JP2006301334A - Image information security method - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide an image information security method capable of protecting image information (secret image) by forming the image information (secret image) to be protected as a computer hologram, embedding the computer hologram into a dummy image (cover image) to form an embedded image, and reproducing the secret image from the embedded image. <P>SOLUTION: The image information security method comprises preparing the secret image which is a target for protection (a step 101); preparing the cover image to be embedded with the secret image prepared in the step 101 (a step 102); synthesizing the secret image prepared in the step 101 with the computer hologram (a step 103); embedding the computer hologram synthesized in the step 103 in the cover image prepared in the step 102 (a step 104); generating the embedded image (a step 105); and reproducing the secret image from the embedded image generated in the step 105 (a step 106). As a result, the secret image can be taken out and verified. <P>COPYRIGHT: (C)2007,JPO&INPIT

Description

  The present invention relates to an image information security method for protecting image information. In particular, image information to be protected (hereinafter also referred to as “secret image”) is a computer generated hologram, and an embedded image is generated by embedding a computer generated hologram in a dummy image (hereinafter also referred to as “cover image”). The present invention relates to an image information security method capable of protecting image information (secret image) by reproducing a secret image.

  In recent years, an infinite number of multimedia contents have been distributed due to the development of information networks such as the Internet. One of the advantages of digital information such as multimedia content is that it is easy to copy, distribute, and browse, but this also means that acts that ignore copyright such as unauthorized copying and unauthorized distribution can be easily performed. It has some disadvantages. Several methods for protecting digital information have been devised to eliminate such drawbacks.

  One example is information encryption. This is a method in which an information distributor encrypts his / her information and cannot be known by anyone other than the user. However, this technology hinders the widespread transmission of information and sacrifices the advantage of multimedia content, the free distribution of information. Therefore, “digital watermarking” was devised as a means to achieve both free information distribution and copyright protection. Digital watermarking is a technique for secretly embedding copyright information of a producer in a certain content so that the content user cannot perceive it. Since users cannot perceive the information embedded in the content, they can be copied and distributed in the same way as normal content, but the copyright information is hidden there, and if the fraud is done, the producer You can take copyright information and claim your rights. In this way, digital watermarking is considered to be effective as a technique that can simultaneously perform free distribution of multimedia contents and copyright protection.

  In addition, “steganography” is a technique related to digital watermarking. Steganography is a technology in which certain multimedia content is used as dummy information and the true information to be transmitted is secretly embedded. Distributing content with embedded information can prevent unauthorized third party information theft and information leakage.

  Furthermore, in recent years, the opportunity to contact digital media has increased rapidly due to the spread of the Internet. According to statistics from the Ministry of Internal Affairs and Communications in 2002, the Internet population penetration rate is 54.5%. The establishment of such a mutual network has dramatically improved the convenience of information collection, but a number of problems have emerged. In particular, copyright infringement due to illegal duplication of image information is a major problem. Unlike images such as paintings, digital images can be easily duplicated, and you can get exactly the same as the original image. This is an advantage of the digital media that distributes image information widely, but it also causes infringement of copyrights and the like. For this reason, there is a need for a method of protecting copyright while taking advantage of digital images. As one of such image copyright protection methods, a digital watermark technique has been proposed. This is a technology in which a content creator embeds copyright information in the content in such a way that it cannot be perceived by the user. By embedding a digital watermark, even if the information is illegally copied, the copyright information of the producer can be extracted from the copied content, and the right as the producer can be claimed. The development of such digital watermark technology is considered to play a major role in protecting the copyright of images. As a method of embedding a digital watermark, there are a method using pixel replacement, a method of embedding in a frequency domain, a method using wavelet transform, a method using statistics, and the like.

  Also, a method of embedding a computer generated hologram in a grayscale image has been studied. In general, in a grayscale image, one pixel is expressed by an 8-bit gradation value. Therefore, since the redundancy of the gradation value is high and the modification in the lower bits has little visual influence, a relatively large number of methods for embedding the secret image can be considered. However, an image subjected to pseudo halftone processing expresses one pixel with a small number of bits, has very little redundancy, and it is difficult to embed a secret image. In particular, it is very difficult to embed a secret image in an image obtained by quantizing a grayscale image into binary values by pseudo halftone processing. In addition, on electronic media, a secret image embedded in the lower bits of a grayscale image can be extracted by digital image processing. However, in the case of a hard copy made by a printer or the like, a method of extracting only embedded information of lower bits is used. Can not. Several methods have been proposed for embedding a secret image even in such a binary image subjected to pseudo halftone processing. For example, there are a method using error diffusion and a method using a dither method.

Further, for example, Japanese Patent Application Laid-Open No. 2003-58856 describes that a printed material is prevented from being counterfeited using a medium having an IC chip or a hologram.
JP 2003-58856 A

  However, according to the above-described conventional method, the secret image is embedded in a part of the cover image, and there is a problem that the secret image is easily lost when the part is cut off.

  In addition, there is a problem that the amount of information to be embedded becomes small in consideration of the influence on the image quality.

  Further, in Japanese Patent Laid-Open No. 2003-58856, since a medium having an IC chip or a hologram is directly built in a printed material or a card, a specification for incorporating a medium having an IC chip or a hologram at a predetermined position of the printed material or the card. This area is necessary, and the built-in IC chip and hologram can be easily removed.

  Furthermore, in Japanese Patent Laid-Open No. 2003-58856, a plurality of media having IC chips and holograms cannot be inserted into the entire printed matter or card.

  Therefore, an object of the present invention is to use image information (secret image) to be protected as a computer hologram, embed a computer hologram in a dummy image (cover image) to generate an embedded image, and reproduce the secret image from the embedded image. It is another object of the present invention to provide an image information security method capable of protecting image information (secret image).

  In order to solve the above-mentioned problem, an image information security method of the present invention is an image information security method for protecting image information. (A) A secret image to be protected is prepared, and (B) the step (A The cover image for embedding the secret image prepared in step (A) is prepared, (C) the secret image prepared in step (A) is combined with the computer generated hologram, and (D) the computer generated hologram combined in step (C) is processed in the step An embedded image is generated by embedding it in the cover image prepared in (B), and (E) a secret image is reproduced from the embedded image generated in step (D).

  In the step (C), the secret image prepared in the step (A) can be combined with a Lohmann hologram.

  In the step (C), the secret image prepared in the step (A) can be combined with a Lee (Lee) hologram.

  In the step (C), the secret image prepared in the step (A) can be combined with a binary phase hologram.

  In the step (C), the secret image prepared in the step (A) can be combined with an error diffusion hologram.

  Further, in step (D), the computer generated hologram synthesized in step (C) can be embedded in the cover image prepared in step (B) by pixel replacement to generate an embedded image.

  In the step (D), the computer generated hologram synthesized in the step (C) can be embedded in the cover image prepared in the step (B) using a gradation value to generate an embedded image.

  Further, the step (D) divides the cover image prepared in the step (B) into a plurality of areas, detects the complexity of the divided areas, and in areas where the complexity exceeds a predetermined threshold, The computer generated hologram in step (C) is embedded to generate an embedded image, and in step (E), the embedded image generated in step (D) is divided into a plurality of regions, and the complexity of the divided regions is determined. The secret image may be reproduced from an area where the complexity exceeds a predetermined threshold.

  Alternatively, in the step (D), the computer generated hologram synthesized in the step (C) is embedded in the cover image prepared in the step (B) using a systematic dither to generate an embedded image, and the step (E ) Can optically reproduce the secret image from the embedded image generated in the step (D).

  According to the image information security method of the present invention, image information to be protected (secret image) is a computer generated hologram, an embedded image is generated by embedding the computer hologram in a dummy image (cover image), and the secret image is reproduced from the embedded image. By doing so, the image information (secret image) can be protected.

  Further, by making the copyright information (secret image) into the form of a computer generated hologram, it has robustness against a modification operation that is performed by a malicious tamper to erase the copyright information. Also, when it is desired to use 3D information as a watermark, it can be expressed in a 2D form by making it a computer generated hologram.

  In addition, computer holograms are highly redundant, so that information will not be lost if there is a slight deterioration, and even if a secret image is three-dimensional, it can be expressed in a two-dimensional form by using a computer hologram. Since it is impossible to grasp the contents of information without taking the procedure of reproduction, it is excellent in safety from the viewpoint of information leakage, and the shape of the computer hologram itself is complicated, so that it is embedded. Because it is difficult to be recognized, it has the following attack resistance.

<1> Influence of noise mixing Regarding the influence of mixing noise with an embedded image, since the embedded image is a binary image, the influence of noise is not a problem at all.

<2> Influence of JPEG compression Even when JPEG compression is performed, if a binary image can be accurately restored, it is not necessary to consider the influence of JPEG compression on an embedded image.

<3> Influence by Median Filter Since a binary image is targeted, only a binary value can be obtained when the median filter is applied. Therefore, it can be considered in the same manner as when noise is superimposed on a binary image. Therefore, even in the case of the influence of the median filter, the attack resistance is strong.

<4> Effect of cropping attack The purpose of a malicious falsifier erasing confidential information embedded in an image when a cropping attack that performs processing to crop a part of the image is added to the embedded image (cutting attack) Therefore, a part of the embedded image may be cut out. In order to be effective as a digital watermark, it is required that copyright information is not lost even if an image is cut out. In the method of the present invention for embedding a computer generated hologram, the secret information is spread over the entire image, and even if a part is lost, the secret information is not lost. Therefore, the method of embedding a computer generated hologram according to the present invention has excellent resistance to a cutting attack.

  Hereinafter, an embodiment of an image information security method of the present invention will be described with reference to the drawings. In the following, the image information security method of the present invention will be described as a specific example in which it is executed by a general PC (Personal Computer).

  FIG. 1 is a flowchart showing the processing of the image information security method of the present invention. In FIG. 1, the image information security method of the present invention first prepares a secret image to be protected (step 101). At this time, the secret image may be any type of image as long as the image can be read by the PC. The secret image may be read by any method such as downloading by a PC, reading by a scanner, or reading from a storage medium.

  Next, a cover image for embedding the secret image prepared in step 101 is prepared (step 102). This cover image can also be prepared in the same manner as the secret image in step 101.

  Next, the secret image prepared in step 101 is synthesized with a computer generated hologram (step 103). The process of combining with the computer generated hologram in step 103 will be described in detail later.

  Then, the computer generated hologram synthesized in step 103 is embedded in the cover image prepared in step 102 (step 104), and an embedded image is generated (step 105). Thereby, the image which protected the secret image is produced | generated. The processing in steps 104 and 105 will also be described in detail later.

  Finally, the secret image is reproduced from the embedded image generated in step 105 (step 106). Thereby, a secret image can be taken out and verified.

A method for synthesizing the secret image with the Lohmann hologram in step 103 will be described. FIG. 2 is a diagram illustrating a Lohmann cell. A Lohmann hologram is divided into cells of one side Δν around each sample point (m, n) on the computer hologram plane. As shown in FIG. 2, each cell Smn opens a small rectangular window according to the amplitude and phase to be expressed. The width c of the window is an arbitrary value, and its height Vmn is proportional to the amplitude. The shift Pmn between the center of the window and the center of the cell corresponds to the phase. The values of Vmn and Pmn can be determined by the following <Equation 1>.
In the above description, the Lohmann-type computer generated hologram (FIG. 3A) and the reproduced image (FIG. 3B) when the width Δν of each cell is 4 dots and Vmn and Pmn are quantized to 4 values. As shown in FIG. The image in FIG. 3A shows the Lohmann computer hologram completed in step 105, and the image in FIG. 3B shows the reproduced image (secret image) reproduced in step 106.

A method for synthesizing the secret image with the Lee hologram in step 103 will be described. FIG. 4 is a diagram illustrating a Lee type cell. The Lee (Lee) hologram divides the complex amplitude distribution of the hologram into a real part and an imaginary part, and determines the position of the opening depending on whether they are positive or negative. As shown in FIG. 4, a positive real part, a positive imaginary part, a negative real part, and a negative imaginary part are assigned from the left end of the cell. The size of the opening is proportional to each component. If the size of the opening is Omn and the value of the real part or imaginary part is Bmn,
It becomes. Here, FIG. 5 shows a Lee type computer generated hologram (FIG. 5A) and a reproduced image (FIG. 5B) when the width Δν of each cell is 4 dots. The image in FIG. 5A shows a Lee-type computer generated hologram completed in step 105, and the image in FIG. 5B shows a reproduced image (secret image) reproduced in step 106.

A method of synthesizing the secret image with the binary phase hologram in step 103 will be described. In the phase hologram, the amplitude is assumed to be 1, and only the phase is displayed in gradation. This is generally called kinoform. Here, for the sake of simplicity, a computer generated hologram that is phase-quantized into two values is used. The binary phase hologram Hmn is
It shows. The image in FIG. 6A shows the binary phase computer hologram completed in step 105, and the image in FIG. 6B shows the reproduced image (secret image) reproduced in step 106.

A method of synthesizing the secret image with the error diffusion hologram in step 103 will be described. FIG. 7 is a diagram showing the concept of the error diffusion method. An error diffusion method is applied to an error diffusion hologram when quantizing a binary phase hologram into a binary hologram. An error between the binarized value Hmn and the actual value Dmn is defined as Emn. The value is multiplied by the diffusion coefficients a, b, c, and d so that the sum is 1, and added to the surrounding four Dm + 1, n, Dm + 1, n + 1, Dm, n + 1, Dm-1, n + 1. These are again referred to as Dm + 1, n, Dm + 1, n + 1, Dm, n + 1, Dm-1, n + 1. The same operation is repeated in the order of raster scanning. This is expressed as follows.
The image in FIG. 8A shows the error diffusion type computer hologram completed in step 105, and the image in FIG. 8B shows the reproduced image (secret image) reproduced in step 106.

  Next, a method for embedding a computer generated hologram in the cover image prepared in step 102 to generate an embedded image (steps 104 and 105) will be described.

  In the following, a description will be given of embedding a computer generated hologram into a cover image by pixel replacement to generate an embedded image.

  As a procedure for embedding the computer generated hologram in the pixel replacement type, the pixel replacement type digital watermark and steganography method using the computer hologram is almost the same as the case of embedding information by the normal pixel replacement. The method is shown below.

<Embedding procedure>
A secret image to be embedded is prepared (step 101). A computer generated hologram of the image is synthesized (step 103). At this time, the type of the computer generated hologram is arbitrary, and the four types of computer generated holograms described above can be used. The synthesized computer generated hologram is embedded by a pixel replacement method (step 104). That is, the bit plane with the cover image is replaced with a computer generated hologram.

<Removal procedure>
When extracting the secret information, the bit plane information at the embedded position is extracted, which is regarded as a computer generated hologram, and a reproduction procedure is performed (step 106).

  FIG. 9 is a diagram showing image quality evaluation. As shown in FIG. 9 of PSNR, which is an evaluation of image quality, the higher the bit plane to be embedded, the worse the image quality of the image after embedding, and the easier it is to recognize the secret image. The same. However, compared to the normal method in which the letter “A”, which is a secret image when embedded in the upper bit plane, appears on the image, the secret image has been converted into a computer-generated hologram. Is not recognized as being embedded. When comparing PSNR in terms of image quality, there is no significant difference between the method of embedding a computer generated hologram and the normal method. In addition, the effect on the image quality is considered to be about the same depending on the type of computer hologram to be embedded. However, when embedded in the upper bit plane, a characteristic pattern appears in the Lohmann type, Lee type, and binary phase type. A characteristic pattern does not appear in the error diffusion type. This is due to the fact that the error diffusion type is complicated, and a meaningful shape cannot be seen. Therefore, it is better to embed the secret information as a computer generated hologram than in the normal method, and it seems that the error diffusion type is suitable for the computer generated hologram.

  Next, it will be described how to embed a computer generated hologram in a cover image using gradation values to generate an embedded image.

<Embedding procedure>
As the embedding of the computer generated hologram using the gradation value, the computer generated hologram can also be used by the method using the gradation value as in the pixel replacement type. The embedding is the same as the method using normal gradation values, but first, the secret image is made into the form of a computer generated hologram. That is, an embedded image is generated by “secret image computer generated hologram + cover image”.

<Removal procedure>
When taking out the secret image, the cover image is subtracted from the embedded image to “embed image-cover image” to restore the secret image computer generated hologram and reproduce it.

  FIG. 10 is a diagram showing image quality evaluation (PSNR). In FIG. 10, the higher the embedding strength is, the lower the image quality is, no matter which computer hologram is used, the same as in the normal case where no computer hologram is used. However, even if it is embedded with high strength, the secret information does not appear directly on the cover image as in the normal case. This has the advantage that the secret information is converted into the form of a computer generated hologram, so that at first glance it is not possible to know what secret information is embedded. Considering the influence on the image quality due to the difference of the computer hologram to be embedded, the image quality is low in the phase type and error diffusion type. This is because, since there are many white portions that are openings of the computer generated hologram, in principle, the image quality is lower than that of the Lohmann type and the Lee type. However, looking at the pattern appearing in the image, the characteristic pattern of the computer hologram appears on the cover image in the Lohmann type, the Lee type, and the binary phase type more than the pixel replacement type, Get an unnatural impression. On the other hand, in the error diffusion type, the cover image has a slightly rough texture, but because of its high complexity, a characteristic pattern does not appear. Also, when compared with the pixel replacement type, the PSNR is lower in the gradation value change type, but the change between a certain pixel and its neighboring pixels is smooth. Since the pixel replacement type changes the entire bit plane, a certain pixel and its neighboring pixels have uniform gradation values, and smoothness is lost. In this respect, it can be said that the gradation value change type is more suitable for embedding secret information.

  Next, embedding of a computer generated hologram using complexity will be described. As described above, there is no information of a shape that can be recognized by humans for images with higher complexity. Therefore, even if a high complexity image area is replaced with a high complexity image having a different shape, a human cannot recognize it. This can be used to embed information. The pixel replacement type information embedding is realized by replacing a bit plane with information to be embedded as it is. The method using the gradation value is realized by changing the gradation value of the grayscale image according to the watermark information. However, the information that can be embedded by these methods depends on the size of the bit plane of the image, and a lot of information cannot be embedded. Here, a method of embedding information focusing on the complexity α is proposed. That is, the complexity of each bit plane is checked, and an area having a certain complexity that is recognized as being complicated is replaced with secret information. Here, the secret information is converted into a form having a certain level of complexity.

  FIG. 11 is a diagram illustrating the complexity α of each bit plane of an image. FIG. 12 is a diagram illustrating the relationship between the complexity α and the number of pixels that can be embedded. According to this method, information can be embedded if the complexity is high not only in the lowest-order bit plane but also in the higher-order bit plane. By so doing, information can be embedded across all the bit planes, so that the amount of information that can be embedded is increased as compared with pixel replacement type or gradation value change type embedding. In the conventional method using complexity, as described above, the secret information to be embedded must have a certain level of complexity, but an arbitrary image does not necessarily have a certain level of complexity. Therefore, processing such as conjugate calculation is necessary, and when extracting information, the reverse conjugate calculation must be performed, which complicates the operation. Therefore, consider making the embedded information into a hologram. As described above, the computer generated hologram does not need to be reversely operated during the reproduction even if the conjugate calculation is performed. Error diffusion holograms are inherently very complex. Therefore, by keeping the secret information in the form of a computer generated hologram, information can be embedded by exchanging a complicated area of a certain image with a computer generated hologram, so that it is possible to omit reverse conjugate calculation processing and the like. The procedure of a method for embedding a computer generated hologram using complexity is shown below.

<Embedding procedure>
Examine the complexity of each bit plane in the grayscale image. In the following, an example is given in which each bit plane is divided into 8 × 8 blocks and the complexity in that area is examined. First, when the complexity of each block exceeds a certain threshold value, that portion is replaced with a computer generated hologram. The complexity threshold value varies depending on each image, but the complexity value that can be regarded as a noise region is approximately around 0.5. It is necessary to find a threshold value that does not affect the image quality depending on the image. The computer generated hologram also needs to exceed the complexity threshold value. Therefore, when the threshold value is not exceeded, the conjugate calculation is performed. In the case of the error diffusion type, since the complexity is close to 0.7, it can be considered that the error can be embedded as it is. Therefore, it is considered that the error diffusion type is suitable for embedding.

<Removal procedure>
When extracting information, an area of the bit plane or more is extracted from each bit plane, and the original computer generated hologram is constructed.

  In the pixel replacement type and gradation value change type described above, only computer holograms having the same size as the image could be embedded. However, when complexity is used, a computer generated hologram can be embedded if each bit plane has a certain complexity or more. In practice, a computer generated hologram having a size twice the image size can be embedded. How much secret information can be embedded depends on the cover image used. Looking at the effect on image quality, image quality deteriorates as the amount of embedding increases. In addition, when the threshold value is lowered, the amount of embedding increases as a matter of course, but the image quality decreases. According to experiments, if it is a natural image, it is possible to embed a hologram more than twice the image size. Since it is a computer generated hologram, three-dimensional information can be embedded in a two-dimensional form. In the case of a three-dimensional hologram, the safety as steganography can be enhanced by using the depth information as key information.

  Next, a method for embedding a secret image in pseudo halftone processing that has strong resistance to various modification operations will be described. In this method, the secret image is converted into a computer generated hologram, and the dither matrix of the systematic dither method is changed according to the shape, thereby embedding the hologram in the entire image. By this method, it is possible to embed information that is not affected by modification operations such as rotation and cutting. Even if the hard copy is made, the watermark information can be taken out by optical hologram reproduction processing. The specific method is described below.

  Hereinafter, the principle of the systematic dither method will be described. The systematic dither method is a method in which a threshold value for quantizing to binary is changed for each pixel to express a grayscale image in a pseudo manner. In the systematic dither method, a threshold matrix called an N × N dither matrix is superimposed on an input image, and corresponding pixels are binarized using elements of the dither matrix as thresholds. Since it is a two-tone expression of white and black, grayscale display cannot be performed in pixel units, but the entire image can be perceived as if grayscale is displayed due to the visual integration effect. There are several types of dither matrices, but in the following, three types of dither matrices of Bayer type, halftone type, and magic square type are used.

  FIG. 13 is a schematic diagram of the systematic dither method. FIG. 14 is a diagram showing specific values of the three types of dither matrices and an image displayed in pseudo gray scale.

  Hereinafter, a method for embedding a hologram in the systematic dither method will be described. Here, a method of embedding a hologram using a systematic dither method is specifically shown.

<Procedure 1>
In order to use the systematic dither method, the cover image is quantized to 16 gradations. (Here, 16 gradations are used in order to use a 4 × 4 dither matrix.) The image is divided into 4 × 4 blocks.

<Procedure 2>
The secret image is converted into a computer generated hologram according to the computer hologram composition method described above. Here, the computer generated hologram is displayed in two gradations. Any type of computer generated hologram can be used, but considering the effect of embedding, a high-complexity error diffusion type or a two-gradation phase type with conjugate calculation is suitable. The size of the computer generated hologram is made equal to the number of blocks when the cover image is divided by 4 × 4. For example, in the case of a 256 × 256 size cover image, a 64 × 64 size hologram is used. This is to correspond to each block of a 16-gradation cover image obtained by dividing each pixel of the hologram by 4 × 4. That is, the upper left pixel of the hologram is considered to correspond to the upper left block of the input image.

<Procedure 3>
Two types of dither matrices are prepared. Any dither matrix can be used, but the combination of the matrices used also affects the image quality when the computer hologram of the secret image is embedded. Here, for convenience, it is assumed that two types of dither matrices D1 and D0 are used.

<Procedure 4>
One block of the 16 gradation cover image is taken out, and the luminance level of the corresponding hologram is examined. If the corresponding hologram pixel is 1 (white), the dither matrix D1 is used. If it is 0 (black), the dither matrix D0 is used. Once the dither matrix to be used is determined, it is used to output 1 if the luminance level of each pixel in the block of the input image exceeds the threshold value of the corresponding dither matrix, and 0 if it is below the threshold value. Is output (see FIG. 15).

  By repeating the above procedure for each block, an image displayed in a pseudo gray scale by a systematic dither method in which computer hologram information is embedded is synthesized.

  When taking out the secret image, the embedded image is considered as a Fourier transform type hologram, and Fourier transform is performed. Then, the embedded secret image (reproduced image of the computer hologram of the secret image) can be taken out by displaying the magnitude of the obtained complex amplitude distribution in grayscale.

  In this method, two types of dither matrices are used. Depending on the combination of dither matrices, the quality and texture of the output image obtained, the quality of the reproduced image, and the like differ. Here we compared what kind of images can be obtained by combining various dither matrices. 16 and 17 show combinations of dither matrices. The title here is the name of the dither matrix combination. D1 and D0 in the table are dither matrices used when the hologram is 1 (white) and 0 (black) as described above.

  By using two dither matrices, an error of a minute shape is generated between each block, thereby embedding the hologram, so that the hologram shape appears strongly in the dither matrix greatly different, and weak in the similar dither matrix Will appear. Therefore, the closer the two types of matrices are, the better the image quality.

  Next, optical reproduction of embedded information will be described. Since the image itself, which is displayed in pseudo gray by the dither method, has a hologram shape inside, the secret image can be reproduced optically in the same way as a normal computer hologram without digital processing on the computer. Can be confirmed. That is, if the embedded image itself is considered as a Fourier transform type hologram and is irradiated with reproduction light by a laser or the like, the secret image is reproduced by the light diffraction phenomenon.

  FIG. 18 is a diagram illustrating an example of a hologram reproducing optical system. In FIG. 18, a computer generated hologram 4 is placed between two lenses 3 by using a He—Ne laser 1 of an optical system laser, which is a light source, a beam expander 2 and two lenses 3. A reproduced image (reproduced image) is output on the plane 5. That is, in order to reproduce the embedded image as the computer generated hologram 4, the image is taken as a photograph, and is reproduced after being reduced to the size of the film. This is for reproduction by the laser 1. The created film is irradiated with a laser 1 and a video camera is installed on a lens focal plane (focal plane) 5 to observe a reproduced image. As the laser, a helium-neon (He-Ne) laser 1 is used, and a lens 3 having a focal length f = 400 [mm] is used. The secret image computer hologram 4 uses an error diffusion type hologram, and is embedded using a halftone dot type (1) dither matrix shown in FIG.

  Although the image information security method of the present invention has been described above, the present invention can be applied to ID cards and electronic keys. It can also be used for watermarks and numbers such as banknotes and securities. In a conventional IC card, confidential information embedded due to scratches or breakage cannot be used, but in the case of the present invention, since it can be embedded in a plurality of locations, it is highly resistant to scratches and breakage. In addition, when an embedded image is created by embedding a secret image in a copyrighted work, when the embedded image (copyrighted work) is copied, the secret image is also copied, so that unauthorized copying of the copyrighted work can be easily proved.

It is a flowchart which shows the process of the image information security method of this invention. It is a figure which shows a Lohmann (Roman) type cell. It is a figure which shows (A) Lohmann (Roman) type | mold computer hologram and (B) reproduced image. It is a figure which shows a Lee (Lee) type | mold cell. It is a figure which shows (A) Lee (Lee) type | mold computer hologram and (B) reproduction | regeneration image. It is a figure which shows (A) binary phase type | mold computer hologram and (B) reproduced image. It is a figure which shows the concept of an error diffusion method. It is a figure which shows (A) error diffusion type | mold computer hologram and (B) reproduced image. It is a figure which shows evaluation (PSNR) of image quality. It is a figure which shows evaluation (PSNR) of image quality. It is a figure which shows the complexity (alpha) of each bit plane of an image. It is a figure which shows the relationship between complexity alpha and the number of pixels which can be embedded. It is a schematic diagram of the systematic dither method. FIG. 14 is a diagram showing specific values of the three types of dither matrices and an image displayed in pseudo gray scale. It is a figure which shows the determination of the dither matrix by a computer generated hologram. It is a figure which shows the combination of a dither matrix. It is a figure which shows the combination of a dither matrix. It is a figure which shows an example of the reproduction | regeneration optical system of a hologram.

Explanation of symbols

1 He-Ne laser 2 Beam expander 3 Lens 4 Computer generated hologram 5 Focal plane (reproduced image)

Claims (9)

An image information security method for protecting image information,
(A) Prepare a secret image to be protected,
(B) Prepare a cover image in which the secret image prepared in step (A) is embedded,
(C) combining the secret image prepared in step (A) with a computer generated hologram;
(D) Embed the computer generated hologram synthesized in step (C) in the cover image prepared in step (B) to generate an embedded image;
(E) A secret image is reproduced from the embedded image generated in the step (D).
An image information security method characterized by the above.   2. The image information security method according to claim 1, wherein the step (C) combines the secret image prepared in the step (A) with a Lohmann-type hologram. 3.   2. The image information security method according to claim 1, wherein the step (C) combines the secret image prepared in the step (A) with a Lee hologram.   The image information security method according to claim 1, wherein the step (C) combines the secret image prepared in the step (A) with a binary phase hologram.   The image information security method according to claim 1, wherein the step (C) combines the secret image prepared in the step (A) with an error diffusion hologram.   The step (D) includes embedding an image by embedding the computer generated hologram synthesized in the step (C) in the cover image prepared in the step (B) by pixel replacement. 5. The image information security method according to any one of 5.   The step (D) embeds the computer generated hologram synthesized in the step (C) in the cover image prepared in the step (B) using a gradation value to generate an embedded image. The image information security method according to any one of 1 to 5. In the step (D), the cover image prepared in the step (B) is divided into a plurality of areas, the complexity of the divided areas is detected, and the complexity exceeds a predetermined threshold. Embed the computer generated hologram synthesized in (C) to generate an embedded image,
The step (E) divides the embedded image generated in the step (D) into a plurality of regions, detects the complexity of the divided regions, and extracts a secret image from the region where the complexity exceeds a predetermined threshold. Reproduce,
6. The image information security method according to claim 1, further comprising: In the step (D), the computer generated hologram synthesized in the step (C) is embedded in the cover image prepared in the step (B) using a systematic dither to generate an embedded image,
The step (E) optically reproduces a secret image from the embedded image generated in the step (D).
6. The image information security method according to claim 1, further comprising: