JP2000350007A - Digital watermarking method, digital watermarking device, and recording medium - Google Patents

Abstract

(57) [Summary] [PROBLEMS] To extract watermark information even when a watermarked image is enlarged or reduced. SOLUTION: An input image (1) is converted into key information (3)
(4), and watermark information (2) is assigned to each rectangle.
(5). Embedding is performed by quantizing the average value of the pixels in the rectangle. Decoding (6) divides the input image into rectangles based on the key information, calculates an average of pixel values in the rectangle, quantizes the average value, and determines a bit value corresponding to the quantized average value. To extract embedded information.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

TECHNICAL FIELD The present invention relates to a digital watermarking method,
The present invention relates to a digital watermark device and a recording medium storing a digital watermark processing program.

[0002]

2. Description of the Related Art In recent years, digital image data (hereinafter referred to as "images") has been developed due to the development of network infrastructure such as the Internet, expansion of image databases represented by digital libraries, and the spread of digital devices such as personal computers. There is an urgent need to protect and manage the copyright of "image data".

It is also possible to illegally and easily copy a large amount of digital image data obtained, and measures to prevent this are required. Here, the digital watermark is, as shown in FIG. 7, invisiblely embeds copyright information or the like as “watermark information” in digital image data itself, and a copyright owner extracts the information as necessary. This is a technology that allows you to claim your own copyright.

[0004] With this, even if an illegal copy is made unintended by the copyright holder, the copyright can be asserted by extracting information embedded at the time of distribution from the image.

[0005] The main requirements to be satisfied by the digital watermark are as follows. (1) The original image is not disturbed (not visually detected) (2) The watermark information remains even after editing / processing (3) Extraction / rewriting of the watermark information (by a third party) is difficult (4) It is difficult to remove watermark information without significantly damaging the image

[0006]

As a digital watermarking technique, for example, there is an "electronic watermark insertion method" described in JP-A-9-191394. This method is a method of mixing information into an original image by frequency conversion (specifically, Fourier transformation). However, frequency conversion processing generally requires a large amount of processing and is difficult to be put into practical use.

In addition, when the encoded image is enlarged or reduced, the rectangle at the time of encoding is not preserved, so that the embedded information cannot be taken out even if the frequency conversion is performed (that is, decoding is impossible). Therefore,
The watermark information can be intentionally erased by performing enlargement or reduction so as not to damage the image, and does not satisfy the above requirements.

An object of the present invention is to provide a digital watermarking method, a digital watermarking device, and a recording medium on which a digital watermarking processing program is recorded, which can extract watermark information even if a watermarked image is enlarged or reduced, which has been difficult in the past. It is in.

[0009]

According to the present invention, the entire original image is divided into rectangles, and the average value of the pixel values in the rectangle is made to correspond to the embedding information. The characteristic that the average value does not change significantly is used.

[0010]

DESCRIPTION OF THE PREFERRED EMBODIMENTS One embodiment of the present invention will be specifically described below with reference to the drawings.

[0011] As described above, digital image data can be easily digitally copied between various types of media. Is very large and it is very difficult for copyright owners to manage the copyright of digital images.

Therefore, by using the digital watermarking technique of the present invention, it is possible to embed arbitrary data such as a signature of a copyright holder in the image data in advance without damaging the image data in the original image data. By performing decoding on the image data, the embedded signature or the like can be extracted to claim the copyright.

Although it is prohibited by law to take in securities and the like with a digital copier or scanner, at present, there is no means to prevent copying of only banknotes that can be registered in advance. Therefore, when the digital watermarking technology of the present invention is used,
Since arbitrary data (for example, the machine number and date of the copying machine) can be embedded in the image data without being noticed by the user, the digital data can be tracked based on the embedded data. .

Further, various utilization methods such as confidential communication, in which confidential information is embedded as digital watermarks in image data for communication, are considered, and are important and have a wide range of applications.

The present invention provides a digital watermarking method capable of reliably extracting (decoding) embedded information from all types of image data even when an encoded image is enlarged or reduced. As described above, in the conventional method using frequency conversion, embedded information can be extracted by performing frequency conversion on a rectangle having the same position and the same size as during encoding at the time of decoding. However, when the image is enlarged or reduced, the size of the rectangle at the time of encoding becomes unknown, and information about which rectangle should be subjected to the decoding process is lost, and decoding becomes impossible. Therefore, even if the encoded image is enlarged or reduced, the rectangular information at the time of encoding can be obtained, and decoding is possible if the information embedded in the rectangle is represented by a feature that does not change due to the enlargement or reduction of the image. Becomes

In the present invention, the entire original image is divided into rectangles,
In order to make the average value of pixel values in a rectangle correspond to embedded information (for example, a bit value sequence of 0 or 1), the property that the average value of pixel values does not change significantly even if the encoded image is enlarged or reduced. We are using.

According to the present invention, the type of digital image data (2
The present invention can be applied irrespective of values, multi-values, colors, moving images, and the like.

FIG. 1 shows the configuration of an image processing apparatus according to an embodiment of the present invention. The image processing apparatus includes an image input unit 1 for inputting digital image data into the apparatus, a watermark information input unit 2 for inputting watermark information, a key input unit 3 for inputting a key, and a rectangular division unit for dividing an input image into rectangles. 4. Encoding section 5 for embedding watermark information, decoding section 6 for extracting watermark information from a watermarked image, control section 7 for performing various controls, data display section 8 for displaying data such as images, image data It comprises a data storage unit 9 for storing embedded information and processing results, an external information storage device 10 for driving an information storage medium, and a data communication path 11.

FIG. 2 shows a flowchart of the encoding process. The encoding process will be described for each step with reference to FIG. In step 101, digital image data is fetched from the image input unit 1 such as a scanner or the external storage device 10 into the data storage unit 9 in the present apparatus. In step 102, embedded information (watermark information) is input from the watermark information input unit 2. In step 103, key information is input from the key input unit 3.

In step 104, the whole image is divided into rectangles by the rectangle division unit 4. Hereinafter, embodiments of the dividing method will be described using specific examples. The target image is multi-valued (8 bits / pixel = 256 gradations). The number of pixels is 1500 pixels in width and 1000 pixels in height.

Assuming that the number of bits to be embedded is 100 bits, one bit is embedded in one rectangle.
What is necessary is just to divide into * 10 (or 5 * 20 etc.) rectangles. Therefore, when the number of divided rectangles is fixed, the size of the rectangle (the number of pixels) changes depending on the size of the image.

The number of divisions may be appropriately selected according to the number of bits to be embedded. However, this information must be held as common information at the time of embedding and at the time of extraction.

The first method of division is all equal division,
In the above example, the size of one rectangle is the width 1500/10 =
150 pixels and a height of 1000/10 = 100 pixels.

The second method of division is not equal division but division at a predetermined ratio. For example, if 10 ratios are 0.1
1, 0.09, 0.08, 0.09, 0.07, 0.1
3, 0.10, 0.09, 0.11, and 0.13 (the total of these is 1.0). The number of pixels of the width is obtained by multiplying the image width (1500) by this ratio. The same applies to the height. The advantage of not using equal division is that information cannot be extracted without knowing the ratio of this division, and it can be said that security is higher.

Further, in the third method, the ratio is determined using a random number, instead of determining the ratio in advance. Here, a method of dividing into rectangles at a ratio based on key information will be described. First, the number of image divisions is m in the horizontal direction and n in the vertical direction.
Since one bit is embedded in one rectangle, m × n bits can be embedded in all (see FIG. 3).

A random number is initialized based on the key information. Next, the ratio of division into rectangles is expressed in the horizontal direction: {rh0, rh
1,. . . , Rhm-1}, vertical direction: {rv0, rv
1,. . . . rvn-1}, these are random values according to random numbers, and are determined so as to satisfy the following equation. rh0 + rh1 +. . . . rhm-1 = 1 (1) rhi≥W / (2m) for all 0≤i≤m-1 (2) rv0 + rv1 +. . . . . rvn-1 = 1 (3) rvi≥H / (2n) for all 0≤i≤n-1 where W is the number of pixels in the horizontal direction of the image, and H is the number of pixels in the vertical direction. Represents Equations (2) and (4) are to prevent generating very small rectangles.

The size and position of the rectangle can be obtained using the ratio determined in this way. Note that a random number generator that always generates the same random number if the seed (initialized value of the random number) is the same is used for the random number. This random number generator may be a known one. Using this random number, ten ratios as in the above example are generated. However, the sum must be 1.0. You can use the same random number both vertically and horizontally,
Another random number may be used.

As described above, according to the third method, if the key information is used as a value based on the key information, that is, for example, if the key information is used as a value for initializing a random number sequence, the ratio changes each time encoding is performed. Those who do not know can not decode.

In step 105 and subsequent steps, watermark information is embedded. This method embeds watermark information, but there is an average value of pixels in a rectangle as a relatively stable feature amount even when an image is enlarged or reduced. In the present invention, an arbitrary bit value can be obtained by manipulating this average value. Embed

In step 105, the average of the pixel values (luminance value or density value) in the rectangle is calculated. Here, the average of the pixel values is obtained using all the pixels in the rectangle. In step 106, the average of the pixel values obtained in step 105 is quantized. That is, the average value of the pixel values that take any value is changed to take only the quantized value. By associating a bit value 0 or 1 with this quantized value,
Bit value information can be provided.

In step 107, the pixel value is changed so as to become the average value quantized in step 106. In step 108, it is determined whether or not all rectangles have been processed. If all rectangles have been processed, all processing ends. If there is an unprocessed rectangle, the process proceeds to step 105.

The method of embedding the watermark information will be described with reference to FIG. 4. The average value of the pixel values of the rectangle of interest is 11
Suppose that it was 7. When the bit value to be embedded in this rectangle is 0, the pixel value of the entire rectangle is changed so that the average value becomes 120, or when the bit value to be embedded is 1, the average value becomes 110. Thus, the pixel value of the entire rectangle is changed. In order to change the average value by x, all pixels may be changed by x uniformly.

In the above-described method, the average value is obtained by using all the pixels in the rectangle, and the average value is set as the pixel to be operated. However, the present invention can employ another method. For example, the average value is obtained using only predetermined pixels (for example, every other pixel) in a rectangle, and a method of setting the average value as a pixel to be operated or only a pixel determined using a random number in a rectangle is used. The average value may be obtained using the average value, and the pixel may be used as an operation target.

FIG. 5 shows a flowchart of the decoding process. The decoding process will be described for each step with reference to FIG. In step 201, digital image data is fetched from the image input unit 1 or the external storage device 10 into the data storage unit 9 in the present device. In step 202, key information is input from the key input unit 3. In step 203, the entire image is divided into rectangles by the rectangle division unit 4. The method of division is the same as that used for encoding. That is, the method described in step 104 is used.

In step 204, the average of the pixel values in the rectangle is calculated. Here, the average of the pixel values is obtained using all the pixels in the rectangle. In Step 205, Step 2
The average value of the pixel values obtained in step 04 is quantized. Step 20
In step 6, the bit value embedded in the rectangle of interest is determined based on whether the average value quantized in step 205 corresponds to the bit value 0 or 1. In step 207,
The bit value determined in step 206 is stored.

In step 208, it is determined whether or not all rectangles have been processed. If all rectangles have been processed, all processing is terminated. If there is any unprocessed rectangle, the flow advances to step 204.

The present invention can be realized by software. FIG. 6 shows a system configuration example when the present invention is realized by software. The digital watermark processing program of the present invention is recorded on a recording medium such as a CD-ROM, and is installed in the system. The watermark information is embedded in the image read from the scanner or the image stored in the system, and the watermarked image data is output to another medium or transmitted to another device via a network. Also, the watermark information is correctly restored from the watermarked image data taken into the system.

As described above, according to the present invention, the embedded information can be reliably extracted even if the encoded image is enlarged or reduced for any image data. In particular, since the information can be extracted even if the magnification when the encoded image is enlarged or reduced is unknown, the original image is not required when extracting the information, and the versatility is high. Therefore, according to the present invention, when image processing such as enlargement and reduction is conventionally performed, embedded information cannot be extracted and the copyright protection function does not always work. Can be applied to a wider range of applications and fields.

[0039]

As described above, claims 1, 9 and
According to the tenth aspect, since the rectangular division position is determined by the ratio to the entire image, even if the encoded image is enlarged or reduced, the position of the rectangle at the time of encoding is clear, so the embedded information is extracted ( Decoding).

According to the second aspect of the present invention, since the sizes of the rectangles into which the image is divided are all the same, the position of the rectangle can be determined by simple processing.

According to the third aspect of the present invention, since the sizes of the rectangles are different, it is difficult for a person who does not know the division ratio to know the position of the rectangle at the time of decoding even if the decoding is attempted. Are better.

According to the fourth aspect of the invention, the size of the rectangle is determined by the ratio based on the key information input by the user. It is difficult to know the rectangular position, and the confidentiality is more excellent than that of the third aspect.

According to the fifth aspect of the present invention, since the characteristic that the average of the pixel values of the image does not change greatly even if the image is enlarged or reduced is used, even if the encoded image is enlarged or reduced, it is embedded. Information can be retrieved.

According to the sixth aspect of the present invention, since the target pixel for calculating the average of the pixel values is all the pixels in the image, the average value can be stably extracted. That is, the encoded information can be stably extracted. Further, since the process of selecting a pixel is not required, the process can be performed with a small processing amount.

According to the seventh aspect of the present invention, since the target pixel for calculating the average of the pixel values is a predetermined pixel in the image, the information is extracted if the position of the target pixel cannot be known. It is impossible and confidentiality is high.

According to the eighth aspect of the present invention, since the target pixel for calculating the average of the pixel values is a pixel determined based on the key information in the image, the target pixel changes each time encoding is performed. If this key cannot be known, decoding becomes impossible, and the confidentiality is further improved as compared with the invention of claim 7.

[Brief description of the drawings]

FIG. 1 shows a configuration of an embodiment of the present invention.

FIG. 2 shows a flowchart of an encoding process.

FIG. 3 shows an example of dividing an image into rectangles.

FIG. 4 is a diagram illustrating embedding of watermark information.

FIG. 5 shows a flowchart of a decoding process.

FIG. 6 shows a configuration example when the present invention is realized by software.

FIG. 7 is a diagram illustrating the concept of a digital watermark.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 1 Image input part 2 Watermark information input part 3 Key input part 4 Rectangular division part 5 Encoding part 6 Decoding part 7 Control part 8 Data display part 9 Data storage part 10 External information storage device 11 Data communication path

Continued on front page F term (reference) 5B057 CA08 CA12 CA16 CB08 CB12 CB16 CC01 CE08 5C053 FA13 FA27 GB01 GB05 KA01 LA11 LA15 5C076 AA14 AA36 5D044 AB07 BC03 CC04 DE49 DE50 EF05 GK11 GK17 HL06 HL14

Claims (10)

[Claims] 1. A digital watermarking method for embedding watermark information in digital image data and extracting the embedded information, wherein the digital image data is divided into a plurality of rectangles, and each of the plurality of rectangles is divided by using a feature amount in each of the rectangles. A digital watermarking method characterized by embedding watermark information in a rectangle. 2. The digital watermarking method according to claim 1, wherein all of the divided rectangles have the same size. 3. The digital watermarking method according to claim 1, wherein the image data is divided into a rectangular size calculated from a size of the image data and a predetermined ratio. 4. The electronic watermarking method according to claim 3, wherein the ratio is determined based on key information. 5. The electronic watermarking method according to claim 1, wherein the watermark information to be embedded represents an average value of pixel values of target pixels in the rectangle as bit information. 6. The digital watermarking method according to claim 5, wherein the target pixels are all pixels in a rectangle. 7. The digital watermarking method according to claim 5, wherein the target pixel is a predetermined pixel in a rectangle. 8. The digital watermarking method according to claim 5, wherein the target pixel is a pixel determined based on key information. 9. A means for dividing digital image data into a plurality of rectangles, a means for embedding watermark information in each of the rectangles by using an average value of pixel values in the rectangle as a feature amount, and extracting the embedded watermark information. And a digital watermarking device. 10. A function of dividing digital image data into a plurality of rectangles, a function of embedding watermark information in each of the rectangles by using an average value of pixel values in the rectangle as a feature, and extracting the embedded watermark information. A computer-readable recording medium on which a program for causing a computer to realize a function is recorded.