JP2010016529A - Method, device and program for embedding digital watermark for alteration detection, method, device and program for alteration detection using digital watermark, and recording medium having these programs recorded therein - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a method for embedding a digital watermark for alteration detection, which is capable of detecting alteration of video images by embedding the digital watermark without spoiling the beauty. <P>SOLUTION: An input video image is subjected to YCbCr conversion, down-sampling, DCT, and DCT coefficient quantization in normal JPEG coding process (S31, S32), and redundancy is given to digital watermark data to obtain error detection code data (S33, S34), and this data is converted to encrypted code data (S35) by a prescribed encryption method, and DCT blocks of a selected color component are divided into groups (S36, S37), and the encrypted code data is embedded into the DCT block belonging to each group (S41), and entropy compression of quantization coefficients after the completion of embedding is performed in the same manner as the normal JPEG coding process, and the result is shaped so as to conform to a prescribed JPEG format, and thus JPEG coding is completed (S42, S43). <P>COPYRIGHT: (C)2010,JPO&INPIT

Description

  The present invention relates to a digital video recording apparatus, and more particularly to an alteration detection digital watermark embedding method, apparatus, program, falsification detection method using an electronic watermark, apparatus, program, and a recording medium on which these programs are recorded.

  In recent years, with the widespread use of digital cameras and camera-equipped mobile phones, it has become possible for anyone to easily take digital photo images. The digital video holding method in a digital camera or a camera-equipped mobile phone has a high compression ratio and a good trade-off between image quality, so the JPEG encoding format is the de facto standard. Some models can hold JPEG2000 encoded or uncompressed video, but the most popular is the JPEG encoding method.

  Digital cameras, especially camera-equipped mobile phones, can be easily photographed, so digital images taken with them can be used as evidence of personal property security, evidence of traffic accidents, process proofs at construction sites, and garbage disposal. There is a need to make a proof that was correctly processed in the field. A similar but not digital camera is a drive recorder, and the video taken is used as reference evidence of a traffic accident.

On the other hand, in order to enhance the evidence, a means for adding a digital watermark to a digital video and improving the credibility of the video depending on whether the digital watermark is not broken is proposed (see Non-Patent Document 1). ).
“Other Reference Information JPEG”, searched on June 19, 2008, Internet URL <http: // www. jpo. go. jp / shiryou / s_sonota / hyyoujun_gijutsu / denshi_sukawa / 1 / 1_a_4_1. htm> “Drawing Laboratory: JPEG (Principle 1, 2, 3)”, retrieved on June 19, 2008, Internet URL <http: // webs. lanset. com / crazy17 / jp / lab / jpeg1. htm> Book “Industrial Mathematics for Redoing, Information Communication and Signal Analysis / Cryptography, Error Correcting Code, Integral Transformation”, Masaaki Mitani, CQ Publisher, ISBN 4-7898-3318-6 Specification “128-bit block cipher Camellia algorithm specification version 2.0”, Kazuo Aoki et al., Nippon Telegraph and Telephone Corporation, Mitsubishi Electric Corporation, September 26, 2001 Specification, “Announcing the ADVACED ENCRYPTION STANDARD (AES)”, Federal Information Processing Standards Publication 197, November 26, 2001

  However, because digital video has the property of being easily duplicated and edited, it is difficult to prove whether the video has been tampered with. There was a problem of not getting.

  Further, the method described in Non-Patent Document 1 has a problem that image quality deterioration due to watermark embedding is large because watermark data having a large value (> 1) is embedded immediately before the zero-run portion of the quantization coefficient. Large deterioration in image quality is undesirable because it reduces the authenticity of evidence photographs.

  The present invention solves the above-mentioned problems, and its purpose is to embed a digital watermark for falsification detection without damaging the aesthetics, and to detect a falsification of a video, It is an object to provide an apparatus, a program, and a falsification detection method using an electronic watermark, an apparatus, a program, and a recording medium on which the program is recorded.

  The present invention is closely related to ordinary JPEG encoding and JPEG decoding techniques (see Non-Patent Document 2).

  A normal JPEG encoding process is executed according to the flowchart of FIG. First, a digital video is input (step S11), and the image is converted to YCbCr (luminance scalar (Y) / hue vector (CbCr)) (step S12).

  In step S13, Cb and Cr information is reduced by down-sampling processing, and then DCT (Discrete Cosine Transform) processing is performed to obtain DCT coefficients (step S14).

  Next, the DCT coefficient is quantized (step S15), entropy compression is performed (step S16), and shaping is performed so as to conform to a predetermined JPEG format, for example, JFIF (JPEG File Interchange Format) (step S17).

  The normal JPEG decoding process is executed according to the flowchart of FIG. First, JPEG-encoded digital video data is input (step S21), and the JFIF format is converted into a data string (step S22).

  Next, entropy decoding is performed in step S23, inverse quantization (step S24) and inverse DCT (step S25) are performed, and color space conversion is performed if necessary (step S26).

  In the method of embedding a digital watermark in a video according to the present invention, the video is divided into blocks of 8 × 8 sampling pixels (referred to as DCT blocks), and 1 bit is assigned to each block, and the digital watermark data is encoded with error detection encoding and encryption. It is basic to embed bits.

  The bit of the digital watermark data does not correspond directly to the DCT block but is made to correspond after error detection coding is performed once to provide redundancy. By using an error detection code, it is possible to prevent obtaining the same result as if it was accidentally embedded even if nothing was embedded, and to know whether there was an error (falsification) without knowing the original digital watermark data at the time of decoding Two benefits are obtained.

Since the total number of DCT blocks included in a video is generally large (4800 luminance components of VGA video), DCT blocks are divided into groups and error detection coding is performed independently in one group. As a result, if there is an error, the group including the error can be specified, and the merit of knowing the location of the error (falsification) can be obtained. In the modulation method of the quantization coefficient in the DCT block, the absolute value of the quantization coefficient is obtained. And a reference value selected by the random number generation means from the quantization coefficients excluding the maximum value. The even / odd difference from the reference value is adjusted by adding −K, 0, or K (K is an arbitrary positive / odd number) to the maximum absolute value of the quantization coefficient. The advantage of this method is that only a small variation is given to a coefficient having a large value, so that the change of the quantization coefficient is small, and the aesthetic appearance after embedding is hardly affected. Although there are a plurality of methods for modulating the quantization coefficient in the DCT block, including the one in Reference Document 1, the change in the quantization coefficient due to the modulation is large, which is aesthetically disadvantageous.

  The falsification detection digital watermark embedding method according to claim 1 of the present invention is a system for embedding a falsification detection digital watermark simultaneously with JPEG encoding of a video, and a step of inputting video by JPEG encoding. Means for performing a YCbCr conversion process, a downsampling process, a DCT process and a DCT coefficient quantization process in a normal JPEG encoding process on the input video; and a digital watermark data input means A step of inputting, a step of digital watermark data error detection encoding means, adding the redundancy to the input digital watermark data by a predetermined error detection encoding method to obtain error detection code data, and encryption means Is a code that has already been encrypted by a predetermined encryption method. A color component selecting unit selecting a color component to embed the encoded code data; and a DCT block group dividing unit encoding the DCT block of the selected color component into encoded code data The number of bits is divided into groups equal to the number of bits, and the remaining DCT blocks are grouped into only one group, and the maximum value selecting means includes the quantization coefficients of the DCT blocks belonging to the divided groups. The step of selecting a value having the maximum absolute value from the reference value, and the reference value selecting means 1 based on the random number value obtained from the predetermined random number generating means, from the quantization coefficient excluding the one having the maximum absolute value of the quantization coefficient If the encrypted code data embedding means is a DCT block that embeds bit 0 of the encrypted code data If the DCT block embeds bit 1 of the encoded code data so that the difference between the maximum value of the quantization coefficient of the DCT block and the reference value is an even number, the maximum value of the quantization coefficient of the DCT block and the Encrypted code data embedding step of adding any of −K, 0, K (K is an arbitrary positive / odd number) to the maximum value of the quantized coefficient absolute value so that the difference between the reference values becomes an odd number, and the remainder A remainder group data embedding step for embedding a predetermined pattern of 0 and 1 in the DCT block of the group by the same method as the encrypted code data embedding step, and a JPEG encoding means, Entropy compression is performed in the same way as the normal JPEG encoding process, and the JPEG encoding is completed by shaping it so that it conforms to the predetermined JPEG format. And a step of finishing.

  Further, the falsification detecting digital watermark embedding method according to claim 2 is characterized in that, in claim 1, performing a YCbCr conversion process, a downsampling process, a DCT process and a DCT coefficient quantization process in the normal JPEG encoding process; Between the step of inputting the digital watermark data to be embedded, the step of JPEG encoding means compressing the quantized coefficient so as to conform to a predetermined JPEG format, and the data shaping / entropy decoding means include the JPEG And performing a data shaping and entropy decoding on the format to obtain a DCT coefficient value.

  According to a third aspect of the present invention, there is provided a method for detecting falsification using a digital watermark from a JPEG-encoded image in which the falsification detection digital watermark is embedded by the falsification detection digital watermark embedding method according to claim 1 or 2. In the system for detecting whether the video has been falsified by detecting a digital watermark, a step in which video input means inputs the JPEG encoded video, and a JPEG decoding means applies to the input JPEG encoded video. A step of performing data shaping processing and entropy decoding processing in a normal JPEG decoding step, a step of selecting a color component in which the error detection code data is embedded, and a step of dividing a DCT block group The DCT block of the selected color component is the same as the number of bits of the encrypted encoded data. A step of dividing the number into groups including DCT blocks, a step of selecting a maximum value from a quantization coefficient of a DCT block belonging to each of the divided groups, and a reference value selection; Means for selecting one reference value from the quantized coefficients excluding the one having the maximum absolute value of the quantized coefficient based on the random number value obtained from the predetermined random number generating means; and the difference between the maximum value and the reference value Means for obtaining bit data from the step of outputting 0 if the difference between the maximum value and the reference value is an even number, and 1 if the difference is an odd number; The step of performing the corresponding encryption / decryption as the input bit of the error, and the error detection / decryption means performs the predetermined error detection / decryption on the encrypted / decrypted data, and if there is an error and there is an error, A group including the generated DCT block and embedded digital watermark data if there is no error, and JPEG decoding means performs a normal JPEG decoding process on the input JPEG encoded video. And a step of displaying the presence / absence of the error and the position of the DCT block belonging to the group in which the error occurs prominently together with the decoded video.

  The falsification detection digital watermark embedding device according to claim 4 is a system for embedding a falsification detection digital watermark at the same time as JPEG encoding of a video, and a video input means for inputting video, and the input video, JPEG encoding means for performing YCbCr conversion processing, downsampling processing, DCT processing and DCT coefficient quantization processing in a normal JPEG encoding process, electronic watermark data input means for inputting embedded watermark data, and the input electronic Digital watermark data error detection coding means for providing watermark data with redundancy by using a predetermined error detection coding method to make the data error detection code data, and the error detection code data encrypted by a predetermined encryption method And encryption means for embedding the encrypted code data. And a DCT block that divides the DCT block of the selected color component into groups of the same number as the number of bits of the encrypted code data, and the remaining DCT blocks are only one group. Based on a random number value obtained from a group dividing means, a maximum value selecting means for selecting a maximum absolute value among the quantized coefficients of DCT blocks belonging to each of the divided groups, and a predetermined random number generating means, DCT block in which reference value selecting means for selecting one reference value from quantized coefficients excluding the one with the maximum absolute value of the quantized coefficients, and the encoded code data embedding means embed bit 0 of the encrypted code data If so, the bit 1 of the encoded code data is embedded so that the difference between the maximum value of the quantization coefficient of the DCT block and the reference value is an even number. In the case of a DCT block to be embedded, the maximum value of the quantization coefficient absolute value is set to −K, 0, K (K is an arbitrary value so that the difference between the maximum value of the quantization coefficient of the DCT block and the reference value becomes an odd number. And a coded code data embedding unit that embeds a predetermined pattern of 0 and 1 in the DCT block of the remainder group, and the quantization coefficient after the embedding is completed by ordinary JPEG coding It is characterized by comprising entropy compression means for compressing similarly to the entropy compression processing in the process, and shaping means for shaping the compressed data so as to conform to a predetermined JPEG format.

  The falsification detecting digital watermark embedding apparatus according to claim 5 is a system for embedding a falsification detecting digital watermark at the same time as JPEG encoding of a video, and a video input means for inputting the video, and the input video is converted into a normal JPEG code. JPEG encoding means for encoding in the encoding step, data shaping / entropy decoding means for obtaining DCT coefficient values by performing data shaping and entropy decoding on the JPEG format encoded by the JPEG encoding means, and embedded electronic Digital watermark data input means for inputting watermark data, and digital watermark data error detection encoding means for providing redundancy to the input digital watermark data by using a predetermined error detection encoding method to obtain error detection code data The error detection code data is encrypted by a predetermined encryption method. Encrypting means for making encoded code data; color component selecting means for selecting a color component in which the encrypted code data is embedded; and the same number of bits as the number of bits of encrypted code data for the DCT block of the selected color component The DCT block group dividing means for dividing the remaining DCT blocks into one group and the quantization coefficient of the DCT block belonging to each of the divided groups has a maximum absolute value. A maximum value selection means for selecting, a reference value selection means for selecting one reference value from the quantized coefficients excluding the one with the maximum absolute value of the quantized coefficients based on a random number value obtained from a predetermined random number generating means; If the encrypted code data embedding means is a DCT block in which bit 0 of the encrypted code data is embedded, the quantization coefficient of the DCT block is If the DCT block embeds bit 1 of the encrypted code data so that the difference between the large value and the reference value is an even number, the difference between the maximum value of the quantization coefficient of the DCT block and the reference value is an odd number. Thus, any one of −K, 0, K (K is an arbitrary positive / odd number) is added to the maximum value of the quantized coefficient absolute value, and a predetermined pattern of 0 and 1 is embedded in the DCT block of the remainder group. Encrypted code data embedding means, entropy compression means for compressing the quantized coefficients after completion of embedding in the same manner as the entropy compression processing in the normal JPEG encoding process, and the compressed data conforming to a predetermined JPEG format It is characterized by comprising shaping means for shaping.

  According to a sixth aspect of the present invention, there is provided a falsification detection apparatus using a digital watermark from a JPEG-encoded image in which the falsification detection digital watermark is embedded by the falsification detection digital watermark embedding apparatus according to the fourth or fifth aspect. In a system for detecting whether or not a video has been tampered with by detecting a digital watermark, a video input means for inputting the JPEG encoded video, and a normal JPEG decoding process for the input JPEG encoded video Means for performing data shaping processing and entropy decoding processing; color component selection means for selecting a color component in which the error detection code data is embedded; and a DCT block of the selected color component is converted into a bit of the encoded data. DCT block group dividing means for dividing a DCT block into groups each including the same number of DCT blocks, A maximum value selecting means for selecting a maximum absolute value from among the quantized coefficients of DCT blocks belonging to each divided group; and an absolute value of the quantized coefficient based on a random number value obtained from a predetermined random number generating means. A reference value selection means for selecting one reference value from the quantization coefficient excluding the one with the largest value, and 0 if the difference between the maximum value and the reference value is an even number and 1 if the difference is an odd number. Means for obtaining bit data from the difference; encryption / decryption means for performing encryption / decryption using the output 0 or 1 as an input bit for predetermined encryption / decryption; and a predetermined error for the encrypted / decrypted data Error detection and decoding means for performing detection and decoding to obtain the presence or absence of an error, a group including a DCT block in which an error has occurred if there is an error, and embedded digital watermark data if there is no error, and the input JPEG decoding means for converting the JPEG-encoded video into a normal JPEG decoding process and displaying the presence / absence of the error and the position of the DCT block belonging to the group in which the error occurred together with the decoded video And a display means.

  A falsification detection digital watermark embedding program according to a seventh aspect is a falsification detection digital watermark embedding program for causing a computer to execute the steps according to the first or second aspect.

  A falsification detection program using a digital watermark according to claim 8 is a falsification detection program using a digital watermark for causing a computer to execute each step according to claim 3.

  A recording medium according to claim 9 is a computer-readable recording medium in which the falsification detecting digital watermark embedding program according to claim 7 is recorded.

  A recording medium according to a tenth aspect is a computer-readable recording medium in which a falsification detection program using the digital watermark according to the eighth aspect is recorded.

(1) In the present invention, when embedding a digital watermark in a quantization coefficient of a JPEG video, a coefficient having the maximum value is selected, and a given variation is K at most, so that it is possible to embed without impairing aesthetics. Also, if the error detection code detects that the watermark is broken, it can be seen that the video group (= video range) containing the broken watermark data has been tampered with. It is possible to provide a means of knowing whether or not it has been done.

Furthermore, the present invention encrypts embedded data, and the embedding algorithm also uses a random number to determine the position of the embedding coefficient. Therefore, unless the knowledge of the encryption key and the seed value of the random number is known, the algorithm of the present invention is used. Even those who are familiar with it are difficult to forge. For this reason, if the encryption key and the seed of the random number are kept secret, it is possible to prevent an attack that makes a false image a genuine image.
(2) Also, since an error detection code is used, it is possible to prevent the same result from being accidentally embedded even if nothing is embedded, and an error (falsification) without knowing the original digital watermark data at the time of decoding. ) Can be obtained.

  Hereinafter, embodiments of the present invention will be described with reference to the drawings, but the present invention is not limited to the following embodiments. First, an embodiment of a digital watermark embedding method for falsification detection and its apparatus according to the present invention will be described with reference to FIGS.

  FIG. 2 shows an example of the configuration of a JPEG encoding apparatus with watermark embedding to which the present invention is applied. In FIG. 2, 31 is a video input means for inputting a digital video to be embedded.

  Reference numeral 32 denotes JPEG encoding means, which performs YCbCr conversion processing, downsampling processing, DCT processing, and DCT coefficient quantization processing in the normal JPEG encoding process on the input video.

  Reference numeral 33 denotes digital watermark data input means for inputting digital watermark data to be embedded in the input video.

  Reference numeral 34 denotes digital watermark data error detection encoding means for converting the input digital watermark data into error detection code data by adding redundancy to the data by a predetermined error detection encoding method.

  Reference numeral 35 denotes encryption means for converting the error detection code data into code data already encrypted by a predetermined encryption method.

  Reference numeral 36 denotes color component selection means for selecting a color component for embedding the encrypted code data.

  Reference numeral 37 denotes DCT block group dividing means for dividing the DCT block of the selected color component into groups of the same number as the number of bits of the encoded code data, and setting the remaining DCT blocks alone as one group.

  Reference numeral 38 denotes maximum value selection means for selecting a value having the maximum absolute value from among the quantization coefficients of the DCT blocks belonging to each of the divided groups.

  Reference numeral 39 is a reference value selection means for selecting one reference value from the quantized coefficients excluding the one with the maximum absolute value of the quantized coefficients based on the random number value obtained from the predetermined random number generating means 40.

  41 is a DCT block in which bit 0 of the encrypted code data is embedded, bit 1 of the encrypted code data is set so that the difference between the maximum value of the quantization coefficient of the DCT block and the reference value is an even number. In the case of a DCT block to be embedded, the maximum value of the quantization coefficient absolute value is set to −K, 0, K (K is an arbitrary value so that the difference between the maximum value of the quantization coefficient of the DCT block and the reference value becomes an odd number. The coded code data embedding means embeds a predetermined pattern of 0 and 1 in the DCT block of the remainder group.

  Reference numeral 42 denotes entropy compression means for compressing the quantized coefficients after completion of the embedding in the same manner as the entropy compression process in the normal JPEG encoding process.

  Reference numeral 43 denotes JFIF format shaping means for shaping the compressed data so as to conform to a predetermined JPEG format.

  The functions of the means 31 to 43 shown in FIG. 2 are achieved by, for example, a computer.

  Next, the processing procedure performed in the apparatus of FIG. 2 will be described with reference to the flowchart of FIG. First, a digital video to be embedded is obtained by a video input step (step S31).

  Next, in step S32, the same processing as that in the normal JPEG encoding process (steps S12 to S15 in FIG. 1A) is performed, and YCbCr conversion, downsampling, and DCT are performed until DCT coefficient quantization. After quantization, 8 × 8 = 64 quantization coefficients exist in each DCT block. FIG. 4 shows an example of the quantization coefficient. A quantization coefficient exists for each color component of Y, Cb, and Cr. Since the Cb and Cr (color) components are generally sampled more coarsely than the Y (luminance) component during the downsampling, the number of DCT blocks is the largest for the Y component. Regardless of Y, Cb, or Cr, the DCT block contains 8 × 8 quantization coefficients.

  Next, the digital watermark data obtained in the digital watermark data input step (step S33) is processed. For example, assume that the digital watermark data is 448 bits. This 448 bits can be arbitrarily determined, and can be arbitrarily selected within a range in which the number of bits resulting from redundancy by error detection coding described later does not exceed the number of DCT blocks in one divided group described later.

  Next, in the watermark data error detection encoding step (step S34), for example, an even-numbered row parity code is selected as the error detection code. 448 bits are arranged in 7 columns and 64 rows, and one even parity bit is added to each row. The even parity bit means that the value of the parity bit is set to 1 or 0 so that the number of 1 in 8 data bits of each row including the parity bit is an even number. As a result of adding even parity bits to each row, 64 parity bits are added to the data subjected to error detection coding, and the total becomes 448 + 64 = 512 bits. The error detection code is not limited to the extended even-numbered row parity, and the essence of the invention does not change even if other than this is used.

  In the next encryption step (step S35), the 512-bit data subjected to error detection coding is encrypted to generate 512-bit encrypted code data. For example, the Camellia cipher (see Non-Patent Document 3) is used as the cipher. Camellia is a 128-bit block cipher, and converts a sequence of 128-bit data into another sequence of 128-bit data (encoded encoded data). When converting, an encryption key of 128/192/256 bits is required. In order to decrypt the encoded data and obtain the original data code, a camellia decryption process is performed using the same encryption key as that used for encryption. Since camellia is a 128-bit block cipher, 512/128 = 4 processes are required to encrypt a 512-bit error detection code. The encryption method is not limited to camelia, and the essence of the present invention does not change even if other methods are used.

  In the next color component selection step (step S36), the color component in which the error detection code data generated from the watermark data is embedded is selected from Y, Cb, and Cr. Each of them can be selected individually, any two can be selected, or all three can be selected. However, if two or more are selected, the number of DCT blocks that can be used increases, but the grouping of DCT blocks becomes complicated. Therefore, in this example, only Y is selected to simplify the explanation. By selecting Y, there is an advantage that black and white images can be handled. Another effective method is to select only Cb. In this case, there is an advantage that human vision is not noticeable because it is difficult to perceive the blue / yellow change.

  Next, group division of the DCT block in the video is performed (step S37). The number of 8 × 8 pixel DCT blocks is calculated from the video size and the selected color component. For example, if the Y component is selected for a video of VGA size of 640 × 480 pixels, there are 640 × 480 / (8 × 8) = 4800 DCT blocks. These DCT blocks are divided into groups each having the number of encrypted encoded data bits (512 in the above example). In this example, 4800 DCT blocks are divided into 512 groups. Since 4800 is not divisible by 512 and the quotient is 9 and the remainder is 192, 9 groups are formed and 192 DCT blocks remain. A group in which only the remaining DCT blocks are collected is also defined as one group.

  It is assumed that the shape of one group is counted in the raster scan format from the upper left with respect to the original image, and 512 are counted. The next group will continue to count 512 in raster scan format. The last remaining DCT block is one group. An example of block division is shown in FIG. In this example, the case where only the Y component is selected is shown, but the group shape is similarly determined when only the Cb or Cr component is selected. When using a plurality of color components, it is assumed that the DCT block of each color component is continuous, such as Cb after Y and Cr after Cb, and 512 groups are selected. Finally, the remainder One minute is a group.

  In this example, the DCT blocks included in one group are adjacent to each other. However, as long as the number of included DCT blocks is the same, the DCT block arrangement in the same group does not need to be adjacent, and various groupings are made. There can be methods. For example, there is a method of dividing and dividing as shown in FIG. However, the method of division may be arbitrary, but it is easier to specify the location when tampering occurs if the blocks are divided as much as possible.

  Next, the encoded code data is embedded in the DCT block included in each group (step S41).

  Here, there is one exception process, and the encoded code data is embedded in a group composed of 512 DCTs, but the first group, ie, the group consisting of remainders with less than 512 DCTs (remainder group), The same encrypted code data as that of the first group is embedded by the number of the remaining DCTs. Data from the first of the first group data to the DCT number of the remainder group is embedded from the first to the last of the remainder group data. Since the number of DCTs in the remaining group is less than 512, encrypted code data that cannot be embedded is generated but is ignored and not embedded.

  Step S41 of FIG. 3 (embedding process of encrypted code data in the DCT block) is processed according to the flowchart of FIG.

  That is, step S52 for selecting the maximum value M1 of the quantized coefficient, step S53 for selecting the reference value M2, and step S54 for adjusting the maximum value M1 with embedded bits are completed for all DCT blocks. This is repeatedly executed until it is performed (step S51).

  The quantization coefficients are arranged in an 8 × 8 square as shown in FIG. 4, and the upper left is a direct current component and the higher the lower right, the higher the frequency component. In FIG. 7, in the absolute value maximum value selection step (S52) of the quantization coefficient, the absolute value maximum value is found out of 64 quantization coefficients. If there are multiple absolute maximum values, they are determined as follows. A flowchart of this algorithm is shown in FIG.

  When there are two or more maximum values, the first value that appears in the reverse zigzag scan (reverse order of normal JPEG zigzag scan) indicated by the arrow in FIG. 9 is set as the maximum value.

  In FIG. 8, first, the maximum value M1 is set to 0 in step S61, then the advance coefficient A is obtained by reverse zigzag scanning (step S62), and it is determined whether or not | A |> M1 (step S63). If A |> M1, M1 = | A | is set (step S64), and the above steps S62 to S64 are repeatedly executed until all the quantized coefficients are scanned (step S65).

  If there is an equivalent value, the reason why the first appearing in the reverse zigzag scan is adopted is to bring the quantization coefficient used for watermark embedding to the high frequency component side as much as possible and make it less noticeable. The maximum value selected as described above is M1.

  In the reference value selection step (step S53 in FIG. 7), a random number generator that outputs an integer β of 1 ≦ β ≦ 63 is used, and 63 of the 64 quantization coefficients in one DCT block, excluding the maximum value M1, are used. The βth coefficient from the inside is selected as the reference value M2. When determining the β-th coefficient, as shown in FIG. 10, it is counted from the upper left in the raster scan format, and the maximum value M1 is skipped and counted. The random number generator uses a method of obtaining the same random number sequence, for example, a linear congruence method if the seeds of random numbers (parameter initial values) are the same. The contents of the linear congruential method will be described in detail in “5. The random number generation method is not limited to the linear congruential method, and the essence of the present invention does not change even if another method is used in which the random number sequence to be output is determined deterministically by determining the seed of the random number. The value of the obtained random number sequence is used in the order of raster scanning the DCT block. That is, the obtained first random number is used in the DCT block in the upper left corner, and the second random number obtained next is used in the DCT block adjacent to the right of the previous DCT block.

  The embedding policy in the step of adjusting M1 by the embedding bit (step S54 in FIG. 7) is that data 0 (bit 1 of encrypted code data) is embedded so that M1−M2 = odd number. When embedding bit 0) of encrypted code data, M1 is adjusted so that M1-M2 = even. The adjustment is selected from three choices: subtracting arbitrary positive / odd numbers, adding arbitrary positive / odd numbers, or nothing. It is better to use a smaller value for the arbitrary positive / odd number.

  The procedure of this adjustment method is shown in FIG. In FIG. 11, for example, when it is desired to embed data 1, if M1 = 15 and M2 = 14, M1-M2 = 1 and already odd, so nothing is done (steps S71 and S72).

  Further, when M1 = 15 and M2 = 15, M1−M2 = 0 is an even number, and therefore M1 is examined for positive / negative (steps S71, S73, S74). (Step S75) M1 = 15 + 1 = 16 (Thus, M1-M2 becomes an odd number).

  When it is desired to embed data 0, if M1 = −18 and M2 = 11, M1−M2 = −29, which is an odd number, and therefore M1 is examined for positive / negative (steps S71, S72, S74) and is negative. For example, the positive / odd number 1 is subtracted (step S76), and M1 = −18−1 = −19 (this makes M1−M2 an even number).

  If M1 = −18 and M2 = 10, M1−M2 = −28, which is an even number, nothing is done (steps S71 and S73).

  As described above, the bits of the encoded data corresponding to all the DCT blocks in one group are embedded. However, as described above, the same encrypted code data as in the first group is embedded in the remaining groups.

  When embedding of the encoded code data bits is completed for all groups in step S41 in FIG. 3, the entropy compression step is performed in the same manner as in the normal JPEG encoding process (steps S16 and S17 in FIG. 1A). A data shaping step (step S43) is executed so as to conform to (step s42) and a predetermined JFIF format and output as a JPEG video.

  Next, an embodiment of a tampering detection method and detection apparatus using a digital watermark according to the present invention will be described with reference to FIGS.

FIG. 12 shows the configuration of a JPEG decoding apparatus with watermark detection to which the present invention is applied. In FIG. 12, 81 is a JPEG encoded video input means for inputting the JPEG encoded video. 82 is a data shaping process and entropy decoding in the normal JPEG decoding process for the input JPEG encoded video. Data shaping / entropy decoding means for performing processing.

  Reference numeral 83 denotes color component selection means for selecting a color component in which the error detection code data is embedded.

  Reference numeral 84 denotes DCT block group dividing means for dividing the DCT block of the selected color component into a group including DCT blocks by the same number as the number of bits of the encrypted encoded data.

  Reference numeral 85 denotes maximum value selection means for selecting a value having the maximum absolute value from among the quantization coefficients of the DCT blocks belonging to each of the divided groups.

  Reference numeral 86 denotes reference value selection means for selecting one reference value from the quantized coefficients excluding those having the maximum absolute value of the quantized coefficients based on the random number value obtained from the predetermined random number generating means 87.

  Reference numeral 88 denotes a means for obtaining bit data from the difference between the maximum value and the reference value for outputting 0 if the difference between the maximum value and the reference value is an even number and 1 for the odd number.

  Reference numeral 89 denotes encryption / decryption means for performing association encryption / decryption using the output 0 or 1 as an input bit for predetermined encryption / decryption.

  90 performs predetermined error detection / decryption on the encrypted / decrypted data, and if there is an error, if there is an error, a group including a DCT block in which an error has occurred, and if there is no error, an embedded electronic Error detection decoding means for obtaining watermark data.

  Reference numeral 91 denotes a JPEG decoding unit that performs a normal JPEG decoding process on the input JPEG encoded video to generate a video.

  Reference numeral 92 denotes error occurrence location superimposing display means (display means of the present invention) that displays the presence / absence of the error and the position of the DCT block belonging to the group in which the error occurred together with the decoded video.

  The functions of the means 81 to 92 shown in FIG. 12 are achieved by, for example, a computer.

  Next, a processing procedure performed by the apparatus of FIG. 12 will be described with reference to a flowchart of FIG. Make sure that the detection side knows the same information about the block shape, number of blocks, error detection encoding method and encryption method, its encryption key, random number generation method, and seed of random number generation used when embedding in advance. .

  The digital video data input in the digital video data input step (step S81) is subjected to the data shaping and entropy decoding steps (steps S22 and S23 in FIG. 1B) which are normal JPEG decoding steps. Conversion to a quantized coefficient is performed (step S82).

  Next, the same color component selection as that used at the time of embedding is performed (step S83), and then the same DCT block grouping as that used at the time of embedding (see FIG. 5) is performed (step S84).

  In step S88, the encrypted code data 1 or 0 is detected in the following procedure for each DCT block in each group. The procedure will be described with reference to the flowchart of FIG.

  In FIG. 14, a process of selecting the maximum absolute value M1 (step S102), a process of selecting the reference value M2 (step S103), and a process of detecting whether the bit of the encrypted code data is 0 or 1 ( Steps S104 to S106) are repeatedly executed until the processing is completed for all DCT blocks (step S101).

  First, in step S102, the 64 quantized coefficient values are examined to find the absolute maximum value M1. If there are multiple maximum values, they are determined as follows. This is exactly the same as in embedding (executes the process of FIG. 8).

  That is, when there are two or more maximum values, the first value that appears in the reverse zigzag scan (reverse order of normal JPEG zigzag scan) as shown in FIG. 9 is set as the maximum value. Next, in the reference value selection step (step S103), a random number generator that outputs an integer β satisfying 1 ≦ β ≦ 63 is used, and 63 of the 64 quantization coefficients in one DCT block are excluded from the maximum value M1. The βth coefficient from the inside is selected as the reference value M2. When the β-th coefficient is determined, it is counted in the raster scan format from the upper left as in FIG. 10, and M1 is skipped and counted (FIG. 12). The random number generator uses, for example, a linear congruential method. The random number seed (parameter initial value) added to the random number generator uses the same value as that used for embedding. Then, the value of β is the same as that at the time of embedding, and the position of M2 is the same as at the time of embedding. The random number generation method is not limited to the linear congruential method, and the essence of the present invention does not change even if another method is used in which the random number value to be output is determined deterministically by determining the seed of the random number.

  The maximum absolute value selected as described above is M1, and the reference value is M2. If M1-M2 is an even number, it can be understood that encrypted code data 0 is embedded, and if M1-M2 is an odd number, encrypted code data 1 is embedded (steps S104 to S106).

  In this way, bit (1 or 0) is detected from all the DCT blocks in one group (in this example, 512 bits in total), and the encryption is decrypted (step S89 in FIG. 13). At the time of decryption, the same encryption key as at the time of embedding is used. An error detection decoding process is performed in the error detection decoding step (step S90) on the 512-bit decoded data code obtained as described above. As a result, the presence / absence of an error is obtained. If there is no error, 448-bit embedded watermark data is obtained, and if there is an error, information indicating which group includes the error is obtained. However, the remaining groups are not subjected to encryption decryption or error detection decryption, but are merely checked whether the bit groups are the same as the first group.

  Subsequently, in step S91, the quantization coefficient of each DCT block is dequantized and returned to the DCT coefficient similarly to the normal JPEG decoding procedure (steps S24 to S26 in FIG. 1B), and inverse DCT is performed on it. Further, color space conversion is performed to obtain an image.

  Using the presence / absence of error, error group, and embedded watermark data obtained in the error detection / decoding step (step S90), the presence / absence of alteration and the location of alteration are known as follows.

  That is, if no error is detected from all the groups (if the remaining groups have the same data as the first group), it is assumed that the video has not been tampered with after watermark embedding (FIG. 15).

  Also, if the video in the group in which the error is detected is different from the original, that is, it has been tampered with, there is a possibility that it has been tampered somewhere in the area occupied by the DCT blocks constituting the group (FIG. 16). ).

  When expressing an error part, that is, a falsified part, as shown in FIG. 16, if a mark (blacked part in FIG. 16) indicating the position of the falsified group is superimposed on the JPEG-decoded video, it can be seen where the falsified part is. Cheap.

(Example)
In the present invention, there can be several configurations other than the configuration described in the description of the embodiment. Variations in the configuration can be achieved by changing the number of bits in the digital watermark data to be embedded, the combination of the color components to be embedded, the type of error detection code, the type of encryption method, and the type of random number generation method. can get. Here, the methods and types that each component can take will be described in detail.

1. Number of embedded digital watermark data bits (1) From the aspect of error detection code, an arbitrary value can be taken under the condition that the error detection code can be configured. For example, if it is desired to minimize the data, the data is 1 bit and the parity is 1 bit. The maximum value is determined by the size of the image. For example, when a VGA image of 640 × 480 pixels is handled, the number of DCTs of the Y component is 4800. Therefore, when only the Y component is used, the maximum number of data bits is 4800 or less including the error detection code. .
(2) From the aspect of the encryption method, the number of bits is restricted by how the encryption method is selected. The camellia cipher described in the detailed description is called a block cipher, and converts data having a 128-bit length into one block into 128-bit encrypted code data. In this case, the number of bits including the error detection code is desirably a multiple of the block length 128. The reason why it is not essential but desirable is that even if it is not a multiple of 128, it can be made a multiple of 128 by adding 0 or 1 data (padding) to the extent that the number of bits is insufficient. However, it can be said that it is more advantageous in terms of application to add some meaningful data than to add meaningless data by padding.
(3) What data to embed depends on the application. For example, when using video as evidence, the camera-specific ID, the shooting time (obtained from the camera clock, if any), the shooting location (by GPS, etc.) May be embedded). Further, in order to specify the alteration location, a serial number from 0 or 1 may be assigned to each group in which the video is grouped and embedded. An example of the data format to be embedded is shown in FIG. The embedded data and format are not limited to this example, and can be arbitrarily designed.

2. Combinations of embedding target color component selection (1) As described in the description of the above embodiment, selection is made from Y, Cb, and Cr. Each of them can be selected individually, any two can be selected, or all three can be selected. When only Y is selected, there is an advantage that black and white images can be handled. When only Cb is selected, the human vision is less likely to perceive the change in the blue-yellow component, so that the digital watermark is less noticeable. It is difficult to select only Cr because there is no particular advantage. In the case of selecting a plurality of color components, the number of DCT blocks increases, so that there is an advantage that more data can be embedded.

3. Error detection code type (1) Row parity. This is the simplest error detection code, in which original bit data is arranged in a matrix form and 1 bit of parity bit is added to each row. That is, for example, as shown in FIG. 18, the embedded data is arranged in 64 rows × 7 columns (= 448), and even parity pits are added to the end of each row. The even parity is a method of setting the parity bit to 0 or 1 so that the number of 1s in the bit arrangement of each row including the parity bit becomes an even number.

If the parity is odd, the parity bit is selected so that the number of 1s is odd. By adding the parity bit, the total number of bits increases. In the description of the embodiment, the original data 448 bits are arranged in 64 rows, so the total bits increase by 64 bits to 448 + 64 = 512 bits. Similar to row parity is column parity. The only difference is that the addition of parity is performed in units of columns, and the others are the same as row parity.
(2) Matrix parity. Original bit data is arranged in a matrix form, and parity bits are added to each row and each column. Also called vertical and horizontal parity. Although more parity bits are added than row parity and column parity alone, the accuracy of error detection is higher.
(3) Hamming code (see pages 47 to 53 of Non-Patent Document 3). In the Hamming code, if the number of original data bits is k and the number of check bits (redundancy bits) is m, the relationship is 2 ^m −m ≧ k + 1. For example, assuming that the total number of bits is about 500 bits, there are combinations of m = 9 and k = 502. That is, the original data is 502 bits, the inspection bit is 9 bits, and the total bit is 511 bits. Although the Hamming code can not only detect errors but also correct errors, the present invention does not use an error correction function.
(4) Cyclic code (CRC code; see pages 54 to 65 of Non-Patent Document 3). Cyclic codes have the feature of being easily implemented as hardware. Although the cyclic code can correct not only errors but also errors, the present invention does not use an error correction function.
(5) BCH code (see pages 66 to 75 of Non-Patent Document 3). The BCH code is a kind of cyclic code and has a high ability to correct random errors. However, since the error correction function is not used in the present invention, the BCH code is adopted only when the BCH coding module already exists and there is a cost advantage by diverting it.
(6) RS code (Reed-Solomon code; see pages 76 to 81 of Non-Patent Document 3). The RS code is a kind of cyclic code and has high burst error correction capability. However, since the error correction function is not used in the present invention, the RS code is employed only when the RS encoding module already exists and there is a cost advantage by diverting it.

4). Types of encryption methods (1) Camellia encryption (see Non-Patent Document 4). The Camellia cipher is a common key 128-bit block cipher, and a common key is used for encryption and decryption. The 128-bit block cipher indicates that 128-bit original data can be encrypted into 128-bit encrypted data, and 128-bit encrypted data can be decrypted into 128-bit original data. A common key is required for encryption and decryption. The common key can be 128-bit length, 196-bit length, or 256-bit length. The longer the key length, the higher the cryptographic strength (hardness to break the encryption).
(2) AES encryption (see Non-Patent Document 5). The AES cipher is a common key 128-bit block cipher, and a common key is used for encryption and decryption. The 128-bit block cipher indicates that 128-bit original data can be encrypted into 128-bit encrypted data, and 128-bit encrypted data can be decrypted into 128-bit original data. A common key is required for encryption and decryption. The common key can be 128-bit length, 196-bit length, or 256-bit length. The longer the key length, the higher the cryptographic strength (hardness to break the encryption).

5). Random number generation method (1) Linear congruence method. This is an algorithm for generating a pseudo-random number sequence. The algorithm is given by the recurrence formula Xn + 1 = (A * Xn + B) Mod M. A, B, and M are constants, and M> A, M> B, A> 0, and B> 0. X ₀ is a seed of random numbers, and when this is determined, X ₁ , X ₂ ,. . . Is obtained in a reproducible form. In the present invention, appropriate values for M, A, and B are determined in advance, and the information is shared between the embedding side and the detection side. The random seed X ₀ may also be determined as an appropriate value and shared between the embedding side and the detection side, but in the present invention, it is easy to use the value used for the encryption key as it is.
(2) Mixed joint method. This is an algorithm for generating a pseudo-random number sequence. In this algorithm, an integer initial value a and values b and c are determined and substituted into the equation ab + c = a ′. A random number is obtained by extracting the same number of digits as a from the center of the obtained a ′. Thereafter, the operation of substituting the random number obtained in the above equation a again is repeated to obtain a pseudo-random number sequence. The initial value a is a seed for random numbers. An appropriate value for the seed of the random number may be determined and shared between the embedding side and the detection side, but in the present invention, it is easy to use the value used for the encryption key as it is.

6). Variations of constituent elements of the invention Examples of the inventions according to claims 2 and 5 will now be described. Some digital cameras are manufactured using a camera module in which an optical unit and an electronic circuit unit are integrated, and an image received by the optical unit is output from the camera module as JPEG format data. In this case, DCT coefficients in the middle of JPEG encoding cannot be taken out. Claims 2 and 5 are configurations for applying the present invention to a digital camera using such a camera module. Since the DCT coefficient cannot be extracted from the camera module, JPEG format data is received, data shaping from the JFIF format and entropy decoding are performed to obtain the DCT coefficient. With this configuration change, the DCT coefficient can be changed as in the first and fourth aspects.

  This embodiment will be described below with reference to FIGS. FIG. 19 shows another example of the configuration of a JPEG encoding apparatus with watermark embedding to which the present invention is applied. In FIG. 19, reference numeral 111 denotes video input means for inputting a digital video to be embedded.

  112 is a JPEG encoding means for JPEG encoding the video. For the input video, YCbCr conversion processing, downsampling processing, DCT processing, DCT coefficient quantization processing, entropy compression processing, and entropy compression processing in a normal JPEG encoding process A shaping process into the JFIF format (that is, the processes in steps S12 to S17 in FIG. 1A) is performed.

  A data shaping / entropy decoding unit 113 receives JPEG format data from the JPEG encoding unit 112, shapes the data into a data string from the JFIF format, and performs entropy decoding to obtain a DCT coefficient.

  Reference numeral 114 denotes digital watermark data input means for inputting digital watermark data to be embedded in the input video.

  Reference numeral 115 denotes digital watermark data error detection coding means for providing the input digital watermark data with error redundancy code data by using a predetermined error detection coding method.

  Reference numeral 116 denotes encryption means for converting the error detection code data into code data that has been encrypted by a predetermined encryption method.

  Reference numeral 117 denotes color component selection means for selecting a color component in which the encoded code data is embedded.

  Reference numeral 118 denotes DCT block group dividing means for dividing the DCT block of the selected color component into groups of the same number as the number of bits of the encoded code data, and setting the remaining DCT blocks alone as one group.

  Reference numeral 119 denotes maximum value selection means for selecting a value having the maximum absolute value from among the quantized coefficients of the DCT blocks belonging to the divided groups.

  Reference numeral 120 denotes reference value selection means for selecting one reference value from the quantized coefficients excluding those having the maximum absolute value of the quantized coefficients, based on the random number value obtained from the predetermined random number generating means 121.

  122 is a DCT block in which bit 0 of the encrypted code data is embedded, bit 1 of the encrypted code data is set so that the difference between the maximum value of the quantization coefficient of the DCT block and the reference value is an even number. In the case of a DCT block to be embedded, the maximum value of the quantization coefficient absolute value is set to −K, 0, K (K is an arbitrary value so that the difference between the maximum value of the quantization coefficient of the DCT block and the reference value becomes an odd number. The coded code data embedding means embeds a predetermined pattern of 0 and 1 in the DCT block of the remainder group.

  Reference numeral 123 denotes entropy compression means for compressing the quantized coefficients after completion of the embedding in the same manner as the entropy compression processing in the normal JPEG encoding process.

  Reference numeral 124 denotes JFIF format shaping means for shaping the compressed data so as to conform to a predetermined JPEG format.

  The functions of the means 111 to 124 shown in FIG. 19 are achieved by, for example, a computer.

  Next, a processing procedure performed by the apparatus of FIG. 19 will be described with reference to a flowchart of FIG. First, a digital video to be embedded is obtained by a video input step (step S111).

  Next, in step S112, the same processing as in the normal JPEG encoding process (steps S12 to S17 in FIG. 1A) proceeds, and is output as JPEG format data.

  Next, in a data shaping / entropy decoding step (step S113), the JPEG format data is shaped from the JFIF format into a data string, and entropy decoding is performed to obtain DCT coefficients.

  Next, the digital watermark data obtained in the digital watermark data input step (step S114) is processed in the same manner as in step S33 of FIG.

  Next, in a watermark data error detection encoding step (step S115), redundancy is added to the digital watermark data by the error detection encoding method described in step S34 of FIG.

  Next, in the encryption step (step S116), the 512-bit data subjected to the error detection encoding is encrypted by the method described in step S35 of FIG. 3 to generate encrypted data.

  Next, in the color component selection step (step S117), the color component in which the encrypted code data is embedded is selected by the method described in step S36 of FIG.

  Next, in the DCT block group division step (step S118), the DCT block in the video is divided into groups by the method described in step S37 of FIG.

  Next, in the encryption code data embedding step (step S122), the encryption code data is embedded in the DCT block belonging to each group by the method described in step S41 of FIG.

  When embedding of the encrypted code data bits is completed for all the groups, the entropy compression step (step S123) is performed in the same manner as the normal JPEG encoding process (steps S16 and S17 in FIG. 1A). A data shaping step (step S124) is executed so as to conform to a predetermined JFIF format and output as a JPEG video.

  Further, some or all of the functions of each means in the tamper detection digital watermark embedding apparatus and the digital tamper detection apparatus of this embodiment are configured by a computer program, and the program is executed using the computer. The present invention can be realized, each procedure in the falsification detection digital watermark embedding method and the digital falsification detection method of the present embodiment can be configured by a computer program, and the program can be executed by the computer. Needless to say, a program for realizing the functions of the computer is recorded on a computer-readable recording medium such as FD (Floppy (registered trademark) Disk), MO (Magneto-Optical disk), ROM (Read Only M). memory), memory card, CD (Compact Disk) -ROM, DVD (Digital Versatile Disk) -ROM, CD-R, CD-RW, HDD, removable disk, etc. for storage and distribution Is possible. It is also possible to provide the above program through a network such as the Internet or electronic mail.

The flowchart which shows the process of normal JPEG encoding and JPEG decoding which are employ | adopted by this invention. 1 is a block diagram showing an embodiment of a digital watermark embedding device for tampering detection according to the present invention. The flowchart which shows one Embodiment of the digital watermark embedding method for tampering detection of this invention. Explanatory drawing which shows the example of the quantization coefficient in the embodiment of this invention. Explanatory drawing which shows an example of DCT block grouping in the embodiment of this invention. Explanatory drawing which shows the other example of DCT block grouping in the embodiment of this invention. The flowchart which shows the encryption code data embedding process performed by FIG.3 S41. The flowchart which shows the selection process of the maximum value M1 performed by step S52 of FIG. Explanatory drawing which shows the mode of the reverse zigzag scan in the embodiment of this invention. Explanatory drawing which shows the mode of the raster scan which skipped the maximum value in the example of embodiment of this invention. The flowchart which shows the adjustment process of the maximum value M1 performed by step S54 of FIG. The block diagram which shows one Example of the tampering detection apparatus using the digital watermark of this invention. 5 is a flowchart illustrating an embodiment of a tampering detection method using a digital watermark according to the present invention. 14 is a flowchart showing error detection code data detection processing executed in step S88 of FIG. Explanatory drawing which shows the case where there is no error as a result of decoding in the embodiment of the present invention. Explanatory drawing which shows the mode of a display of the tampering location in the example of embodiment of this invention. Explanatory drawing which shows the structure of the format of embedded data in the embodiment of this invention. Explanatory drawing which shows the mode of even-numbered row parity bit addition in the example of embodiment of this invention. The block diagram which shows the other Example of an electronic watermark embedding apparatus for tampering detection of this invention. 10 is a flowchart showing another embodiment of a digital watermark embedding method for falsification detection according to the present invention.

Explanation of symbols

  31, 111 ... Video input means, 32, 112 ... JPEG encoding means, 33, 114 ... Digital watermark data input means, 34, 115 ... Digital watermark data error detection encoding means, 35, 116 ... Encryption means, 36, 83, 117 ... color component selection means, 37, 84, 118 ... DCT block group division means, 38, 85, 119 ... maximum value selection means, 39, 86, 120 ... reference value selection means, 40, 87, 121 ... random numbers Generation means, 41, 122 ... Encrypted code data embedding means, 42, 123 ... entropy compression means, 43, 124 ... JFIF format shaping means, 81 ... JPEG encoded video data input means, 82, 113 ... data shaping / entropy Decoding means, 88 ... means for obtaining bit data from the difference between the maximum value and the reference value, 91 ... JPEG decoding means, 92 ... error Raw place superimposed display means.

Claims (10)

In a system that embeds a digital watermark for falsification detection at the same time as JPEG encoding the video,
A step in which video input means inputs video;
JPEG encoding means performs a YCbCr conversion process, a downsampling process, a DCT process, and a DCT coefficient quantization process in a normal JPEG encoding process on the input video;
A step in which digital watermark data input means inputs digital watermark data to be embedded;
Digital watermark data error detection encoding means, the step of giving the input digital watermark data to the data by a predetermined error detection encoding method to make the error detection code data;
An encryption means sets the error detection code data to code data encrypted by a predetermined encryption method;
A step of selecting a color component for embedding the encoded code data;
DCT block group dividing means divides the DCT block of the selected color component into the same number of groups as the number of bits of the encoded code data, and the remaining DCT blocks are made into one group by themselves;
A maximum value selecting means selecting a value having the maximum absolute value from among the quantized coefficients of the DCT blocks belonging to each of the divided groups;
A reference value selection means, based on a random number value obtained from a predetermined random number generation means, to select one reference value from the quantized coefficients excluding the one with the maximum absolute value of the quantized coefficients;
If the encrypted code data embedding means is a DCT block in which bit 0 of the encrypted code data is embedded, the encryption is performed so that the difference between the maximum value of the quantization coefficient of the DCT block and the reference value is an even number. If the DCT block embeds bit 1 of the already-encoded code data, the maximum value of the quantized coefficient absolute value is −K, 0, so that the difference between the maximum value of the quantized coefficient of the DCT block and the reference value becomes an odd number. Encrypted code data embedding step of adding any one of K (K is an arbitrary positive / odd number), and a predetermined pattern of 0 and 1 is the same as the encrypted code data embedding step in the DCT block of the remainder group A remainder group data embedding step embedded in the method,
JPEG encoding means includes a step of entropy compressing the quantized coefficient after completion of the embedding in the same manner as in a normal JPEG encoding process, shaping it so as to conform to a predetermined JPEG format, and completing JPEG encoding. An electronic watermark embedding method for falsification detection characterized by the above. Between the step of performing YCbCr conversion processing, downsampling processing, DCT processing and DCT coefficient quantization processing in the normal JPEG encoding step, and inputting the embedded watermark data,
JPEG encoding means compresses the quantized coefficient and shapes it to conform to a predetermined JPEG format;
The method for embedding a digital watermark for tampering detection according to claim 1, wherein the data shaping / entropy decoding means performs a step of data shaping and entropy decoding on the JPEG format to obtain a DCT coefficient value. . In the system for checking whether or not the video is falsified by detecting the digital watermark from the JPEG-encoded video in which the digital watermark for alteration detection is embedded by the digital watermark embedding method for alteration detection according to claim 1 or 2.
Video input means for inputting the JPEG encoded video;
A step in which JPEG decoding means performs a data shaping process and an entropy decoding process in a normal JPEG decoding process on the input JPEG encoded video;
A step of selecting a color component in which the error detection code data is embedded;
DCT block group dividing means for dividing the DCT block of the selected color component into a group including DCT blocks by the same number as the number of bits of the encrypted encoded data;
A maximum value selecting means selecting a value having the maximum absolute value from among the quantized coefficients of the DCT blocks belonging to each of the divided groups;
A reference value selection means, based on a random number value obtained from a predetermined random number generation means, to select one reference value from the quantized coefficients excluding the one with the maximum absolute value of the quantized coefficients;
Means for obtaining bit data from the difference between the maximum value and the reference value, outputting 0 if the difference between the maximum value and the reference value is even;
Encryption / decryption means performs the associated encryption / decryption using the output 0 or 1 as a predetermined encryption / decryption input bit;
Error detection / decryption means performs predetermined error detection / decryption on the encrypted / decrypted data, and if there is an error, if there is an error, a group including a DCT block in which an error has occurred, and if there is no error, embedding is performed. Obtaining the digital watermark data,
JPEG decoding means performing the normal JPEG decoding process to the input JPEG encoded video as a video; and
A tamper detection method using a digital watermark, characterized in that a display means displays the presence / absence of the error and the position of a DCT block belonging to the group in which the error occurred together with the decoded video. In a system that embeds a digital watermark for falsification detection at the same time as JPEG encoding the video,
Video input means for inputting video;
JPEG encoding means for performing YCbCr conversion processing, downsampling processing, DCT processing, and DCT coefficient quantization processing in a normal JPEG encoding process for the input video;
Digital watermark data input means for inputting digital watermark data to be embedded;
Digital watermark data error detection encoding means for providing redundancy to the input digital watermark data by a predetermined error detection encoding method to obtain error detection code data;
Encryption means for converting the error detection code data into code data encrypted by a predetermined encryption method;
Color component selection means for selecting a color component for embedding the encrypted code data;
DCT block group dividing means for dividing the DCT block of the selected color component into a group of the same number as the number of bits of the encoded code data, and setting the remaining DCT blocks alone as one group;
Maximum value selection means for selecting a value having the maximum absolute value from among the quantization coefficients of the DCT blocks belonging to each of the divided groups;
A reference value selecting means for selecting one reference value from the quantized coefficients excluding the one with the maximum absolute value of the quantized coefficients based on a random number value obtained from a predetermined random number generating means;
If the DCT block embeds bit 0 of the encrypted code data, the DCT block embeds bit 1 of the encrypted code data so that the difference between the maximum value of the quantization coefficient of the DCT block and the reference value is an even number. If so, the maximum value of the quantization coefficient absolute value is −K, 0, K (K is an arbitrary positive / odd number) so that the difference between the maximum value of the quantization coefficient of the DCT block and the reference value is an odd number. Any one of the above, and encrypted code data embedding means for embedding a predetermined pattern of 0 and 1 in the DCT block of the remainder group,
Entropy compression means for compressing the quantized coefficients after the completion of embedding in the same manner as the entropy compression processing in the normal JPEG encoding process;
A tamper detection digital watermark embedding apparatus comprising: shaping means for shaping the compressed data so as to conform to a predetermined JPEG format. In a system that embeds a digital watermark for falsification detection at the same time as JPEG encoding the video,
Video input means for inputting video;
JPEG encoding means for encoding the input video by a normal JPEG encoding process;
Data shaping and entropy decoding means for obtaining DCT coefficient values by performing data shaping and entropy decoding on the JPEG format encoded by the JPEG encoding means;
Digital watermark data input means for inputting digital watermark data to be embedded;
Digital watermark data error detection encoding means for providing redundancy to the input digital watermark data by a predetermined error detection encoding method to obtain error detection code data;
Encryption means for converting the error detection code data into code data encrypted by a predetermined encryption method;
Color component selection means for selecting a color component for embedding the encrypted code data;
DCT block group dividing means for dividing the DCT block of the selected color component into a group of the same number as the number of bits of the encoded code data, and setting the remaining DCT blocks alone as one group;
Maximum value selection means for selecting a value having the maximum absolute value from among the quantization coefficients of the DCT blocks belonging to each of the divided groups;
A reference value selecting means for selecting one reference value from the quantized coefficients excluding the one with the maximum absolute value of the quantized coefficients based on a random number value obtained from a predetermined random number generating means;
If the DCT block embeds bit 0 of the encrypted code data, the DCT block embeds bit 1 of the encrypted code data so that the difference between the maximum value of the quantization coefficient of the DCT block and the reference value is an even number. If so, the maximum value of the quantization coefficient absolute value is −K, 0, K (K is an arbitrary positive / odd number) so that the difference between the maximum value of the quantization coefficient of the DCT block and the reference value is an odd number. Any one of the above, and encrypted code data embedding means for embedding a predetermined pattern of 0 and 1 in the DCT block of the remainder group,
Entropy compression means for compressing the quantized coefficients after the completion of embedding in the same manner as the entropy compression processing in the normal JPEG encoding process;
A tamper detection digital watermark embedding apparatus comprising: shaping means for shaping the compressed data so as to conform to a predetermined JPEG format. In the system for checking whether or not the video is falsified by detecting the digital watermark from the JPEG encoded video in which the digital watermark for falsification detection is embedded by the digital watermark embedding device for alteration detection according to claim 4 or 5.
Video input means for inputting the JPEG encoded video;
Means for performing a data shaping process and an entropy decoding process in a normal JPEG decoding process on the input JPEG encoded video;
Color component selection means for selecting a color component in which the error detection code data is embedded;
DCT block group dividing means for dividing the DCT block of the selected color component into a group including DCT blocks by the same number as the number of bits of the encoded data.
Maximum value selection means for selecting a value having the maximum absolute value from among the quantization coefficients of the DCT blocks belonging to each of the divided groups;
A reference value selecting means for selecting one reference value from the quantized coefficients excluding the one with the maximum absolute value of the quantized coefficients based on a random number value obtained from a predetermined random number generating means;
Means for obtaining bit data from the difference between the maximum value and the reference value for outputting 0 if the difference between the maximum value and the reference value is an even number, and 1 if the difference between the maximum value and the reference value;
Encryption / decryption means for performing the associated encryption / decryption using the output 0 or 1 as an input bit for predetermined encryption / decryption;
Predetermined error detection / decoding is performed on the decrypted data, the presence / absence of an error, a group including a DCT block in which an error has occurred if there is an error, and embedded digital watermark data if there is no error, Error detection decoding means for obtaining
JPEG decoding means for converting the input JPEG encoded video into a video by performing a normal JPEG decoding process;
An alteration detection apparatus using digital watermark, comprising: display means for conspicuously displaying the presence / absence of the error and the position of the DCT block belonging to the group in which the error occurred together with the decoded video.   A digital watermark embedding program for falsification detection for causing a computer to execute each step according to claim 1 or 2.   A tamper detection program using a digital watermark for causing a computer to execute each step according to claim 3.   A computer-readable recording medium on which the falsification detecting digital watermark embedding program according to claim 7 is recorded.   A computer-readable recording medium on which the falsification detection program using the digital watermark according to claim 8 is recorded.

Images