JP4446400B2 - Digital watermark embedding method and authentication method for content - Google Patents

Description

  BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a content digital watermark embedding method and an authentication method, and more particularly, to embed data that certifies that distributed content has not been tampered with a digital watermark, And authentication methods.

  Conventional determination of content falsification has been performed, for example, as shown in FIG. That is, the transmission side obtains a hash value a by applying a hash function to the content 50, and obtains a digital signature b using the secret key. Then, the content 50 and the digital signature b, public key, and public key certificate as the accompanying data are transmitted to the receiving side. On the other hand, the receiving side compares the hash value a obtained by applying a hash function to the received content 51 and the hash value a ′ obtained by decrypting the received digital signature b using the public key, The contents are determined to be falsified depending on whether or not they match.

An example of the background art is described in Japanese Patent Application Laid-Open No. 2001-78013.
JP 2001-78013 A

  The digital watermark is a method of storing data secretly inside the content, that is, a method of embedding data by modifying the content itself. For this reason, the content in which the digital watermark is embedded does not match the original content.

  Therefore, when the transmission side embeds and transmits a digital signature calculated from the original content as a digital watermark, the digital signature calculated from the received content does not match the digital signature on the transmission side at the time of inspection on the reception side, etc. There is a problem that it is not possible to make a tampering determination.

  The present invention has been made in view of the above-described prior art, and an object of the present invention is to provide a digital watermark embedding method and an authentication method for content, which can embed a digital watermark in content and determine whether the content is falsified. There is.

  In order to achieve the above-mentioned object, the present invention provides a means for DCTing image data or audio data as contents, and a signal element not used for digital watermark embedding in a DCT coefficient, which is a coarser quantum signal applied in MPEG coding. Means for obtaining the hash value, means for embedding the hash value in a signal element used for embedding the digital watermark in the DCT coefficient, and means for quantizing the signal element in which the hash value is embedded It is characterized in that it is equipped.

  Another feature is that a method for authenticating the contents embedded with the digital watermark is provided.

  According to the first aspect of the present invention, it is possible to provide a content authentication method capable of determining whether a content is falsified with respect to content such as an image frame or an audio frame.

  According to the second aspect of the present invention, authentication data can be accurately authenticated based on content at the time of use, for example, received content and digital watermark data extracted from the content.

  The present invention will be described below with reference to the drawings. FIG. 1 is a schematic system diagram of a multimedia content file providing service for a mobile phone including an embodiment of the present invention.

  The content file transmission side (content file production side) performs watermarked content creation processing 2 by inserting watermark data into content 1, for example, video, music, and other data. Next, hash value calculation 3 is performed by applying a hash function to the watermarked content, and then encryption processing 4 is performed using a secret key possessed only by the producer. Next, a watermark insertion process 5 is performed to insert the encrypted data into the watermarked content to create a content file 6.

  On the other hand, on the receiving side (mobile device side), the content file is downloaded and temporarily stored in the file storage unit 11. The verification unit 12 verifies whether or not the downloaded content file has been produced with a valid production tool. If it is verified that the content file has been produced with a valid production tool, the verification unit 12 Register with the registration unit 13 and provide it to the user. For example, the video / music source playback function is provided to the user. On the other hand, when it is not verified that the content has been produced with a valid production tool, a measure is taken not to provide the content file to the user.

  Next, the function of an embodiment of the present invention will be described with reference to FIGS. FIG. 2 shows functions on the content production side (or transmission side), and FIG. 3 shows functions on the reception side.

  Content 1 is content obtained by digitizing images, music, and the like. For example, watermark data “000... 0” of all 0 is embedded in the content 1 to form a watermarked content 15 (step S1). If the content 1 is an image, the all-zero watermark data “0000... 0” is stored in the least significant bit of the value of each pixel represented by 8 bits, for example. If 1 is music, for example, it can be put in the least significant bit of each sampling value represented by 16 bits.

  Next, a hash value a by, for example, the MD5 algorithm is obtained from the watermarked content 15 (step S2), and the digital signature b is calculated by encrypting the value a using, for example, RSA encryption (step S3). Next, the public key and the data of the public key certificate are added to the digital signature b to create a watermark c (step S4). For example, several bits can be assigned to the public key and public key certificate data. Thereafter, the watermark c is embedded in the watermarked content 2 as a new digital watermark (step S5). Thus, the content file 6 created by the content production side (or transmission side) is completed.

  The receiving side (user side), for example, a mobile phone, acquires the content file 6 through a wireless network or a wired connection. At this time, the verification unit 12 checks whether the received content has been tampered with. This process will be described with reference to FIG.

  When the content file 6 is received, the digital watermark embedded in the content file 6, that is, the digital signature c is acquired (step S11). Next, the public key certificate is authenticated using a root certificate stored in advance in a receiving device such as a mobile phone (step S12). When the authentication is performed, the digital signature b encrypted with the private key is decrypted using the public key (step S13). By this decryption process, a hash value a ′ is obtained.

  Further, watermark data “000... 0” of all 0 is embedded in the content file 6 at the same position as that embedded on the transmission side (step S14). Thereby, the watermarked content 16 is obtained. If there is no falsification, the watermarked content 16 is the same as the watermarked content 15 of FIG. Next, a hash value a by, for example, the MD5 algorithm is obtained from the watermarked content 11 (step S15).

  Then, the hash value a obtained in step S15 and the hash value a ′ obtained in step S13 are compared (step S16). If they match, there is no falsification, and if they do not match, the digital signature on the producer side is compared. After the embedding, it is determined that the content has been modified.

  If the authentication in step S12 is not successful, it is determined that the content is not produced by a regular content creator. The all-zero watermark data embedded in step S1 is an example, and the watermark data is not limited to this as long as it is known in advance.

  If the received content is altered or is not a content created by an authorized content creator, the receiving side takes measures such as not playing the content, for example. This measure makes it possible to prevent falsified content and content from unauthorized content creators from being provided to, for example, mobile phone users.

  In this embodiment, watermark data known in advance on the transmission / reception side is embedded in a predetermined position of the content to obtain the hash value a. Therefore, it is possible to obtain the hash value by eliminating the influence of embedding the digital watermark. it can. For this reason, the receiving side can correctly determine whether the received content has been tampered with or whether the content is a content by an authorized content creator.

  Next, a second embodiment of the present invention will be described with reference to FIG. For example, it is assumed that an image frame or a music frame is divided into blocks and the DCT coefficient of the block is manipulated.

  Now, it is assumed that the image element 21 is divided into blocks of 8 × 8 pixels, for example, and blocks 21a, 21b and 21c in which watermark data is embedded, that is, signal elements used for digital watermarking are determined in advance. The content 22 is formed by arranging DCT coefficients of blocks of the image frame 21 in time series.

  In the present embodiment, a hash value a is obtained by applying a hash function to a DCT coefficient value (block shaded) of a block in which no DCT coefficient is embedded, that is, a signal element not used for digital watermarking (step S21). Next, the hash value a is encrypted to obtain a digital signature b (step S22). Next, a public key and a public key certificate are added to the digital signature b to create a watermark c (step S23), which is embedded in the blocks 21a, 21b, and 21c. The watermarked content thus formed is uploaded as a content file 6.

  On the receiving side, the hash value a is obtained from the downloaded watermarked content by applying a hash function to the DCT coefficient values (blocks shaded) of the block in which the watermark data is not embedded. Further, the digital signature b, the public key, and the public key certificate are extracted from the blocks 21a, 21b, and 21c in which the watermark data c is embedded. Then, the digital signature b is decrypted using the public key to obtain a hash value a ′. Next, the hash values a and a 'are compared, and if they match, it is determined that there is no falsification, and if they do not match, it is determined that falsification has occurred. If the public key certificate cannot be authenticated, it is determined that the content is not a content created by an authorized content creator.

  According to this embodiment, since a hash value is calculated using a block in which watermark data is not embedded, it is possible to obtain a hash value by eliminating the influence of digital watermark embedding.

  As a modification of the second embodiment, when Fourier transform is performed on the luminance signal component of an image, a hash value is calculated from coefficient values other than the Fourier transform coefficient used for digital watermark embedding.

  Next, a third embodiment of the present invention will be described. In the above-described embodiment, one image frame or audio frame to be transmitted, or data of all image frames or audio frames is used as one content to determine whether or not tampering has occurred. This makes it possible to determine whether or not tampering has occurred for each predetermined number of image frames or audio frames.

  In this embodiment, as shown in FIG. 5, image frames or audio frames to be transmitted are divided into groups A, B,..., And index information is embedded as watermark information for each group. If groups A, B,... Are formed every second, 30 frames are defined as one group for an image. Then, a hash value a is calculated from the 30 frames, and this is encrypted to obtain a digital signature b. A public key, a public key certificate, and index information are added to the digital signature b to create watermark information c. To do. As the index information, for example, group A can be “0010” and group B can be “0100”. The watermark information c is embedded in each of the 30 frames by the method described in the first or second embodiment.

  On the receiving side, the digital signature b, public key, public key certificate, and index information are extracted from the received content by the method described in the first or second embodiment, and a group of frames is extracted from the index information. Ask. Next, the hash value a is calculated from the frames belonging to the group, and the cipher is decrypted by applying the public key to the digital signature b to obtain the hash value a ′. If the hash values a and a 'do not match, it is determined that the content of the group has been tampered with.

  As described above, according to the present embodiment, it becomes possible to determine the presence / absence of falsification for each group of image frames or audio frames. For example, when providing content to a mobile phone user, the first and second In the embodiment, if even a part of the content is tampered with, all the content cannot be provided, but in this example, a group frame that has not been tampered with can be provided.

  Next, a fourth embodiment of the present invention will be described with reference to FIG. This embodiment provides a method for avoiding the influence of a DCT coefficient quantization operation error when the content embedded with watermark information created in the first to third embodiments is encoded and transmitted.

  For example, the DCT coefficient value used for the hash value calculation is quantized coarser than the quantization used in MPEG coding (hereinafter referred to as “coarse quantization than MPEG”), and the hash value is calculated using the quantized value. Try to calculate.

  For example, as shown in FIG. 6A, when attention is paid to one block 31 (for example, 8 × 8 pixels) in the image data that is the content, the block 31 is subjected to DCT to obtain an 8 × 8 element. A DCT coefficient 32 consisting of Next, the coefficient 32a for embedding the watermark data is removed, and the DCT coefficient 32 is quantized coarsely than MPEG, and the hash value a is calculated from the data in which the hash value of the quantized data 33 is not embedded. Then, the hash value a is embedded in the DCT coefficient 32a. In this way, the quantized data 34 in which the hash value a is embedded is obtained. Thereafter, the image data is returned to the image data by inverse DCT, and MPEG encoding is performed for video data distribution.

  Next, when the quantized data 34 is inversely quantized by MPEG on the receiving side, the data that is the basis for the calculation of the hash value a is data that is coarser quantized than the MPEG. Inverse quantization without being affected by. For this reason, the influence of the quantization operation error can be avoided. That is, as shown in the drawing (b), the data 31 'decoded by the inverse quantization is subjected to DCT to obtain data 32'. The data 32 'is subjected again to a coarser quantization operation than MPEG (this operation is the same as that for embedding) to obtain data 32', and a hash value a 'is calculated from data other than the coefficient 32a'. Next, the hash value a extracted from the coefficient 32 a ′ is compared with the a ′ to determine whether or not it has been tampered with. As described above, according to the present embodiment, it becomes possible to perform falsification determination without being affected by the quantization operation error.

1 is a schematic system diagram of a multimedia content file providing service for a mobile phone that is an application example of an embodiment of the present invention. FIG. It is explanatory drawing which shows the function by the side of the transmission of the 1st Embodiment of this invention. It is explanatory drawing which shows the function of the receiving side of this 1st Embodiment. It is explanatory drawing which shows the function by the side of the transmission of the 2nd Embodiment of this invention. It is explanatory drawing which shows the function by the side of the transmission of the 3rd Embodiment of this invention. It is explanatory drawing which shows the function of the 4th Embodiment of this invention. It is explanatory drawing of the conventional system.

Explanation of symbols

  DESCRIPTION OF SYMBOLS 1 ... Content, 2 ... Watermark content creation processing part, 3 ... Hash value calculation processing part, 4 ... Encryption processing part, 5 ... Watermark insertion processing part, 6 ... Content File 11, file storage unit 12, verification unit 13, registration unit 15, 16, watermarked content 21 image frame 21 a, 21 b, 21 c watermark Data embedding block.

Claims (2)

Means for DCTing image data or audio data as content;
Means for performing coarser quantization of signal elements not used for digital watermark embedding in a DCT coefficient, which is applied in MPEG encoding, and obtaining a hash value thereof;
Means for embedding the hash value in a signal element used for embedding a digital watermark in the DCT coefficient;
A digital watermark embedding method for content, comprising: means for quantizing a signal element in which the hash value is embedded. An authentication method for content created by the digital watermark embedding method described in claim 1,
Means for dequantizing and decoding the content;
Means for DCTing the decoded signal elements;
Means for extracting a hash value from a signal element used for digital watermark embedding in a DCT coefficient;
Means for subjecting the DCT signal elements to coarse quantization applied during hash value calculation in claim 1;
Means for calculating a hash value from signal elements not used for watermark embedding in the coarse quantized DCT coefficients;
A content authentication method comprising: a means for comparing the retrieved hash value with the calculated hash value and authenticating that the hash value is not falsified if they match.

Images