JP7162216B2 - Digital watermarking device and method - Google Patents

Description

The present invention is an electronic watermarking method that embeds watermark information such as copyright information in digital image data for the purpose of copyright protection and forgery prevention. The present invention relates to a digital watermarking apparatus and method.

In today's digital information society, information duplication has made it possible for many people to share information, and society has developed significantly. However, this convenience has led to incidents such as copyright infringement due to the illegal duplication and distribution of other people's works. In the case of images, the high image quality of digital cameras and printers in recent years has made it possible to easily obtain exact reproductions of original images. It is a result of encouraging vicious criminal acts called acts.

In such a situation, invisible digital watermark technology embeds other information, such as author information, in image information to protect copyright or warn against illegal duplication. has developed. For example, for the purpose of copyright protection, this digital watermarking technology not only embeds the copyright owner, URL, contact information, terms of use, etc. , and in case of unauthorized use, it is possible to trace it. In this way, electronic watermarks are beginning to be widely used as security measures such as copyright management/protection and unauthorized copying monitoring.

The electronic watermark as a security measure must be robust and cannot be easily lost or removed. This is because if the watermark information can be removed or erased by editing the image or by a simple operation, it can be used illegally and escaped from tracking and surveillance. That is, it is necessary to strengthen resistance to various attacks.

Typically, to enhance robustness, a digital watermark needs to be embedded deep into the image data. Therefore, the embedding strength (gain) is increased for embedding. This increases damage to the original image, resulting in deterioration of image quality. In other words, image quality and durability are in a trade-off relationship, and it is usually difficult to satisfy both.

As a digital watermarking method that satisfies image quality and robustness, we previously proposed a method of embedding a green noise pattern in image data (hereinafter referred to as a green noise diffusion method), which is shown in Patent Documents 1 to 4. The green noise pattern has its spectrum confined to a specific band fmax to fmin, and its main frequency can be set to a frequency band that is difficult to recognize by human visual characteristics. For this reason, when used for digital watermarking, when the embedded image is observed at a distance of clear vision, the dot pattern exhibits low responsiveness and is difficult to recognize. For this reason, even if the embedding strength (gain) is increased to make editing and processing tougher, artifacts are not felt, and high durability can be maintained while maintaining high image quality.

On the other hand, in terms of the amount of information that can be embedded, increasing the resistance reduces the amount of information that can be embedded. Normally, the invisible electronic watermarking method cannot avoid the occurrence of erroneous determinations at the time of extraction, that is, extraction errors. For this reason, the same information is repeatedly embedded, and information with a majority vote or high reliability is selected, or an error correction code is embedded at the same time to avoid this. These methods increase the amount of embedded information. Therefore, watermark information to be originally embedded is reduced.

In such a green noise diffusion method, an embedding operation is performed by dividing an image into R×R blocks and superimposing a green noise dot pattern corresponding to embedding bits on the image data. The embedded capacity is determined by the number of blocks. Therefore, when the image size is small, the number of embedding blocks is small and the embedding capacity is insufficient. Therefore, even if the image quality and durability are satisfied, the embedding capacity is not sufficient.

As a countermeasure for this, there is a so-called multi-value method in which a plurality of bits of embedded information are grouped together and embedded using a plurality of patterns. For example, in the case of four values, since 2-bit information is embedded in each block, it is necessary to prepare four different dot patterns corresponding to 2-bit information (00, 01, 10, 11).

For this reason, it is possible to use dot patterns with multiple spectral shapes that change the maximum frequency fmax and minimum frequency fmin of the green noise band, combine completely different dot patterns with rectangular or circular spectra, blue noise patterns, etc. We proposed a method of preparing four patterns by combining However, with these methods, the spatial frequency band changes for each block, so if the gain is increased to strengthen the tolerance, the boundaries of the blocks will be slightly visible (this is called block distortion), resulting in poor image quality. is not enough.

JP 2018-182471 JP 2018-201134 JP 2018-207247 Patent application 2017-117460

N.Kawamura;”Digital Halftoning by Clustered Dither Screen Masks Representing Green-noise Characteristics”, Proceeding of the 1st ICAI 2015, T109-3, pp.708-711 (2015). N.Kawamura;”Digital Halftoning by Stochastic Dither Screen Mask Representing Green to Blue Noise Characteristics”, Trans. on IEVC, Vol.5, No.1, pp.34-41 (2017)

Therefore, it is a problem to obtain a digital watermarking method that maintains high image quality and high durability and increases the embedding capacity. In other words, it is a digital watermark that does not cause noticeable block distortion even if it is strongly embedded, and has little deterioration in image quality. The challenge is to obtain a digital watermarking method.

In order to solve the above problems, the present invention divides an image into blocks and embeds 2-bit watermark information in each block, i.e., multi-value (quaternarization in this case), thereby doubling the conventional image. To solve the above problem with a dot pattern that secures an embedding capacity of , and does not cause deterioration of image quality such as distortion of each block. In other words, by using dot patterns with the same spatial frequency band (range) for each block and different dot patterns only in different directions, we provide a method that doubles the amount of embedding with high image quality and high durability.

That is, the green noise spectral characteristics of the same elliptical ring shape as the watermark data obtained by developing the watermark information embedded in the digital image data into a bit string, with the main axis directions oriented in the directions of 0°, 45°, 90°, and 135° 4 dot patterns (pattern 0, pattern 1, pattern 2, pattern 3) are prepared, and embedding divides the image data into blocks of R pixel × R pixel units, and 2-bit data is extracted from the watermark data for each block. The information is extracted and embedded in the image data in a pattern obtained by multiplying the dot pattern corresponding to the 2-bit information (00, 01, 10, 11) by gain. After dividing, a spectral pattern is obtained for each block, and the discriminator discriminates the closest dot pattern from among the four dot patterns based on its shape, and extracts the embedded 2-bit information.

The four dot patterns (pattern 0, pattern 1, pattern 2, pattern 3) are four elliptical ring-shaped filters D(u, using v),
(Step 1) Random dots p(x, y) are generated by a pseudo-random number generator.
(Step 2) A two-dimensional Fourier transform is performed on p(x,y) to obtain a spectrum P(u,v). (where (x,y) are spatial coordinates and (u,v) are spatial frequency coordinates.)
(Step 3) Multiply P(u,v) by filter D(u,v) to obtain new P'(u,v).
(Step 4) Perform an inverse Fourier transform on P'(u, v) to obtain a multivalued point profile p'(x, y).
(Step 5) Obtain an error function e(x, y)=p'(x, y)-p(x, y), and perform white and black inversion in descending order of error at each pixel position.
(Step 6) Steps 2 to 5 are repeated until the error is within the allowable amount to obtain a dot pattern.
, whose spectral characteristics exhibit elliptical ring-shaped green noise characteristics with the principal axis directions oriented at 0°, 45°, 90°, and 135°, respectively.

Also, two of the four dot patterns (pattern 0, pattern 1) are obtained by using two elliptical ring-shaped filters D (u, v) whose main axis directions are in the directions of 0 ° and 45 °, The remaining two dot patterns (pattern 2 and pattern 3) can also be generated by rotating the generated pattern 0 and pattern 1 by 90°. Each spectral characteristic exhibits green noise characteristics.

The discriminator uses four elliptical ring-shaped extraction masks (mask 0, mask 1, mask 2, and mask 3) whose main axis directions are 0°, 45°, 90°, and 135°. , the output Qi (i=0,1,2,3) for the image spectral pattern W(u, v) for each block,
Qi=(1/Z) ΣMi(u,v)・W(u,v)
Z=ΣMi(u,v)
(where Mi(u,v) is the distribution of mask i (i=0,1,2,3)), and among such outputs Qi (i=0,1,2,3), the most Find the largest output Qmax and the next largest output Qnext, extract the two embedded bits from the embedding pattern corresponding to Qmax, and obtain the reliability from Qmax- Qnext .

According to the present invention, it is possible to suppress block distortion, maintain high image quality even when the gain is increased, and increase the amount of watermark embedding information with high durability. Therefore, it is possible to embed a sufficient amount of information even in a small-sized image. The suppression of blockiness is due to the fact that the spectral characteristics of the four dot patterns take values between fmax and fmin along the major axis, although in different directions, so that the dot size does not change.

Also, the original image can be restored with a key. This saves the purchaser of the image the trouble of having the original image delivered directly from the distributor. High safety. Furthermore, the purchaser can use the key to embed the copyright information as the second author in the edited image.

Also, of the four dot patterns, two dot patterns (pattern 0, pattern 1) are generated by filtering, and the remaining two dot patterns are generated by rotating 90 degrees, so that the key to removal is two Only the pattern is required and the size of the key can be reduced. Convenience increases as the key size becomes smaller.

1 is a diagram of a watermark embedding and extraction apparatus of the present invention; A diagram showing the distribution of watermarked images over the Internet, A diagram showing the flow of embedding dot pattern generation, A diagram showing the rotation direction and embedded bits of the spectrum of the dot pattern, Diagram showing the shape of the filter when generating a dot pattern for embedding A diagram showing the generated dot pattern and its spectrum, Diagram showing the process flow of watermark embedding, Figure showing the processing flow of watermark extraction, A diagram showing the processing flow of pattern identification, A diagram showing a mask pattern, Figures showing embedded images and their spectra, (a) is an embedded image, (b) is a partially enlarged image, and (c) is a spectral image. Block noise evaluation images, (a) β = 1.3, (b) β = 1.5, (c) β = 1.7. A diagram representing the PSNR of the embedded image against gain, 512×512 image with 512-bit embedding, (a) is the embedded image, and (b) is the spectrum image. Evaluation of robustness by JPEG compression (a) Uncompressed watermarked image and its spectrum, (b) JPEG-compressed image with Q=50 and its spectrum, (c) JPEG-compressed with Q=0.2 image and its spectrum. The figure which evaluated printing resistance.

FIG. 1 is a configuration diagram of a digital watermarking system for information embedding and extraction according to the present invention. In the copyright holder's computer 3, copyrighted image data is stored in a data memory 7 such as a hard disk. The image data is image-processed by the image processing program in the program memory 6 using the CPU 11, ROM 4, RAM 5, etc., and is displayed on the monitor 8. FIG. A scanner 1 and a printer 2 are connected to the computer 3 , and the processed image can be output from the printer and the image can be read from the scanner 1 . Such image processing increases the computational load when handling images of a large size, so there are cases where parallel processing is performed and a processing board such as a GPU is included for speeding up.

In order to embed the copyright information in the image, character information such as the author's name, date, URL, etc. is input from the keyboard 9, and at least 16 characters are required to be embedded. This requires 16 bytes (128 bits) assuming that it is filled with ASCII code characters.

FIG. 2 shows one operation mode, showing a processing procedure for internet distribution. The image embedded with the watermark information is distributed via the Internet 12 from the computer of the distributor or copyright holder 16 (13). The prospective purchaser 17 who sees it informs the distributor of his purchase desire (14). After a predetermined procedure, the distributor distributes the private key for watermark embedding and removal (15). Based on the contract with the distributor, the purchaser can remove the electronic watermark embedded in the sent image and edit the image.

The purchaser can embed information such as his/her own work information and URL as a secondary author in the edited image data using this secret key. Images embedded with such watermark information are released to the public, circulated on the market, and used in various ways.

A third party who sees the image can confirm that the image has the copyright by using watermark extraction software. The watermark extracting software is not confidential and can be freely downloaded from, for example, the homepages of distributors and copyright holders and made widely available for anyone to use. Third parties can know the owner of the image, copyright information, contact information, etc. by this, so it can also prevent unauthorized use and warn.

On the other hand, copyright holders and secondary copyright holders can track and monitor image data as copyrighted works by using the watermark. If a third party uses the image illegally, the copyright holder or secondary copyright holder will read the watermark information from the image using monitoring software, confirm that it is their own work, and that the third party will use it without permission. If so, it can be detected.

A key corresponds to each embedded image on a one-to-one basis. For this reason, the copyright holder uses a different key for each image to prevent abuse. This key cannot be used to remove watermarks on other images. Even if the key is lost or the key is obtained through a collusion attack by a third party, the damage will be limited to one image, and the watermark cannot be removed from the other images. At this time, the watermark reading software can be used in common even if the keys are different. Also. Since keys can be issued almost infinitely, the probability of having the same key is extremely low.

A detailed description will be given below along with examples.
Here, the generation of the dot pattern showing the green noise characteristics used in the present invention will be explained with reference to FIG.
Assume that the matrix size is R pixels × R pixels (R=2^m where ^ is a power), and 1-pixel size black dots are embedded in it. Let the area ratio of black dots be g (0 ≤ g ≤ 1: g = 1 is all black, g = 0 is all white). Assuming that the dot pattern at the point (x, y) is p(x, y), the dot pattern showing the green noise characteristics of g=1/2 is obtained as follows.
(Step 1) First, a pseudo-random number generator generates random dots p(x,y) (20).
(Step 2) A two-dimensional Fourier transform is performed on p(x,y) to obtain a spectrum P(u,v) (21). Here, (x, y) represent spatial coordinates, and (u, v) represent spatial frequency coordinates. For g=0.5, generate (R^2)/2 random dots (initial state is white noise) and let p(x,y). At this time, the initial state can be changed by changing the SEED value of the pseudo-random number generator.
(Step 3) Multiply P(u,v) by filter D(u,v) to obtain new P'(u,v) (22). i.e.
P'(u,v)=P(u,v)*D(u,v)
(Step 4) Perform an inverse Fourier transform on P'(u, v) to obtain a multivalued point profile p'(x, y) (23).
(Step 5) Calculate the error function e(x, y)=p'(x, y)-p(x, y) (24), and perform white and black inversion in descending order of error at each pixel position (25) .
(Step 6) The operations from Step 2 to Step 5 are repeated until the error is within the allowable amount, and finally a dot pattern of g=1/2 is obtained.

Here, the setting of the pattern of the filter D(u,v) will be explained. Now let D(u,v) be a circular ring pattern with its maximum radial frequency fmax and minimum radial frequency fmin. The frequency f0 due to the average dot spacing at g=1/2 is
f0＝√g・fn＝√(1/2)・fn
and fmax and fmin are based on f0,
a ≡ (fmax - f0)/fn
b ≡ (fmin - f0)/fn
Define the parameters (a,b) as Here, fn indicates the Nyquist frequency. By changing (a, b), it is possible to obtain dot patterns with different cluster sizes according to visual characteristics.

For watermark embedding, four types of patterns corresponding to embedding 2 bits (00, 01, 10, 11) are required. Therefore, the pattern of the filter D(u,v) is an elliptical ring pattern, and four patterns are rotated by 45 degrees. FIG. 4 schematically shows a filter pattern, where the minimum value of the minor axis of the elliptical ring is fmin, the maximum value is fmax, and the minimum value of the major axis is β·fmin, and the maximum value is β·fmax. Here, β represents magnification. Filter pattern corresponding to 2-bit (00) upright, (01) rotated by 45°, bit (10) rotated by 90°, and (11) rotated by 135° shall correspond to each other.

FIG. 5 shows the shapes of such four filters (Filter 0, Filter 1, Filter 2, Filter 3). In the figure, black areas represent 0 and white areas represent 1. FIG. 6 shows the pattern and spectrum created through the above steps using this filter D(u,v). Each pattern has a size of R×R, pattern 0 corresponds to 2-bit embedded information (00), pattern 1 corresponds to (01), pattern 2 corresponds to (10), and pattern 3 corresponds to (11).

Next, a method of embedding a watermark in image data will be described. Watermark information is embedded in blue (B) data of color image data consisting of red (R), green (G), and blue (B), or it becomes luminance and color difference signals (Y, Cb, Cr). It is conceivable to convert it into a signal and embed it in luminance (Y) data. When watermark information is handled only in electronic data, embedding it in blue (B) is ideal because it is the least visually recognizable. When considering applications such as reading printed images or monitor images with a camera or scanner, it is difficult to accurately separate blue (B) data (yellow when printing), so luminance (Y) data should be used. is good.

Next, a processing procedure will be described along the flow of FIG. First, the embedded information is converted into a bit string (20). For example, if the information to be embedded is "0123456789abcdef" and the ASCII 16 characters are expanded into a bit string, it becomes a 128-bit string wm="0011000000110001...". Let wm(i) be the i-th bit from the first bit.

Next, the image data is divided into blocks of R pixels×R pixels (21). It should be the same size as the dot pattern size. A block size of 64 pixels x 64 pixels enables highly accurate extraction, but if a large amount of information is to be embedded, a block size of 32 pixels x 32 pixels enables embedding of four times as much information.

The embedding work is done in block units. As embedding in the n-th block (23), first, start from n=1 (22). Consecutive 2 bits are extracted from the head of the embedded watermark information, and this is defined as b=(wm(i), wm(i+1)).

The pattern to be embedded is selected (24) according to the value of b. If b=(00), embed using pattern 0 (25a). Similarly, when b=(01), pattern 1 is used for embedding (25b). If b=(10), pattern 2 is used for embedding (25c). If b=(11), pattern 3 is used for embedding (25d).

Such embedding of watermark information is performed in a real space (pixel space). Let img(x,y) be the image data, gain be the embedding strength, and w(x,y) be the watermark embedded image data.
w(x,y)=img(x,y)+gain・pi(x,y) (i=0,1,2,3)-------------(1)
however,
When b=(00) pi=p0 : Pattern 0
When b=(01) pi=p1 : Pattern 1
When b=(10) pi=p2 : Pattern 2
When b=(11) pi=p3 : Pattern 3
becomes. Although pi is a binary value of (1,0), it is set to (1/2,-1/2) to preserve the average luminance. As a result of embedding, the pixel data may exceed the dynamic range, causing overflow or underflow.
img'(x,y)=(1-gain)・img(x,y)+gain/2--------------------(2)
This process is essential for completely removing watermark information, which will be described later, but may be omitted if the removal is unnecessary or if the frequency of overflow or underflow is low.
It is determined whether or not all blocks are completed (26), and if not completed, the process moves to the next block (27).

All embedding bit strings wm(i) are embedded as described above. If the number of embedding blocks is larger than the number of embedding bits, the reliability can be increased by repeatedly embedding. In this case, a delimiter marker indicating the beginning of the string is required.
Since pi is a random dot, block boundaries are inconspicuous. The dot pattern, which exhibits green noise characteristics, is also used as an FM screen in printing. Dispersive dots are excellent in uniformity, so they are visually uniform and do not give the impression of graininess.

Next, extraction of watermark information will be described along the processing flow of FIG. The extraction work is performed in units of blocks by dividing into R×R blocks (30) as in the case of embedding. Starting from n=1 (31), we extract from the nth block. An R×R image is extracted from the n-th block, FFT (Fast Fourier Transform) is performed, and the image is converted into spectral information Wn (32). Subsequently, normalization by histogram equalization is performed (33). This is because the luminance differs for each block, so that the contrast is improved to improve the identification accuracy, and the degree of reliability, which will be described later, is equally evaluated for each block. Subsequently, the pattern is identified as described later (40), and 2-bit data of embedded information is extracted from the identification pattern (34). It is judged whether or not all the blocks have ended (35), and if not, n is set to n+1 (36), and the next block is started. When finished, all bits are taken out to restore the original character information (37).

FIG. 9 shows the processing flow of pattern identification processing (40). Spectral information Wn in the block is matched with a prepared mask pattern to obtain its output Qi (41). The mask pattern is the same as the filter pattern, as shown in Fig. 10. There are four masks, mask 0 (M0), mask 1 (M1), mask 2 (M2), and mask 3 (M3) rotated by 45 degrees. .
Returning to FIG. 9, with Mi as the i-th mask, the output Qi of matching with Wn is
Qi=(1/Zi)ΣMi・Wn (i=0,1,2,3)-----------(3)
However, Zi = ΣMi
(42). However, Σ is performed for each pixel in the block (corresponding to spatial frequency in this case). As a result, outputs (Q0, Q1, Q2, Q3) corresponding to each mask are obtained. Letting the largest of these four outputs be Qmax,
When Qmax = Q0, b = 00
When Qmax = Q1, b = 01
When Qmax = Q2, b = 10
When Qmax = Q3, b = 11
2-bit watermark information b is obtained (43 and 44a, 44b, 44c, 44d).

Next, if the next largest output after Qmax is Qnext, the reliability rel is
rel=Qmax-Qnext--------------------------(4)
is represented by That is, when Qnext is a value very close to Qmax, the probability of misjudgment is high, and the greater the difference between Qmax and Qnext, the higher the reliability. This difference can be used as an indicator of reliability.
As described above, the 2-bit embedded information b and the reliability rel are obtained.

FIG. 11 shows the image and its spectrum when actually embedded in the image. (a) is an embedded image, in which 16 characters (128 bits) of “0123456789abcdef” are embedded in a 512×512 Lena standard image with a block size of 64×64 with a gain of 0.1875. The embedded pattern is difficult to see at a distance of clear vision. (b) is an enlarged view of part of the embedded image (rectangular portion indicated by A), and the embedded dot pattern can be identified by enlarging it. (c) is the extracted spectrum, and 16 characters were extracted without error. The average reliability (the average of the reliability of each block) is 16.62, and the minimum reliability (the lowest reliability of each block) is 12.19. This indicates that the watermark information could be extracted with very high precision.

FIG. 12 is a result of evaluating block distortion by intentionally embedding a 256×256 uniform image with a strong gain. The original image is a uniform gray image with data of 200/255. , and the four characters "0123" are embedded. (b) shows the embedding pattern and spectrum for β=1.5 and (c) β=1.7. From the embedded images, it can be seen that block distortion is not very discernible in any image, but as β (ellipticity) increases, the anisotropy of the pattern increases and becomes slightly noticeable. On the other hand, if the ellipticity β is small, the reliability decreases (although the block distortion decreases). It can be said that β=1.5 is optimal from the viewpoint of block distortion and reliability.
In normal embedding, the gain is about 0.1 to 0.3, so it can be said that block distortion is hardly conspicuous.

The embedded image quality can be evaluated by PSNR (Peak Signal to Noise Ratio). If the dynamic range of the embedded image is DR, then PSNR is, if the image size is W×H,
PSNR=10 Log (DR^2/MSE)
MSE=1/(WH) Σ(img(x,y)-w(x,y))^2
Therefore,
PSNR=-20 Log (gain/2)------------------------(5)
, which is a gain-only function. where ^ represents exponentiation and Log represents common logarithm. FIG. 13 shows PSNR values for gain.

On the other hand, when the embedded image is viewed at a distance of clear vision, the high spatial frequency region is lowered by the VTF (Visual Transfer Function) of the visual system. It was found that when viewing the image at a distance of clear vision, the PSNR was improved by 5-7 dB due to the effects of VTF. Therefore, if the embedding is performed with a gain of 0.1875 or less, it becomes visually 30 dB or more, and the image quality deterioration is not felt.

Since watermark embedding adds data of ±gain/2 to all pixels, there is no spatial variation in PSNR depending on the position of the image, and large local degradation of image quality does not occur. In a normal digital watermarking method, some parts of the image (for example, edge areas) are embedded and some areas are not embedded, resulting in a high PSNR value. However, large image quality deterioration occurs locally, and as a result, the embedded image quality lacks uniformity and appears as graininess. On the other hand, in this method, since the entire image is uniformly embedded, graininess is not perceived. Also, the image quality is predicted by the gain, and it can be said that it is easy to operate.

Assuming that the image size is W×H, the amount of embedded information V is
V = 2 x int(W/R) x int(H/R) bits. Here, int() indicates truncation below the decimal point. For example, if a 540×530 image is embedded in R=64 blocks, V=128 bits and 16 characters can be embedded in ASCII character code. 64 characters can be embedded if it is embedded in a 32×32 block. When used for copyright protection purposes, at least 16 characters or more are required. Therefore, when embedding in a block size of 64 x 64, the image size is 512 x 512 or more, and when embedding in a block of 32 x 32, it is 256 x 256. Image size or larger is required.

FIG. 14 assumes normal use of the Internet, and (a) is an image when a 32×32 block size is embedded in a 512×512 image with a gain of 0.1875. Up to 64 characters (512 bits) can be embedded, and information such as the author's name, URL, and date can be repeatedly embedded, and error correction codes can be embedded. It is embedded with 16 characters of "0123456789abcdef", which is repeated four times. (b) shows its spectrum. As the embedded information, information with high reliability in repeated embedding is selected, and correct answers are extracted with high reliability.

Next, watermark removal will be described.
Removal of the watermark is possible by doing exactly the opposite of embedding. That is, from equation (1),
img(x,y)=w(x,y)-gain・pi(x,y) (i=0,1,2,3) ------------------ -----(6)
By subtracting the embedding pattern from the watermark-embedded image by multiplying the gain, it is possible to completely restore the original image. However, if linear dynamic range correction is performed to avoid overflow or underflow during embedding, inverse correction must be performed.
Watermark removal is possible as described above, but embedding patterns (pattern 0, pattern 1, pattern 2, pattern 3), gain, and embedding information are required. These pieces of information are exchanged as a "private key".

For example, when the block size is 64×64, the capacity of the private key is 16 Kbits (2 Kbytes) for four patterns from pattern 0 to pattern 3, plus the capacity of gain and embedded information. In the case of 32x32, four patterns from pattern 0 to pattern 3 are 4K bits (512 bytes). Since the gain and embedded information are usually 100 bytes or less, the embedded pattern information occupies most of the key capacity.

It is easier to handle if the capacity of such a key is made as small as possible. To reduce the key capacity, patterns 2 and 3 are generated from pattern 0 and pattern 1, respectively. That is, pattern 2 is obtained by rotating pattern 0 by 90°, and pattern 3 is obtained by rotating pattern 1 by 90°. Since the rotation of the dot pattern is the same as the rotation of the spectrum, the various processes may be exactly the same.

For loss or theft of a key, using a different dot pattern for each image makes it impossible to remove the watermark of another image with the same key. These are obtained by changing the seed value of the initial random number when generating the green noise pattern. The spectral distribution is the same, the extraction software is common, but the dot profile is different. In the case of R = 64, there are 4096 C 2048 different patterns per dot pattern, and when using multiple dot patterns, the product is the product, so the probability of having the same pattern is extremely low. high. Here, C represents a combination.
The copyright holder provides each purchaser with a dot pattern obtained from different random numbers as a secret key. Since this key is different for each purchaser, it is not possible to remove the watermark from another purchaser's image.

Next, the robustness of this method will be explained. Attacks on digital watermarking include signal enhancement (sharpness adjustment), noise addition, filtering (linear, nonlinear), lossy compression (JPEG, MPEG), and transformation (rotation, scaling).
Figure 15 shows the resistance to JPEG compression. (a) is a 256×256 image with a block size of 64×64, a gain of 0.1875, and four characters of “0123” embedded before compression. Average confidence = 16.51, minimum confidence = 15.00, and 4 characters were answered correctly. (b) shows the embedded image and spectrum when the Q factor (Qulity Factor) of JPEG compression is 50. Average reliability = 12.94, minimum reliability = 12.02, and a correct answer is obtained. In (c), Q=20, average reliability = 6.62, minimum reliability = 2.45. Thus, it can be seen that high resistance to JPEG compression is exhibited.

Since this method discriminates from the spectral distribution of a set of dispersed dots, it can be said that it is generally robust against changes in luminance. Therefore, applications such as extracting watermarks by reading embedded images with a camera or scanner are less susceptible to external light. Normally, image processing involves a lot of processing for luminance, and therefore the watermark embedded image watermark information in this method is difficult to remove, and it can be said that this method has high resistance. Tolerance can be further enhanced by increasing gain.

FIG. 16 evaluates printing durability. A 512×512 Lena image with 16-character watermark information embedded with gain=0.25 is printed out on glossy paper at 175 ppi (pixel per inch), read at 300 dpi (dot per inch) with a scanner, and the watermark is removed. Extracted. The upper row is embedded with a block size of R=64, and the extraction gave a correct answer with an average confidence of 10.90 and a minimum confidence of 3.79. On the other hand, those with R = 32 in the lower row obtained correct answers with an average confidence of 12.00 and a minimum confidence of 6.83. In both cases, the reliability is considerably lowered because of the paper output, but it was confirmed that the watermark can be maintained even on the paper medium. The improvement in reliability at R=32 is the effect of repeated embedding.

The digital watermarking method of the present invention has been described above, but since this method embeds a dot pattern that exhibits green noise characteristics that are difficult to see, it is possible to embed strongly with less visual deterioration in image quality. Strong resistance to attacks such as processing. The embedding capacity is doubled compared to the two-pattern case, and the key information can be used to restore the original image. Useful for various applications with links.

In addition, since the watermark is embedded in the real space in units of blocks, there is no restriction on the size of the image such as a large image, and processing can be performed at a high speed equivalent to that of the dither method. Therefore, it is possible to embed moving images in real time, and it is expected to be applied to surveillance cameras and drive recorders.

The present watermark embedding and extraction method is algorithm-friendly. All confidentiality is encapsulated in the dot pattern, which is stored as the private key. Even if the key is lost or stolen, if a different dot pattern is used for each image, since there is, in principle, one key for one image, another image can be watermarked using that key. cannot be removed.
For this reason, security is extremely high, and there are no problems with disclosure of algorithms, standardization, and disclosure of extraction software.

1 is scanner, 2 is printer, 3 is computer system, 4 is ROM, 5 is RAM, 6 is program memory, 7 is data memory, 8 is monitor, 9 is keyboard, 10 is communication function, 11 is CPU, 12 is Internet, 13 image data distribution, 14 purchase request notification, 15 secret key delivery, 16 copyright holder's computer, and 17 purchaser's computer.

Claims (3)

In the digital watermarking method for embedding watermark information in digital image data,
watermark data obtained by expanding the watermark information to be embedded into a bit string;
Four dot patterns (pattern 0, pattern 1, pattern 2, pattern 3) exhibiting green noise spectral characteristics with the same elliptical ring shape and with their principal axes oriented in the directions of 0°, 45°, 90°, and 135°. prepare,
For embedding, the image data is divided into R pixel × R pixel blocks, 2-bit information is extracted from the watermark data for each block, and a dot pattern corresponding to the 2-bit information (00, 01, 10, 11) is created. Embed in the image data with the pattern multiplied by gain,
The watermark information is extracted by dividing the watermarked image data into blocks, obtaining a spectrum pattern for each block, and using a discriminator to discriminate the closest dot pattern from among the four dot patterns. It extracts 2 bits of information ,
The four dot patterns (pattern 0, pattern 1, pattern 2, pattern 3) are four elliptical ring-shaped filters D (u, v )Using,
(Step 1) Random dots p(x, y) are generated by a pseudo-random number generator.
(Step 2) A two-dimensional Fourier transform is performed on p(x,y) to obtain a spectrum P(u,v). (where (x,y) are spatial coordinates and (u,v) are spatial frequency coordinates.)
(Step 3) Multiply P(u,v) by filter D(u,v) to obtain new P'(u,v).
(Step 4) Perform an inverse Fourier transform on P'(u, v) to obtain a multivalued point profile p'(x, y).
(Step 5) Obtain an error function e(x, y)=p'(x, y)-p(x, y), and perform white and black inversion in descending order of error at each pixel position.
(Step 6) Steps 2 to 5 are repeated until the error is within the allowable amount to obtain a dot pattern.
generated in a process that
A digital watermarking method , wherein the spectral characteristics thereof exhibit elliptical ring-shaped green noise characteristics with principal axis directions oriented in directions of 0°, 45°, 90°, and 135°, respectively . Of the four dot patterns , two dot patterns (pattern 0, pattern 1) are obtained by using two elliptical ring-shaped filters D (u, v) whose main axis directions are 0 ° and 45 °. ,
(Step 1) Random dots p(x, y) are generated by a pseudo-random number generator.
(Step 2) A two-dimensional Fourier transform is performed on p(x,y) to obtain a spectrum P(u,v). (where (x,y) are spatial coordinates and (u,v) are spatial frequency coordinates.)
(Step 3) Multiply P(u,v) by filter D(u,v) to obtain new P'(u,v).
(Step 4) Perform inverse Fourier transform on P'(u, v) to obtain a multivalued point profile p'(x, y).
(Step 5) Obtain an error function e(x, y)=p'(x, y)-p(x, y), and perform white and black inversion in descending order of error at each pixel position.
(Step 6) Steps 2 to 5 are repeated until the error is within the allowable amount to obtain a dot pattern.
generated in a process that
The remaining two dot patterns (pattern 2, pattern 3) are generated by rotating the generated pattern 0 and pattern 1 by 90°,
2. A digital watermarking method according to claim 1, wherein each spectral characteristic exhibits a green noise characteristic. The discriminator uses four elliptical ring-shaped extraction masks (mask 0, mask 1, mask 2, mask 3) whose main axis directions are 0 °, 45 °, 90 °, and 135 °,
Output Qi (i = 0, 1, 2, 3) for the image spectral pattern W(u, v) for each block is
Qi=(1/Z) ΣMi(u,v)・W(u,v)
Z=ΣMi(u,v)
(where Mi(u,v) is the distribution of mask i (i=0,1,2,3)),
Among these outputs Qi (i = 0, 1, 2, 3), the largest value output Qmax and the next largest output Qnext are obtained, and the embedded 2 bits are extracted from the embedding pattern corresponding to Qmax,
3. A digital watermarking method according to claim 1 or claim 2 , wherein the reliability is obtained from Qmax -Qnext .

Images