JPH1056557A - Signature method and decoding method to gray level image - Google Patents

Abstract

PROBLEM TO BE SOLVED: To embed signature picture to a gray level image giving visually almost no change and to easily decode the signature picture in signing and decoding methods to the gray level image. SOLUTION: A digital original image whose each pixel has the gradation of 2<m> is divided into a plurality of blocks which comprise the prescribed number of pixels vertically and horizontally, and one of 0 to (p-1) is calculated by a hash function of modulo p (p<=m) in accordance with the coordinate position of the block to divided each block separately. Bit plane of weight which corresponds to a calculated value is fetched among m pieces of bit planes which correspond to the same weight of a plurality of each pixel that comprises the block, the value of one pixel in a signature picture which corresponding to the block of the original image is embedded in bits that comprise the bit plane, and a signed image is acquired by embedding each pixel of the signature picture in each block of the original image.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a method for signing and decoding a grayscale image. In recent years, various types of image data are used by many people via a network, and users of the network can easily acquire and copy these images, so that there is little awareness of rights as copyrighted works of others.
Similarly, digital images recorded on a medium such as a CDROM can be easily copied, so copyright infringement is a problem, and development of an effective protection method is desired.

[0002]

2. Description of the Related Art Images transmitted on a network and C
A so-called multimedia (or electronic media) image such as an image recorded on a recording medium such as a DROM can be easily obtained by a general user, and it is easy to copy the same original as the original. It is very likely that the copyright of the image will be violated.

For this reason, in recent years, various proposals have been made as research for protecting the copyright of multimedia, including the following reports. "Construction Method of Document Image Data with Copyright Protection Function", Hidehisa Matsutani, 3 IEICE Technical Report, ISEC94-58, PP59-68,
March 1995 "Proposal of System Structure for Copyright Processing", Kitamura, Jisho Kenho, Vol.95.NO.37, AVM-8, PP.129-134, March
1995 "Digital Image Copyright Protection System"
SCIS93-13C, Jan. 1995 In each of the above-mentioned conventional proposals, in each case, a matter relating to copyright is embedded in the original image in advance so that the system can constantly monitor.

[0004] As a method of embedding a matter indicating copyright in an image, a method using brightness, a method using darkness, a method using change, a method using edge, and a method using data compression. For orthogonal transformation (such as Fourier transform) for the purpose.

[0005]

In the above-mentioned conventional method, since the signature information is embedded in the image, there is a problem that the information becomes noise and the image quality is deteriorated. Therefore, as a measure for preventing deterioration, a method of limiting an embedding position to a portion having little visual influence has been adopted.

However, limiting the embedding position is
There is a problem that it is relatively easy to detect embedded information by a third party who does not have a copyright, and there is a high possibility that the embedded information is attacked by removing the embedded information and copying the original image.

Similarly, there is no assurance that the conventional method as a signature method for authentication can withstand an attack from a third party who attempts to delete or falsify the embedded information.

SUMMARY OF THE INVENTION It is an object of the present invention to provide a signature method and a decryption method for a gray-scale image in which a signature image can be embedded with almost no visual change.

[0009]

According to the present invention, a digital original image in which each pixel has a gradation of 2 ^m is divided vertically and horizontally into a plurality of blocks each having a predetermined number of pixels. For each, a hash function of modulo p (p ≦ m) corresponding to the coordinate position of the block
(P-1) is calculated and m is calculated from m bit planes composed of values corresponding to the same weight of 2 ⁰ to 2 ^m−1 of a plurality of pixels constituting the block. The value of the pixel corresponding to the block in the signature image corresponding to the block of the original image is embedded in the value of the bit plane, and the value of the pixel at the position corresponding to the block is embedded in the value of the bit plane. To update the corresponding plane of the original image and embed the signature image.

FIG. 1 shows the principle configuration of signature image embedding of the present invention, and FIG. 2 shows the principle configuration of signature image decoding of the present invention. In FIG. 1, 1 is an original image blocking unit, 2 is a block corresponding position determining unit, 3 is a signature pixel selecting unit, 4 is a block corresponding signature pixel value embedding unit, 10 is original image data, 11 is block data, and 12 is block data. Signature image data,
Reference numeral 13 denotes signed image data.

The original image data 10 in which the signature is to be embedded is n
× consist n pixels, each pixel weights 2 ⁰ representing a 2 ^m gradation
It consists of 2 ^m-1 m bits. First, assuming that the original image data 10 is G by the original image blocking means 1,
The whole is divided into blocks of q × q pixels (q ≧ 3). Each of the divided block data 11 is assumed to be g _ij (i and j are numerical values representing the positions of the blocks in the original image, and are incremented by +1 for each q pixels). Here, q × of the block g _ij
q number of 2 ⁰ to 2 m pieces of bit planes corresponding to each weight of the ^m-1 g _ij of the pixel (k) (k = 0,1 ... , m-1) and when, g _ij (k) is It is a collection of q × q bits. Here, in the block-corresponding position deciding means 2, the dispersion of the output is uniform and the embedding position of the signature can be uniquely determined.
Using a predetermined hash function as a position determination function f shown in the following equation (1), substituting i and j, calculating modulo p (p ≦ m), and embedding bit plane k ′ Ask for.

K ′ = f (g_ij) (K '= 0, 1,..., P-1) (1) The signature, face photograph,
Binary or dark proof of copyright such as company mark
Signature image to embed using an image expressed in light (multi-valued)
The size of the image is (n / q) x (n / q) pixels, and the signature image
The element selection means 3 matches the position of the block to be the target of the original image.
One pixel (s) of the corresponding signature image data 12 _ijAnd i,
j is the coordinates of the signature image, and the block g of the original image_ijI, j
Corresponding). Next, block signature pixel value is filled.
Embedding means 4 stores the value of the selected signature pixel in the block
It is embedded in a bit plane k 'for embedding. Sand
If the signature image data 12 is S, each pixel s of S_ij
, A bit plane k ′ for embedding a block corresponding to
Embedded in q × q bits. At this time,
I will embed it in bits other than the center bit of lane q × q
You can do it.

In this manner, the operations of the block corresponding position determining means 2 and the block corresponding signature pixel value embedding means 4 are repeated for all the target blocks,
The signed image data 13 in which the signature is embedded is obtained.

The output value of the position determination function f is k = 0,
,..., P−1, but if the bit on the bit plane is changed, the pixel value in the same block will be up to 2 ^p−1
The gradation may fluctuate, and it is desirable that k ′ is small. On the other hand, the lower bits are easily targeted for an attack (such as an act of erasing a signature) and are not suitable as an embedding position. However, in the present invention, since this position fluctuates in accordance with a block,
You can fully oppose the attack.

Next, in the configuration shown in FIG. 2, 1a is a signed image blocking means for blocking the signed image data 13 created in FIG. 1, and 2a is a corresponding embedding unit for each block of the signed image. Block corresponding position determining means for determining a position, 3a is a signature pixel value identification means for detecting and storing a signature pixel value, 13 is signed image data as in FIG. 1, 14 is block data, and 15 is decoded. This is signature image data.

In the case of decryption, the signed image data 13 is blocked by the signed image data 13 with the number of pixels (n / q × n / q) having the same size as in the case 1 in FIG. . For each block data 14, the block corresponding position determining means 2a calculates a position determining function f using the same hash function as 2 in FIG. 1 to obtain a bit plane k 'for embedding. Next, the data of the bit plane corresponding to k 'is extracted from the target block by the signature pixel value identification means 3a, and the value of the embedded pixel is identified. The identified pixels are stored in the signature image data 15. By performing this decoding operation on each block, all the signature pixels are decoded into the signature image data 15.

[0017]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS FIG. 3 is a processing flow in the case where the original image is a grayscale image and the signature image is binary. In this embodiment, pictures original image having 256 (= 2 ⁸⁾ grayscale, the image of the painting, etc., composed of eight bits corresponding to each weight of each pixel 2 ^0, 2 ^1, 2 ² ... 2 ⁷ is the signature image is (0 or 1 only the weight of 2 ⁰⁾ image binary denote the graphic sign or binary. When the original image is divided into blocks, 1
One block has 3 × 3 = 9 pixels (when q = 3 in the above-described principle configuration of FIG. 1).

In the embodiment in which the binary signature image is signed, the binary image corresponding to the binary pixels "1" and "0" in the selected bit plane in each block of the original image. , "1" is set to an odd number.

The following processing can be executed in an information processing apparatus (not shown) including a CPU, a memory, an image reader, and the like. First, input the original image (S in FIG. 3).
1) Next, when a signature image is input (S2), each image is stored in each area of the memory. In this case, the number of pixels of the original image G is n × n and 2 ^m gradation, the signature image S to be embedded has a length of 1/3 of the length and width of the original image G, and the number of pixels is (n / 3). ) × (n / 3) pixels 2 ⁰ gradation (2
Value), and the pixel is represented by s _ij . Each pixel s _ij of the signature image S is “0” or “1” and i, j = 0, 1,
..., n / 3.

Next, the original image is divided into blocks (S
3). In this embodiment, 3 × 3 pixels form one block,
The 3 × 3 pixels of each block are represented by g _ij , and i, j are 0, 1,..., N / 3 similarly to i, j of the signature image.

Assuming that the coordinates of the original image are i and j, first i
= 0 and j = 0 (S4 in FIG. 3), and k ′ is obtained by calculating the hash function f for the first block of the original image (S5 in FIG. 3).

As the hash function f, various functions can be used. To be specific, the following equation (2),
The function represented by (3) generates values that are simple but vary, and one of them is used in the embodiment. However,
It is possible to use various functions using i and j other than (2) and (3).

F (g _ij ) = i + j (modulo p) (2) f (g _ij ) = i × j (modulo p) (3) where p = 1, 2,..., M, and p = m Then,
Any one of 0, 1, 2,..., m−1 is obtained as a calculation result of f (g _ij ), and m = 8 in the case of 256 gradations.

A bit plane corresponding to the obtained k 'is obtained (S6 in FIG. 3). FIG. 4 is an explanatory diagram of blocks and bit planes. FIG. 4A shows the original image as 3 ×
One block when it is divided into three pixels (a total of nine pixels) is represented by _gij . Is composed of 8-bit data corresponding to each weight of 2 ^0, 2 ^1, 2 ² ... 2 ⁷ each is nine pixels which constitute the block represents one tone in 256 gradations, each weight Has a value of “1” or “0”. A 3 × 3 (bit) plane in the horizontal direction is a bit plane corresponding to the same weight in this block, and the bit plane is 8 bits in the vertical direction.
A indicates that the sheets are stacked. The upper bit plane corresponds to the MSB (the most significant digit 2 ⁷ ),
The lower bit plane corresponds to the LSB (the least significant digit 2 ⁰ ).

When the value of k ′ corresponding to the block (including the coordinates g _ij ) is obtained by the above calculation of the hash function, B. in FIG. The corresponding bit plane is extracted as shown in FIG. 7 and is indicated as g _ij (k ′). Each bit data of this bit plane is represented by gbit _t ,
When t represents one (position) of the nine bit data in the bit plane, gbit _t = 0 or 1, t = 0,
1, ... 8.

Returning to the processing flow of FIG. 3, when the bit plane is obtained as shown in FIG. 4B, embedding of the signature pixel is started. First, the variables z and t are set to 0 (S7). z is a variable representing a count value of the number of “1” (or “0” in the following description but “1”) in the bit plane, and t is a total number of bit data in the bit plane. This is a variable that represents a numerical value.

[0027] Repeat the operation of step S8 to S10, in the case of each bit in the bit plane (Gbit _t) for "1" and the count of +1, count value to the variable z is obtained. When the counting is completed, it is determined whether the signature image s _ij (binary) at the position corresponding to the block g _ij of the original image is “1” (S11 in FIG. 3). If "1", to determine whether the count value of z is odd (the S12), all bits Gbit _t if even in intact, any if odd 1
One of the bit (gbit _t) to reverse the (same S13). If s _ij is “0”, it is determined whether or not the count value of z is an even number (S14), and if it is even, an arbitrary bit (gbit
_t ) is inverted (S15) to be an odd number.

By the processing of S11 to S15, the discrimination corresponding to any of the four cases (1) to (4) shown as the signature pattern shown on the right side of FIG. 3 is performed. bit (gbit _t)
Is inverted. Therefore, when the bit of the signature image is “1”, the number of “1” in the corresponding bit plane of the original image is set to an odd number, and when the bit is “0”, the number is set to an odd number.

FIG. 5 is an example showing changes in a signature pixel and a bit plane. In this example, as shown in A, the bit (gbit _t ) in the bit plane g _ij (k ′) is “1”
Is an example of an even number. In this case, when the pixel s _ij of the corresponding signature image is “1”, the contents of the bit plane are kept as shown in B, and when the pixel s _ij is “0”, one “ The bit (gbit _t ) of “1” is inverted, and the total number of “1” becomes 3
(Odd number).

When the embedding of one pixel of the signature image in one block is completed, i and j are sequentially updated in S16 to S19 in FIG. 3 and the processing of S5 and subsequent steps is executed for each block. When each pixel of the signature image is embedded, the embedding of the signature image in the original image ends, and a signed image is obtained.

By decoding a binary signature image from the signed image created by the processing flow shown in FIG. 3, it can be proved that the original image was created by the signer.

FIG. 6 is a processing flow for decoding a binary signed image from a signed image. First, a signed image (n × n pixels, 256 gradations) is input (S1 in FIG. 6), and block division is performed (S2). Note that an area for a signature image to be decrypted is secured in the memory in advance. This block division is
As in the case of embedding the signature image, one block is divided into three blocks.
× 3 image, each block is g _ij , i and j are 0,
1,..., N / 3. Next, i and j are set to 0 (S
3) For the pixel g _ij of the first block, k ′ (embedding bit plane) is determined using the hash function f (S4). The hash function at this time uses the same function as when the signature image is embedded. When the bit plane (3 × 3 bits) obtained by the calculation is obtained, all of the bit planes are calculated in S6 to S9 using the variable z (the count value of the number of “1”) and the variable t (the count value of the number of bits). The number of “1” in the bit (gbit _t ) is counted. Next, it is determined whether or not z representing the count value of "1" is an odd number (S10 in FIG. 6). If it is an odd number, the pixel _sij of the signature image corresponding to the target block is set to "1" (S11). ), In the case of an even number, the pixel _{sij is set} to “0” (S12).

Next, the value of i is updated (S13 in FIG. 6), and
It is determined whether i> n / 3. If i does not exceed n / 3, the process returns to step S4 to repeat the same processing. If i exceeds n / 3, j is updated next (step S14). It is determined whether or not j> n / 3 (j in FIG. 6). If not, the process returns to step S4 to repeat the same processing. If j exceeds n / 3, all of the original images are deleted. When decoding of the block is completed, the decoded image R is synthesized and displayed.

FIG. 7 is a flow chart of a signature process when the original image and the signature image are both grayscale images. In this case, the original image and signature image photo with both 256 (= 2 ⁸⁾ grayscale, painting,
In the image etc., assumed to be composed of eight bits corresponding to each weight of each pixel 2 ^0, 2 ^1, 2 ² ... 2 ^7.

In the embodiment in which the signature image having the gradation is signed on the original image, the principle is that all the bit values of the gradation pixels of the signature image are set in the selected bit plane in each block of the original image. I do.

The following processing is performed by the CPU and the CPU as in FIG.
It can be executed in an information processing device (not shown) including a memory, an image reader, and the like. First, when an original image is input (S1 in FIG. 7) and then a signature image is input (S2 in FIG. 7), each image is stored in each area of the memory. In addition, the number of pixels in the original image G is 2 ⁸ gradations by n × n, vertical signature image S is the original image G to embed, lateral 1
/ 3 long Satoshi, the number of pixels is 2 ⁸ gradation (n / 3) × (n / 3) pixels represent the pixels in _{s ij, i, j = 0,1} ,
..., n / 3.

Next, the original image is divided into blocks (S
3). Also in this embodiment, 1 × 3 × 3 pixels as in FIG.
Each block is represented by g _ij with 3 × 3 pixels in each block.
i, j are 0, 1,..., n in the same manner as i, j of the signature image.
/ 3.

Assuming that the coordinates of the block are i and j, i = 0 and j = 0 first (S4 in FIG. 7), and k 'is obtained by calculating the hash function f for the first block of the original image. The hash function f in this case is determined by i and j using the same function as in FIG. k 'is 0,
It is any one of 1, 2, ..., 7 and the obtained k '
Are obtained from the data of the original image (S5 and S6 in FIG. 7).

The configuration of the bit plane in this case is shown in FIG. 4B as g _ij (k ′). Each bit data of the bit plane is represented by gbit _t , and t is represented by 9 bits in the bit plane. Expressing the position in the bit data,
gbit _t = 0 or 1, t = 0, 1,...

When the bit plane is obtained, signature pixel embedding is executed. In this case, the multi-level data sbit pixel s _ij signature image corresponding to the block of interest
_t (consisting of values corresponding to respective weights of 2 ⁰ , 2 ¹ ,..., 2 ^{m -1} ) (S7 in FIG. 7), and set as t = 0,
Set the sbit _t as Gbit _t bitplane (same S8), and sets the Gbit _t bitplane the 8-bit pixel _ij signature image below t value is updated and the S9~S11 loop. However, since the bit plane has 9 bits, the 8-bit multi-valued data sbit _t of the pixel s _ij is located at an 8-bit position excluding the center bit.
Place.

FIG. 8 shows an example of embedding pixels of a light and dark signature image in a bit plane. In this case, the bit plane g selected from the block of the original image shown in FIG.
respect _ij (k ') 9 one bit (Gbit _t), imparts number indicating the position of each 8-bit ambient excluding the bit center 0-7. On the other hand, assuming that 8-bit data corresponding to each of the weights 0 to m−1 of the pixel s _ij of the signature image corresponding to the target block of the original image has a value shown in FIG. When the values of .about.7 are set at the positions of 0.about.7 shown in A, the bit plane shown in FIG.

When embedding of one pixel of the signature image in one block of one original image is completed, S1 in FIG.
I and j are updated through the processing of 2 to S15 and S5 to S1
Step 1 is repeated, and when the pixels of the corresponding signature image are embedded in all the blocks, a signed image is obtained and the process ends.

In the embodiment of FIG. 7, the signature image also has the same 256 (2 ⁸ ) gradation as the original image, but when the signature image is 2 ^r , 2 ⁰ , 2 ¹ , 2 The pixels of the signature image can be embedded by setting a total of r data of ² ,..., 2 ^{r−1 in} the bit plane.

In this embodiment, a grayscale image has been described. However, a single color grayscale signature image or a color signature image can be embedded in a color original image by the same processing. In that case, the original image is divided into blocks and R,
For the bit planes selected corresponding to the G and B colors, the values of the corresponding pixels of the grayscale signature image are set, and the R, G, and B values of the corresponding pixels of the color signature image are set, respectively. It can be realized by setting.

By decoding a light and shaded signature image from the signed image created by the processing flow shown in FIG. 7, it can be proved that the original image was created by the signer. FIG. 9 is a processing flow for decoding a multi-valued signature image from a signed image.

The signed image is input (S1 in FIG. 9), and 3 ×
3 (S2). First, i = 0, j
= 0 (S3), and k ′ = f (g _ij ) for the target block in the signed image using the hash function f.
Is calculated (S4), and a bit plane corresponding to k ′ is obtained from the target block (S5). After this,
The process of S6 to S9, sets the data of the pixel of the signature image as sbit _t removed sequentially Gbit _t from t = 0 in a bit-plane to m-1 (= 7). This processing corresponds to obtaining the pixel of the signature image shown in FIG. 8B from the bit plane including the signature pixel shown in FIG. 8C.

When a total of 8 bits from 0 to 7 are obtained for one bit plane, it is assumed that the value (s _ij ) of one pixel of the signature image has been obtained (S10 in FIG. 9), and the following S11 to S14 In the above processing, i and j are sequentially updated, and the processing of S4 to S9 is repeatedly executed for the other blocks, and the pixel value (s _ij ) of the signature image corresponding to each block is obtained from all the blocks. The decrypted image (signature image) is synthesized and displayed (S15), and the process ends.

According to the present invention, the standard image database SID
BA "girl", "moon" and "aeria"
l "(256 x 256 pixels,
256-level gradation, a signature image (85 ×
As a result of embedding a gray-scale signature image (85 gradations of 85 × 85 pixels) representing a face photograph of a person (2 gradations of 85 pixels), the experimental results showed that almost no image quality degradation was observed in any case. I got it. FIGS. 10 and 11 show the results of calculating the signal-to-noise ratio (SNR) in order to evaluate the image quality in more detail. Here, SNR is defined by the following equation (4), err (error) is defined by the following equation (5), org _i in equation (5) is the luminance level of the original image, and emb _i Represents the luminance level of the signature image.

[0049]

(Equation 1)

FIG. 10 shows the SN ratio corresponding to p (modulo value of the hash function) in the case of a binary signature image, and FIG. 11 shows the SN ratio corresponding to p in the case of a multi-valued signature image. In FIGS. 10 and 11, (a) is a value for each p (modulo) when f = i + j (modp) in the above equation (2) is used as a hash function f for determining the position of the bit plane for embedding. SN ratio (SNR),
(b) shows the SN ratio (SNR) for each p when f = i × j (modp) in the above equation (3) is used.

According to FIGS. 10 and 11, as p increases, the chance of using the upper bit plane increases, so that the image quality deteriorates. However, when the binary signature data is embedded with p ≦ 7, the SN ratio is 40 dB each. It can be seen that a good screen is held. Further, according to the embedding method of the present invention, there is an advantage that the embedded signature data can be completely decoded.

In order to confirm the security of the present invention, as a result of evaluating the resistance to an attack aimed at deleting or falsifying a signature, the following three types of attacks were performed and the strength was verified. Attack 1: The lower α bits of each pixel are deleted.

Attack 2: The brightness level of each pixel is ± β. Attack 3: Noise of γ or less is added to each pixel. However, an attack that significantly damages the image does not make sense from the user's point of view (the image is not visible at all), so it was excluded.

For attack 1, as a result of deleting 1, 2, 3, and 4 as lower bits α by experiments,
FIG. 12 shows a signature image (binary) decrypted after the attack. According to this, it exerts considerable strength against attack 1.
This is because the embedding position k ′ is randomly assigned to 0 to (p−1) by the hash function f (position determination function). Attack 2 is easily damaged by the luminance correction performed to maintain the image quality in the embedding algorithm. However, if the luminance level β of a pixel is equal to 2 k ′, the signature information of the partial block is used. However, if β is smaller than 2 (k′−1), an attack can be prevented by manipulating the lower bits. This results in deterioration of image quality, and is a trade-off between image quality maintenance and intensity maintenance. The same can be said for attack 3.

[0055]

According to the present invention, it is possible to completely embed signature information for protecting the copyright of an image in a simple manner without depending on the original image while suppressing deterioration of the image quality. it can.

Also, it is difficult to estimate the embedding position of a signed image for various types of attacks for the purpose of deleting or falsifying the image because the embedding position is random corresponding to the block. Cannot be tampered with.

It is possible to easily embed both binary and multi-valued (shaded) signature images in a shading original image.

[Brief description of the drawings]

FIG. 1 is a diagram showing the principle configuration of embedding a signature image according to the present invention.

FIG. 2 is a diagram showing the principle configuration of signature image decoding according to the present invention.

FIG. 3 is a diagram showing a signature processing flow when an original image is a grayscale image and a signature image is binary.

FIG. 4 is an explanatory diagram of a block and a bit plane.

FIG. 5 is an example showing changes in a signature pixel and a bit plane.

FIG. 6 is a diagram showing a processing flow of decoding a binary signed image from a signed image.

FIG. 7 is a flowchart of a signature process when both the original image and the signature image are grayscale images.

FIG. 8 is a diagram illustrating an example of embedding pixels of a light and dark signature image in a bit plane.

FIG. 9 is a diagram showing a processing flow for decoding a multi-valued signature image from a signed image.

FIG. 10 is a diagram showing an SN ratio corresponding to p in the case of a binary signature image.

FIG. 11 is a diagram illustrating an SN ratio corresponding to p in the case of a multi-valued signature image.

FIG. 12 is a diagram illustrating an example of a signature image (binary) decrypted after an attack.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 1 Original image blocking means 2 Block corresponding position determining means 3 Signature pixel selecting means 4 Block corresponding signature pixel value embedding means 10 Original image data 11 Block data 12 Signature image data 13 Signed image data

Claims (7)

[Claims] 1. A digital original image in which each pixel has a gradation of 2 ^m is divided vertically and horizontally into a plurality of blocks each composed of a predetermined number of pixels. Modulo p corresponding to the coordinate position of
One of 0 to (p−1) is calculated by a hash function of (p ≦ m), and 2 ^{0 of} a plurality of pixels constituting the block is calculated.
A bit plane having a weight corresponding to the value calculated by the hash function is extracted from m bit planes composed of values corresponding to the same weight of ~ 2 ^{m-1 and} the extracted bit plane is configured. In the bit
A shaded image obtained by embedding a value of one pixel of the signature image at a position corresponding to the block of the original image and embedding each pixel of the signature image in each block of the original image; How to sign. 2. The method according to claim 1, wherein when the signature image is a binary image, a total even or odd of “1” (or “0”) is obtained for all bits constituting the extracted bit plane. The number of “1” (or “0”) in the bit plane is even (or odd) depending on whether the value of the pixel at the position corresponding to the block of the signature image is “1” or “0”. A signature method for a grayscale image, wherein the value of one arbitrary pixel in a bit plane is changed so as to embed the value of one pixel such that 3. The method according to claim 1, wherein when the signature image is a grayscale image, except for a bit at a center position in the extracted bit plane, each bit corresponding to each peripheral pixel includes: A signature method for a grayscale image, wherein one pixel is embedded by setting all bits corresponding to each weight of one pixel at a position corresponding to the block of the signature image. 4. The hash function according to claim 1, wherein a function of adding or multiplying a numerical value (i, j) representing a position of the block by a modulo p is used as the hash function. How to sign a grayscale image. 5. A signed image signed in accordance with claim 1 is divided vertically and horizontally into a plurality of blocks each having a predetermined number of pixels, and each of the divided blocks corresponds to the position of the block. Then, one of 0 to (p-1) is calculated by the hash function used for embedding the signature pixel, and 20 ⁰ to 2 of each of a plurality of pixels constituting the block are calculated.
A bit plane having a weight corresponding to the value calculated by the hash function is extracted from the block from m bit planes composed of values corresponding to the same weight of ^m-1 from the block, and The method is characterized in that the value of one pixel of the embedded signature image is detected, the pixel value of the signature image at the position corresponding to the block is detected, and the entire image of the corresponding signature image is decoded from all blocks of the signed image. Decoding method of signature image. 6. A signed image signed with a binary signature image according to claim 2 is divided vertically and horizontally into a plurality of blocks each consisting of a predetermined number of pixels, and for each of the divided blocks, A bit plane corresponding to the value calculated by the hash function used for embedding the signature pixel corresponding to the position of the block is extracted from the block, and “1” (or “0”) in the extracted bit plane is extracted. ) Is determined, the pixel of the signature image corresponding to the block is set to “0” (or “1”) if the number is odd, and “1” (or “0”) if the number is even. A method for decoding a signature image, comprising: determining and decoding an entire image of a corresponding binary signature image from all blocks of the signed image. 7. A signature image obtained by signing a gray-scale signature image according to claim 3 is divided vertically and horizontally into a plurality of blocks each having a predetermined number of pixels, and each of the divided blocks is divided into blocks. The bit plane corresponding to the value calculated by the hash function used for embedding the signature pixel corresponding to the position of the extracted bit plane is extracted from the block, and except for the bit at the center position in the extracted bit plane, surrounding bits are extracted. The multiple bits corresponding to each pixel are
A signature image, which is extracted as bits corresponding to each weight of a pixel at a position corresponding to the block of the signature image, and decodes the entire binary signature image from all blocks of the signed image. Decryption method.