JP2001144935A - Watermark data imbedding device, watermark data imbedding method, watermark data reader and watermark data read method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a method for managing keys by having only to manage a sole common key because a session key used to imbed watermark to contents data cannot be known even when a session key used to imbed other watermark to other contents data is indexed through the analyses of the other contents data to which the other watermark data are imbedded. SOLUTION: Characteristic data are extracted from contents data and the characteristic data are synthesized with the common key to obtain a session key, and using this session key imbeds watermark data to the contents data.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION In a system for distributing digitized content data, in order to prevent illegal copying and resale of the content, information on the creator such as the copyright of the content is watermarked (digital wa).
watermark data embedding device to record as termark)
The present invention relates to a watermark data embedding method, a watermark data reading device, and a watermark data reading method.

[0002]

2. Description of the Related Art With the advance of digital technology, various works such as movies, music, and photographs are distributed as digital data. These digital data are digitized image data and audio data such as still images and moving images, and content data such as MIDI data. The digital data is recorded on a recording medium such as a CD-ROM and sold, or is distributed by a network such as the Internet. Have been sent and received.

[0003] Such digital data is characterized in that a complete copy can be easily and easily made in large quantities. This is because a user who has obtained the digital data illegally makes a copy equivalent to the original and re-creates it. It shows the possibility of distribution and resale. As a result, the price to be paid to the creator of the digital data or a person who has been legally entrusted to sell the digital data is not paid, and the copyright may be infringed.

[0004] The problem of the copyright of such digital data is becoming more serious, and in recent years, digital watermarking technology has been spotlighted as a technology for preventing this problem.

According to this technique, in order to protect the rights of a data creator and a provider, information about the creator and the user is inserted into the data in a form in which the user cannot distinguish it, and the copyright holder is identified. It is for confirming or, in the case of unauthorized use, identifying the user.

[0006] The conditions necessary for this digital watermark include:
Embedding the watermark data so that if the watermark data is forcibly removed, the content data itself will be broken (watermark information indicating the copyright remains in the content data), and where the watermark information is embedded in the content data Is difficult to understand, and even if the watermark data is embedded, the content data may be kept in the original state and quality.

A digital watermarking technique satisfying these conditions is to embed watermark data in a part (redundant part) that is not important for human perception in content data such as image data and audio data, thereby reducing the total data amount. It is considered to embed watermark data without changing.

[0008] Further, a method has been considered in which attention is paid to specific data in the content data, and a part of the specific data is replaced with other data to form a digital watermark.

[0009] Information on the specific spectrum and the embedding position of the watermark data to be embedded as described above must be agreed between the side that embeds the watermark data and the side that reads out the watermark data permitted by the copyright holder. is there. The watermark data embedded in the content data can be uniquely derived by the key data, and the key data is transferred separately from the content data, so that the watermark data can be read.

[0010]

However, when embedding watermark data in a large amount of content data, using different key data for each content data complicates the management of the key data. Absent. Also, the use of the same key data for all content data is the same if the key data is known by a malicious third party analyzing certain content data with embedded watermark data. The watermark data of all content data in which the watermark data is embedded using the key data of
This makes it possible to easily remove the watermark data.

[0011]

SUMMARY OF THE INVENTION Accordingly, the present invention provides a feature extracting unit 1 for extracting feature data from content data.
A synthesizing unit 2 for synthesizing the characteristic data with a common key commonly used for all content data in which watermark data is embedded, and a one-way function operation on an output of the synthesizing unit 2 to obtain a session key. Direction function calculation means 3
And a watermark embedding unit 4 for inputting the content data and the watermark data and embedding the watermark data in the content data using the session key.

Further, by extracting characteristic data from the content data, reflecting the characteristic data on a common key commonly used for all the content data in which the watermark data is embedded, and calculating by a one-way function. , A session key is obtained, and the watermark data is embedded in the content data using the session key.

Further, feature extracting means 1 for extracting feature data from the content data in which the watermark data is embedded.
A synthesizing unit 2 for synthesizing the characteristic data with a common key commonly used for all content data when embedding watermark data, and a one-way function operation of an output of the synthesizing unit 2 to obtain a session key. The obtained one-way function calculating means 3 and the watermark reading means 5 for inputting the content data with the watermark data embedded therein and reading the watermark data from the content data with the watermark data embedded using the session key. Watermark data reading device.

Further, feature data is extracted from the content data in which the watermark data is embedded, and the feature data is reflected on a common key commonly used for all the content data when the watermark data is embedded. A watermark data reading method is provided in which a session key is obtained by performing a calculation using a one-way function, and the watermark data is read from the content data in which the watermark data is embedded using the session key. .

[0015]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS A watermark data embedding device, a watermark data embedding method, a watermark data reading device, and a watermark data reading method according to the present invention will be described below with reference to the drawings.

In the watermark data embedding apparatus and the watermark data embedding method according to the present invention, the characteristics of the content data into which the watermark data is to be embedded so that it is difficult to analyze the key data (common key) used when embedding the watermark data. Is extracted and combined with common common key data to create a different session key for each piece of content data in which watermark data is to be embedded, and the watermark data is embedded based on this.

FIG. 1 is a block diagram showing the concept of a watermark data embedding device and a watermark data embedding method according to the present invention. The content data is input to the feature extracting means 1 and also to the watermark embedding means 4. The feature extracting means 1 extracts feature data of the content data. This characteristic data differs from the content of the content, but, for example, when the content data is audio data, the frequency or the like where a spectrum having a large power at the time of performing the frequency analysis is changed, or is reduced by compression. It is desirable to extract, as the characteristic data of the content data, data that greatly deteriorates the data quality.

Also, the output from the above-described feature extracting means 1 and the synthesizing means 2 to which a common key used commonly for all content data are inputted, the above-mentioned feature data of the content data and the common key are synthesized. However, a method such as exclusive OR is used. In this embodiment, the characteristic data of the content data and the common key are combined. However, the present invention is not limited to this as long as the characteristic data of the content data can be reflected on the common key.

Further, the key data obtained by synthesizing the characteristic data of the content data and the common key is input to the one-way function calculating means 3 and is calculated by a one-way function which is difficult to execute the inverse function. By performing such processing, the key data described above becomes a session key used for the watermark embedding unit 4 described later.

The watermark data is input to the watermark embedding means 4 while the content data described above is input. Then, the watermark data is embedded in the content data by the session key.

Next, a watermark data reading device and a watermark data reading method will be described with reference to FIG. According to the figure, the content data in which the watermark data is embedded is input to the watermark reading means 5 and also to the feature extracting means 1. In the feature extracting means 1, feature data of the content data is extracted from the content data in which the watermark data is embedded. The feature extraction is performed in the same manner as the feature extraction unit 1 used when embedding the watermark data.

Next, the characteristic data of the extracted content data and the common key are synthesized by the synthesizing means 2 and are applied to the one-way function calculating means 3. Thus, a session key is created, and the watermark data is read from the content data in which the watermark data input to the watermark reading means 5 is embedded using the session key.

As described above, the characteristic data of the content data to be embedded with the watermark data is extracted and combined with the common key, thereby generating a different session key for each content data, and embedding the watermark data based on the session key. ing. In addition, since the session key is obtained by using a one-way function instead of directly combining the characteristic data of the content data and the common key, it is difficult to estimate the common key from the session key. I have. When the watermark data is read, similar processing is performed to generate a session key from the common key. Since this session key is different for each content data and is disposable only for one time, even if the content data in which the watermark data is embedded can be analyzed to determine the session key, the common key is not determined. Damage is limited only to the content data, and other content data is protected.

As described above, the characteristic data of the content data is extracted from a very important part of the content data so that the same characteristic data is obtained before embedding the watermark data and after embedding the watermark data. At the same time, it is necessary to avoid a state in which the characteristic data is changed by altering the content data in which the watermark data is embedded, the session key is not correctly generated, and the watermark cannot be read. The same function is used for the feature extraction means 1 used when embedding the watermark data in the content data and the feature extraction means 1 used when the watermark data is read from the content data. The same function is used for the one-way function calculation means 3.

Next, the MI which is the actual content data is
A method of creating a session key used when embedding watermark data in DI data will be described. MIDI
Is a unified standard for computer music, and allows different instruments or models of different instruments to be played simultaneously or with the same performance. MI
The DI data is performance information for an electronic musical instrument, a sound source, and the like to emit sound.

First, note the deviation of the pitch distribution of note events as feature data of MIDI data, and extract it. In the MIDI data, changing, adding, or deleting a pitch of a note event significantly impairs an important part of the MIDI data, and thus can be said to be relatively stable feature data. Specifically, a session key is obtained by performing the following calculation. The note number of the nth note event is NoteNum [n]
Assuming that the number of note events is N, the average value H of note numbers is

[0027]

(Equation 1) And the characteristic data C of the MIDI data is

[0028]

(Equation 2) And The common key CKey {0 to 15} is 16 bytes, and the upper 8 bits Ct of the characteristic data C are set to 8 for each byte.
In a finite field GF (2 ⁸ ) of bits, CKey の 0 to 1
The session key SKey by raising to a power of 5
{0-15} can be obtained.

[0029]

(Equation 3) Using this session key, arbitrary watermark data
It can be embedded in DI data.

[0030]

As described above, according to the watermark data embedding apparatus, the watermark watermark data embedding method, the watermark data reading apparatus, and the watermark data reading method according to the present invention, when embedding watermark data in a large amount of content data. However, it is only necessary to manage one common key, and when actually embedding watermark data in each piece of content data, different session keys are used. Even if the session key is analyzed and determined, the session key of other content data cannot be known, so that the risk that the watermark data is erased from the content data in which the watermark data is embedded is reduced. Play. Further, since the common key cannot be determined even if the session key is determined, there is an effect that the risk of encountering damages such as impersonating the copyright holder is reduced.

[Brief description of the drawings]

FIG. 1 is a block diagram showing the concept of a watermark data embedding device and a watermark data embedding method according to the present invention.

FIG. 2 is a block diagram illustrating the concept of a watermark data reading device and a watermark data reading method according to the present invention.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 1 Feature extraction means 2 Synthesis means 3 One-way function calculation means 4 Watermark embedding means 5 Watermark reading means

Claims (4)

[Claims] 1. A feature extracting means for extracting feature data from content data; a combining means for combining the feature data with a common key commonly used for all content data in which watermark data is embedded; A one-way function calculating means for obtaining a session key by performing a one-way function calculation on the output of, and a watermark embedding means for inputting the content data and the watermark data and embedding the watermark data in the content data using the session key. Watermark data embedding device. 2. Extracting characteristic data from content data, reflecting the characteristic data on a common key commonly used for all content data in which watermark data is to be embedded, and calculating by a one-way function. Obtaining a session key, and embedding the watermark data in the content data using the session key. 3. A feature extracting means for extracting feature data from content data in which watermark data is embedded, and synthesizing the feature data with a common key commonly used for all content data when embedding watermark data. A one-way function calculating means for obtaining a session key by performing a one-way function calculation on the output of the combining means, and inputting the content data in which the watermark data is embedded, and using the session key to A watermark data reading device comprising: watermark reading means for reading watermark data from content data in which watermark data is embedded. 4. A method for extracting feature data from content data in which watermark data is embedded, reflecting the feature data on a common key commonly used for all content data when embedding watermark data, and A watermark data reading method, wherein a session key is obtained by performing an operation using a direction function, and the watermark data is read from content data in which the watermark data is embedded using the session key.