiarity_test is larger than a threshold, it means no watermark.

15 is a detailed block diagram of another embodiment of the watermark information detector 412 shown in FIG. 12 and the watermark information detector 415 shown in FIG. 13, wherein the watermark information detector includes a DCT converter 417; Block classifier 419, insertion mask extractor 421, multiplier 423, subtractor 425, IDCT converter 427, demodulator 429 and pseudo noise string generator 431, and circuit 433. .

The operation of the watermark information detector shown in FIG. 15 is described in connection with FIG. 13 for convenience of description, and the detected watermark information includes both author information and copy control information. In FIG. 15, the DCT converter 417 performs DCT conversion on the decoded video data supplied from the buffer 413 shown in FIG. In this case, it is assumed that data input to the DCT converter 417 is R _mn (i, j).

The block classifier 419 may be configured in the same manner as the block classifier 235 illustrated in FIG. 7, and the DCT block supplied from the DCT converter 417 may be a flat block, an edge block, or a complex block using Equation 6 above. Classify cognition.

The insertion mask extractor 421 extracts an insertion mask t _mn (i, j) from Equation 8 from blocks classified by the block classifier 419. The multiplier 423 multiplies the coefficients of the DCT block output through the block classifier 419 and the insertion mask value t _mn (i, j) extracted by the insertion mask extractor 421, and the result of the multiplication It can be expressed as Equation 14.

The subtractor 425 subtracts the multiplication result of the multiplier 423 from each coefficient of the DCT block output through the block classifier 419 to remove the DC coefficient of the DCT block and the AC coefficient of the classified block. The result of this subtractor 425 is IDCT-converted by the IDCT converter 427 into integer precision.

The demodulator 429 demodulates from the output E _mn (i, j) of the IDCT converter 427 by using the pseudo noise string p _mn generated by the pseudo noise string generator 431, and the demodulation result b _mn ) can be expressed as Equation 15. This pseudo noise stream generator 426 can be shared with the pseudo noise train generator used in the encoder.

The sum circuit 433 selects only one piece of information of {-1,1} of the block unit output from the demodulator 429, and adds the result by the chip rate unit to take a signium function. . In other words, if the sine of the result of the combining circuit 433 is positive, the detected watermark information is "1", and if it is negative, the detected watermark information is "-1". This may be represented as in Equation 16.

Also, the sum circuit 433 detects the inserted author information and the copy control information by selecting only one of the information of the final {-1,1} on a chip rate basis using Equation 16 above. Since zero may occur during the sum process instead of {-1,1}, Equation 17 is used to compensate. That is, the watermark information is detected by taking the sign of the block having the largest value among the values obtained by adding the information of {-1,1} in each block in the chip rate unit.

In the watermark information detector illustrated in FIG. 15, when the watermark information is copy control information, a copy control information pattern generator and a comparator may be further configured as shown in FIG. 14.

Here, another embodiment of the method for detecting watermark information is as follows.

First step: DCT conversion of the input intra picture data is performed and block classification is performed on the DCT region using Equation (6).

Second step; The insertion mask is extracted using the characteristics of the classified blocks.

Third step; By using the extracted insertion mask, the DC coefficients of the current DCT block and the AC coefficients of the classified block are removed to be IDCT transformed into integer precision.

A fourth step; Demodulate using the pseudo noise string from the result of IDCT conversion.

Fifth step: The watermark information is detected by adding the demodulated data in a chip rate unit.

The present invention can be applied to both block-based video codecs as well as MPEG-2 video compression / restoration apparatus. In addition, both the compressed data and the uncompressed data can insert and detect watermark information using the present invention, and the watermark information can be inserted and detected using the present invention not only for moving image data but also for still image data. .

As described above, the present invention minimizes the deterioration of the image quality of the moving picture compression / restoration apparatus employing the MPEG moving picture coding algorithm, and also prevents illegal duplication using a digital recorder or the like.

In addition, the present invention inserts watermark information such as author information into video data as well as copy control information transmitted to a digital recorder that is difficult to remove watermark information by a user, and also detects inserted watermark information without an original image. There are effects that can be applied and can be applied to a real-time video compression / restore device.

Claims (35)

In copy control apparatus using watermark for copyright protection: A modulation unit for spreading the watermark information on a spectrum in sub-picture units of a predetermined size, and modulating the pseudomarker string to generate modulated watermark information; An insertion unit for inserting the modulated watermark information into an input image as an invisible signal to generate watermarked data; And And an insertion strength adjusting unit for controlling the insertion strength of the modulated watermark information. The apparatus of claim 1, wherein the input image is a blocked input image, and the sub-picture unit is a chip rate unit including a predetermined number of blocks having a predetermined correlation. The modulated water of claim 2, wherein the insertion intensity control unit is further configured to determine that the variance of the blocked input image is greater than the average variance of the corresponding chip rate using the variance of the blocked input image and the average variance of the chip rate unit. A duplication control device, characterized in that for adjusting the intensity of the mark information. The block of claim 2, wherein the insertion strength adjusting unit has a small insertion strength for a block having a relatively small energy according to the total energy of each block of the blocked input image, and a block having a relatively large energy has a large insertion strength. Replication control device, characterized in that for adjusting. The method of claim 1, A demodulation unit for demodulating watermark information from the watermarked data using the pseudo noise string to generate demodulated watermark information; And And a detection unit which adds the demodulated watermark information to the sub-picture unit to detect watermark information. The comparison unit according to claim 5, further comprising comparing the detected watermark information with a reference pattern and outputting the detected watermark information as actual watermark information or generating information indicating that there is no watermark information according to a comparison result. Replication control device further comprising. A DCT converter for applying a DCT coefficient by performing discrete cosine transform (DCT) on an input image, a quantizer for quantizing the DCT coefficient, for applying quantized data, a variable length encoder for variable length encoding the quantized data, and the quantized A moving picture data encoding apparatus comprising a local decoder for locally decoding data, and a motion predictor and a compensator for predicting and compensating motion from the locally decoded data. And a watermark information inserter for inserting watermark information into intra (I) picture data in the input image and applying the watermark information to the DCT performer. The method of claim 7, wherein And a watermark information remover for removing watermark information from the locally decoded I picture data and applying the I picture data from which watermark information has been removed to the motion predictor and the compensator. The method of claim 8, wherein the watermark information inserter, A first DCT converter for DCT transforming the I picture data; A first calculator for accumulating the output of the first DCT converter and calculating an average variance of a chip rate unit consisting of a predetermined number of DCT block units in consideration of correlation; A second calculator for calculating a variance of a current DCT block supplied from the first DCT converter; A first amplifier for adjusting a visual sensitivity coefficient when the variance of the current DCT block is greater than the average variance of the corresponding chip rate; A first spectrum spreader for spreading watermark information within the chip rate; A first modulator for modulating the watermark information output from the first spectrum spreader using a pseudo noise string to generate modulated watermark information; And And embedding means for inserting the modulated watermark information in consideration of the visual sensitivity coefficient into the I picture in the DCT region. The method of claim 9, wherein the insertion means, A second DCT converter for DCT converting the modulated watermark information; A first multiplier for multiplying the DCT coefficient and the visual sensitivity coefficient with respect to watermark information supplied from the second DCT converter; An adder for inserting a DCT coefficient of watermark information multiplied by a visual sensitivity coefficient output from the multiplier to a DCT coefficient for an I picture supplied from the first DCT converter; And And a first IDCT transformer for inverse discrete cosine transforming (IDCT) the output of the adder. The method of claim 10, wherein the watermark information remover, A third DCT converter for DCT transforming the locally decoded I picture data; A third calculator for accumulating the output of the third DCT converter and calculating an average variance of the chip rate unit; A fourth calculator for calculating a variance of the current DCT block supplied from the third DCT converter; A second amplifier amplifying the visual sensitivity coefficient if the variance of the current DCT block supplied from the fourth calculator is greater than the average variance of the corresponding chip rate supplied from the third calculator; A second spectrum spreader for spreading the watermark information within a chip rate; A second modulator for modulating watermark information output from the second spectrum spreader using the pseudo noise string; And And extracting means for extracting the modulated watermark information in consideration of a visual sensitivity coefficient from the locally decoded I picture on a DCT region. The method of claim 11, wherein the extraction means, A fourth DCT converter for DCT converting the modulated watermark information supplied from the second modulator; A second multiplier that multiplies the DCT coefficients for the watermark information supplied from the fourth DCT converter with the visual sensitivity coefficients supplied from the second amplifier; A subtractor for subtracting the DCT coefficient of the watermark information multiplied by the visual sensitivity coefficient supplied from the multiplier from the DCT coefficient for the local decoded I picture data supplied from the third DCT converter; And And a second IDCT converter for IDCT converting the output of the subtractor. The method of claim 8, wherein the watermark information inserter, A first DCT converter for performing DCT conversion on the I picture data in units of DCT blocks; A first block classifier for classifying each DCT block supplied from the first DCT converter into an edge block, a complex block, or a flat block using a classification mask according to its energy distribution; A first insertion strength calculator for adaptively calculating the insertion strength according to the total energy of the sorted blocks; A first insertion mask generator for generating an insertion mask indicating an insertion position of watermark information according to the classified block; A first spectrum spreader for spreading the watermark information into a chip rate consisting of a predetermined number of DCT blocks in consideration of correlation; A first modulator for modulating the watermark information output from the first spectrum spreader using a pseudo noise string to generate modulated watermark information; And And embedding means for inserting the modulated watermark information into the I picture in the DCT region according to the embedding intensity and the embedding mask. The method of claim 13 wherein the insertion means A second DCT converter for DCT converting the modulated watermark information; A first multiplier that multiplies the DCT coefficient, the insertion strength, and the insertion mask with respect to watermark information supplied from the second DCT converter; An adder for inserting a DCT coefficient of watermark information multiplied by an insertion intensity output from the multiplier and an insertion mask to a DCT coefficient for an I picture supplied from the first DCT converter; And And a first IDCT transformer for inverse discrete cosine transforming (IDCT) the output of the adder. The watermark information remover of claim 13, wherein the watermark information remover comprises: A third DCT converter for DCT transforming the locally decoded I picture data; A second block classifier for classifying each DCT block supplied from the third DCT converter according to its energy distribution using an classification mask as an edge block, a complex block, or a flat block; A second insertion strength calculator for adaptively calculating the insertion strength according to the total energy of the sorted blocks; A second insertion mask generator for generating an insertion mask indicating an insertion position of watermark information according to the classified block; A second spectrum spreader that spreads the watermark information within the chip rate; A second modulator configured to generate modulated watermark information by modulating watermark information output from the second spectrum spreader using the pseudo noise string; And And extracting means for extracting, on the DCT region, the modulated watermark information in consideration of the insertion strength and position from the locally decoded I picture. The method of claim 15, wherein the extraction means, A fourth DCT converter for DCT converting the modulated watermark information; A second multiplier that multiplies the DCT coefficient, the insertion strength, and the insertion mask for the watermark information supplied from the fourth DCT converter; A subtractor for subtracting the DCT coefficient of the watermark information multiplied by the insertion intensity supplied from the second multiplier and the insertion mask from the DCT coefficient for the I picture supplied from the third DCT converter; And And a second IDCT converter for IDCT converting the output of the subtractor. 16. The copy control apparatus according to claim 15, wherein the first and second insertion strength calculators have a block having a relatively small energy to have a small insertion strength, and a block having a relatively large energy to have a large insertion strength. A variable length decoder for variable length decoding the received encoded bitstream to apply variable length decoded data and a motion vector, an inverse quantizer for dequantizing the variable length decoded data and applying dequantized data, the inverse An apparatus for decoding moving picture data, comprising: an IDCT converter for performing inverse discrete cosine transform (IDCT) on quantized data to apply decoded data, and a motion compensator for performing motion compensation using the motion vector from the decoded data. And a watermark information detector for detecting watermark information from the decoded intra (I) picture data. 19. The apparatus of claim 18, wherein the device is And a watermark information remover for removing the watermark information inserted into the decoded I picture data and applying the watermark information to the motion compensator. The method of claim 19, wherein the watermark information detector, A first DCT converter for DCT transforming the decoded I picture data; A first IDCT converter for IDCT transforming the DCT coefficients excluding the DC coefficients; A demodulator for demodulating the output of the first IDCT converter using a pseudo noise string; And And a summation circuit for restoring watermark information by adding the demodulated data in a chip rate unit consisting of a predetermined number of DCT blocks in consideration of correlation. The method of claim 20, wherein the watermark information detector, And a comparator for comparing the restored watermark information with a predetermined copy control information pattern to detect copy control information according to a comparison result or to generate information indicating that there is no watermark information. The method of claim 21, wherein the watermark information remover, A second DCT converter for performing DCT conversion on the decoded I picture data; A first calculator for accumulating the output of the second DCT converter and calculating an average variance of the chip rate unit; A second calculator for calculating a variance of a current DCT block supplied from the second DCT converter; An amplifier for amplifying a visual sensitivity coefficient if the variance of the current DCT block is greater than the average variance of the corresponding chip rate; A spectrum spreader for spreading the watermark information within a chip rate; A modulator for modulating watermark information output from the spectrum spreader using a pseudo noise string; A third DCT converter for DCT converting the modulated watermark information supplied from the modulator; A multiplier for multiplying the DCT coefficient and the visual sensitivity coefficient with respect to watermark information supplied from the third DCT converter; A subtractor for subtracting the DCT coefficient of the watermark information multiplied by the visual sensitivity coefficient output from the multiplier from the DCT coefficient for the decoded I picture data supplied from the second DCT converter; And And a second IDCT converter for IDCT transforming the output of the subtractor to apply I picture data from which watermark information has been removed to the motion compensator. The method of claim 19, wherein the watermark information detector, A first DCT converter for DCT transforming the decoded I picture data in units of DCT blocks; A first block classifier for classifying each DCT block supplied from the first DCT converter into an edge block, a complex block, or a flat block using a classification mask according to its energy distribution; An insertion mask extractor for extracting a value of an insertion mask indicating an insertion position of watermark information according to the classified block; A first multiplier multiplying the coefficients of the DCT block output through the block classifier and the insertion mask value; A first subtractor for subtracting a multiplication result of the multiplier from each coefficient of the DCT block output through the block classifier to remove the DC coefficient of the DCT block and the AC coefficient of the classified block; A first IDCT converter for IDCT transforming the result of the first subtractor; A demodulator for demodulating the output of the first IDCT converter using a pseudo noise string; And And a summation circuit for restoring watermark information by adding the demodulated data in a chip rate unit composed of a predetermined number of DCT blocks in consideration of correlation. The method of claim 23, wherein the watermark information detector, And a comparator for comparing the restored watermark information with a predetermined copy control information pattern to detect copy control information or generating information indicating that there is no watermark information according to a comparison result. The method of claim 23, wherein the watermark information remover, A second DCT converter for performing DCT conversion on the decoded I picture data; A second block classifier for classifying each DCT block supplied from the second DCT converter according to its energy distribution using an classification mask as an edge block, a complex block, or a flat block; An insertion strength calculator for adaptively calculating the insertion strength according to the total energy of the sorted blocks; An insertion mask generator for generating an insertion mask indicating an insertion position of watermark information according to the classified block; A spectrum spreader for spreading watermark information within the chip rate; A modulator for modulating watermark information output from the spectrum spreader using the pseudo noise string; A third DCT converter for DCT converting the modulated watermark information; A second multiplier that multiplies the DCT coefficient, the insertion strength, and the insertion mask with respect to watermark information supplied from the third DCT converter; A second subtractor for subtracting the output of the second multiplier from the DCT coefficients for the I picture supplied from the second DCT converter; And And a second IDCT converter for IDCT converting the output of the second subtractor. In the method for encoding and decoding video data by block-based compression encoding scheme: (a) assigning bits and patterns of watermark information to construct watermark information; (b) spreading the watermark information into a chip rate consisting of a predetermined number of blocks in consideration of correlation to generate spread watermark information; (c) inserting the spread watermark information into the video data; And (d) controlling the embedding strength of the modulated watermark information Replication control method comprising a. 27. The method of claim 26, wherein the video data is intra picture data. 27. The copy control method according to claim 26, further comprising modulating the spread watermark information using a pseudo noise string to generate modulated watermark information. 27. The modulated watermark of claim 26, wherein in the step (d), if the variance of the currently blocked input image using the variance of the blocked input image and the average variance of the chip rate unit is greater than the average variance of the corresponding chip rate, the modulated watermark A replication control method, characterized in that the intensity of the information is adjusted. The method of claim 29, wherein in step (d), a block having a relatively small energy according to the total energy of each block of the blocked input image has a small insertion strength, and a block having a relatively large energy has a large insertion strength. Replication control method, characterized in that for adjusting. The method of claim 27, (e) applying DCT coefficients by performing DCT conversion on the intra picture data into which the watermark information is inserted; (f) generating IDCT data by performing IDCT conversion on only AC coefficients among the DCT coefficients; (g) demodulating the IDCT data using a pseudo noise string to generate demodulated data; And and (h) detecting the watermark information by adding the demodulated data in units of the chip rate. The method of claim 31, wherein (i) comparing the detected watermark information with a predetermined copy control information pattern and detecting copy control information or generating information indicating that there is no watermark. The method of claim 27, (e1) DCT converting the intra picture data into which the watermark information is inserted and classifying the block in the DCT region; (f1) extracting an insertion mask using the characteristics of the classified block; (g1) applying IDCT coefficients by performing IDCT transformation on DCT coefficients excluding DC coefficients of the current DCT block and AC coefficients of the classified block by using the extracted insertion mask; (h1) demodulating data from the IDCT coefficients using a pseudo noise string and applying demodulated data; And and (i1) detecting the watermark information by adding the demodulated data in units of the chip rate. 34. The method of claim 33, wherein in step (i1), when the sum of the units in the chip rate unit becomes zero, detecting the watermark information using a sign of a block having the largest absolute value in the chip rate unit is made. A replication control method characterized by the above-mentioned. The method of claim 33, wherein (j1) The duplication control method further comprises detecting the duplication control information by comparing the detected watermark information with a predetermined duplication control information pattern or generating information indicating that there is no watermark.