JP3785166B2 - Image reproducing apparatus and image reproducing method - Google Patents

Description

  The present invention relates to an image processing apparatus, and more particularly to a reproducing apparatus for reproducing an additional image from an image in which another additional image is superimposed and recorded on a main image.

  The technology that superimposes other additional information (mainly images) on an image and records the additional information together with the image in an invisible state is, for example, an ID card on which an image such as a face photograph is recorded, a logo mark, or a seal image. It is effective in preventing falsification and forgery of documents and other recorded items recorded as. As a method for superimposing additional information on a recorded image, methods shown in the following documents are known.

  Non-Patent Document 1 discloses a method of superimposing information on a digital image expressed in pseudo gradation. In this method, when the density is expressed by pseudo gradation, additional information is superimposed using a degree of freedom that allows the same density to be expressed by a plurality of gradation patterns.

  In Non-Patent Document 2, a second image subjected to a predetermined conversion is digitally superimposed on the first image. In this method, by converting the low-frequency component of the second image into the high-frequency component, the second image cannot be seen by humans or is observed as noise-like meaningless. The second image can be restored by applying a predetermined second conversion to the superimposed image of the first image and the second image.

  Japanese Patent Application Laid-Open No. 2004-228561 discloses a method that can specify a recorded copying machine or the like from a hard copy output of a color copying machine. In this method, a small yellow dot pattern is recorded on the hard copy output. This dot pattern has a shape corresponding to conditions such as the model number of the copying machine. By reading the hard copy output with a scanner or the like, extracting the dot pattern recorded in an overlapping manner and performing predetermined signal processing, the recorded copying machine can be identified.

  Patent Document 2 discloses a method of superimposing additional information on a color image as a high-frequency color difference signal. In this method, additional information is coded, and a color difference component having a high spatial frequency peak corresponding to the code is superimposed and recorded on the original image. Since the color difference component of high spatial frequency is difficult to be seen by humans, the superimposed additional information hardly degrades the original image. In addition, since a general image contains almost no high-frequency color difference component, the superimposed additional information can be reproduced by reading the recorded image and extracting the high-frequency color difference component by signal processing.

  Non-Patent Document 3 is a method known as a similar example, although it is not the superimposition of additional information on an image. In this method, two random images are formed, and when these are overlapped, a significant image appears. However, invisible information is not superimposed on a significant image.

In Non-Patent Document 4, when a certain image is represented by pseudo gradation, two images having different density gradation expression patterns are created only in a specific region, and when the two images are overlapped, the specific region is obtained. Has proposed a method that makes the color appear dark.
JP-A-4-294862 JP-A-7-123244 “Synthetic coding method of text data to image by color density pattern”, Journal of the Institute of Image Electronics Engineers of Japan, 17-4 (1988), pp194-198 Komatsu et al .: "Proposal of digital watermarking in document image communication and its application to signature", IEICE J72-B-1, pp208-218. Mutsuura: “Simple image encryption method for decryption”, IEICE J72-B-1,12 Oka: "Visual Encryption of Information", 1995 IEICE Fundamental / Boundary Society Conference Preliminary Proposal A-123

  Among the known techniques mentioned above, the techniques of Non-Patent Document 1 and Non-Patent Document 2 superimpose other additional information on a digital image, but perform superimposing operation on a digital signal, and hard copy It is not intended for output. Moreover, complicated signal processing and arithmetic processing are required to extract the superimposed additional information, and it is difficult to easily reproduce the additional information. Furthermore, it is possible to combine hard copy recording with images created by these methods, but by performing the recording and reading operations, the superimposed additional information is significantly degraded, and it is difficult to restore the additional information. is there.

  The techniques of Patent Document 1 and Patent Document 2 are intended for hard copy images, but in order to reproduce the superimposed additional information, it is necessary to perform image reading, signal processing, and computation. It was difficult to easily reproduce the additional information.

  In the technique of Non-Patent Document 3, information can be easily reproduced by stacking sheets on a recorded material, but this is a random image by itself, and superimposing other information on a significant image. It cannot be used for.

  In the technique of Non-Patent Document 4, two images to be superimposed are paired, and information cannot be reproduced unless images that are not paired are superimposed. That is, if information is to be reproduced for a plurality of images, it is necessary to prepare corresponding images for the number of images.

  An object of the present invention is to provide an image reproduction apparatus for reproducing an image superimposed and recorded on a hard copy by a simple operation that does not require complicated signal processing.

An image reproduction apparatus according to the present invention is a pattern-modulated image obtained by modulating a color difference pattern image signal having a pattern in which gain information given to a predetermined color difference amount is represented for each image unit using a first image signal. An image reproducing apparatus for reproducing the first image signal as a visible image from a recorded matter in which an image signal superimposed on a second image signal after being multiplied by a predetermined color difference amount is recorded as an image, A sheet having a transmittance distribution of the same pattern as the pattern of the color difference pattern image signal and provided on the recorded matter is provided.

The image reproduction apparatus according to the present invention also provides a pattern obtained by modulating a color difference pattern image signal having a pattern in which gain information given to a predetermined color difference amount is represented for each image unit using a first image signal. An image reproduction apparatus for reproducing a first image signal as a visible image from a recorded matter in which an image signal obtained by multiplying a modulated image signal by a predetermined color difference amount and superimposed on a second image signal is recorded as an image. And an optical element having the same thickness distribution as the pattern of the color difference pattern image signal and disposed on the recorded matter.

  When a composite image in which the first image is embedded on the second image is reproduced, the embedded image can be easily reproduced without requiring complicated signal processing.

(First embodiment)
A first embodiment in which the present invention is applied to an image composition recording / reproducing apparatus will be described below. This apparatus is an apparatus that synthesizes and records two images and reproduces them. More specifically, the two images are combined by embedding the first image in the second image. Then, when a synthesized image to be recorded is viewed by a human in a normal manner, it looks almost the same as the second image, the first image cannot be seen, and the first image is reproduced only by a special device or method. Make it visible. In the following description, for convenience, performing image synthesis in this way is expressed as “embedding” the first image in the second image, the first image is “embedded image”, and the second image is “covered”. These will be called “embedded images”. In addition, an operation of making an embedded image visible to a human is expressed as “playing”.

  FIG. 1 shows a configuration of an image composition recording apparatus according to the present embodiment. The image composition recording apparatus includes a CPU 101, an image memory 102, an image input unit 103, a program memory 104, and an image recording unit 105, all of which are connected by a bus 106. The CPU 101, the image memory 102, the image input unit 103, and the program memory 104 constitute an image processing unit 107.

  The operation of this image composition recording apparatus will be briefly described. First, the embedded image and the embedded image are written into predetermined areas of the image memory 102 through the image input unit 103, respectively. Then, based on the algorithm shown below, these images are subjected to calculation processing to create a composite image, and this composite image is recorded as a color hard copy by the image recording unit 105.

  These series of processes are all performed by the CPU 101 in accordance with a program stored in the program memory 104. A dedicated device may be used for the processing, but a general-purpose computer such as a personal computer may be used. In this case, the image memory 102 and the program memory 104 generally use the same memory in divided areas.

  Next, the configuration of the embedded image and the embedded image, which are input images of the present apparatus, and the contents and meaning of image processing will be described in detail.

  The input image is usually expressed as digital information having a defined density on each grid point in the Cartesian coordinate system in the same manner as used for expression in a computer. Here, the two axes of the orthogonal coordinate system are the x-axis and the y-axis, and are expressed as horizontal and vertical for convenience.

  In the present embodiment, a monochrome binary image such as a figure or a character is used as the embedded image. The density value of the pixel (x, y) is represented as R (x, y). On the other hand, a full-color image is used as an embedded image, and pixel values of R, G, and B color components are expressed as Pr (x, y), Pg (x, y), and Pb (x, y). These pixel values represent black when Pr = 0, Pg = 0, and Pb = 0, and white when Pr = 1, Pg = 1, and Pb = 1.

  Next, an image processing algorithm in the present embodiment will be described in detail. The processing procedure is shown in the flowchart of FIG.

[First step (pattern generation)]
First, a pattern image Q (x, y) is generated in the first step S11. The pattern image is an image that is modulated by the embedded image and superimposed on the embedded image, and is preferably an image with a high spatial frequency that is difficult to be felt by human eyes.

  In the present embodiment, a checkered pattern image as shown in FIG. 3 is used as the pattern image Q (x, y). Here, each pixel of the pattern image Q (x, y) is represented by a numerical value of 1 or −1, but its physical meaning is given to a predetermined color difference amount (Vr, Vg, Vb) for each pixel. It is gain. Such a pattern image Q (x, y) is referred to as a color difference pattern image. The pattern image Q (x, y) shown in FIG. 3 is a pattern image in which pixels of gain 1 and gain −1 are arranged in a checkered pattern with (4 × 4) pixels as a unit, and the generation equation thereof is expressed by equation (1). Shown in

(Int (x / 2) + int (y / 2)) mod 2 = 0
Then Q (x, y) =-1
(Int (x / 2) + int (y / 2)) mod 2 = 1
Then Q (x, y) = 1 (1)
Here, int (x) is an operation that takes an integer part of x, and x mod y is an operation that represents a remainder when x is divided by y. The pattern image Q (x, y) is an image having a direct current component of 0 and a small low frequency component, that is, an image having a high spatial frequency.

[Second step (pattern modulation)]
Next, in the second step S12, the pattern image Q (x, y) is modulated by the embedded image R (x, y). At this time, the embedded image R (x, y) is first processed by the smoothing filter according to the equation (2) to obtain a smoothed embedded image R ′ (x, y).

  Here, xi, yi, and Ai represent the smoothing filter kernel. In the present embodiment, a smoothing filter of (5 × 5) pixels of the kernel as shown in FIG. That is, −2 ≦ xi, yi ≦ 2, and A (xi, yi) = 1/25.

  With this smoothed embedded image R ′ (x, y), the pattern image Q ′ (x, y) is modulated according to the equation (3) to obtain the pattern modulated image Q ′ (x, y).

Q ′ (x, y) = Q (x, y) · (−2 · R ′ (x, y) +1) (3)
As a result of this processing, a pattern-modulated image Q ′ (x, y) is obtained by −1 times the pattern image Q (x, y) in the region R ′ = 1, that is, an image obtained by inverting the pattern image Q (x, y). As a result, in the region where R ′ = 0, the pattern image Q (x, y) itself is a pattern modulation image. In the region where R ′ takes a value between 1 and 0, the pattern-modulated image Q ′ (x, y) has an intermediate value between them. Since R ′ is a smoothed signal, it takes between 1 and 0 in the edge region of the embedded image R (x, y). Therefore, by the above processing, the polarity of the amplitude is inverted according to the pixel value of the embedded image R (x, y), and an image in which the amplitude changes gently at the edge portion is obtained as the pattern modulation image Q ′ (x, y). It is done.

  FIG. 5 shows the relationship among the embedded image R (x, y), the smoothed embedded image R ′ (x, y), the pattern image Q (x, y), and the pattern-modulated image Q ′ (x, y). However, in FIG. 5, for convenience, the image is represented in one dimension.

  In the above description, the amplitude modulation is performed on the pattern image Q (x, y) by the smoothed embedded image R ′ (x, y). However, as another example, it is expressed by the equation (4-1). Phase modulation may be used.

Q ′ (x, y) = Q (x + g (R ′ (x, y)), y) (4-1)
However, g (x) is a function that takes 0 when x = 0, 3 when x = 1, and 1 when 0 <x <1.

  When phase modulation is performed in this way, the pattern image Q (x, y) is shifted by 4 pixels in the x-axis direction in the region of R ′ = 1, and the pattern image Q (x, y) in the region of R ′ = 0. ) Itself becomes a pattern-modulated image Q ′ (x, y). Since the pattern image Q (x, y) is an x-axis symmetric periodic image with a period of 4 pixels, a 4-pixel shift and −1 times the amplitude have the same meaning. Therefore, the processing result by the phase modulation and the processing result by the amplitude modulation described above are equivalent in the portion where R ′ is 0 or 1, and the embedded image R (x, y) in which the value of R ′ takes an intermediate value. Only the edge part will be different. In this phase modulation processing, an image in which the phase changes gently at the edge portion is obtained.

  FIG. 6 shows an embedded image R (x, y), a smoothed embedded image R ′ (x, y), a pattern image Q (x, y), and a pattern modulated image Q ′ (x, y) in the case of this phase modulation processing. 6) is a one-dimensional representation of the relationship in FIG.

  Further, the above description is an example of phase modulation in the x-axis direction, but phase modulation in the y-axis direction may be performed as shown in Expression (4-2).

Q ′ (x, y) = Q (x, y + g (R ′ (x, y))) (4-2)
The smoothing of the embedded image R (x, y) is not limited to the smoothing by the two-dimensional reference area of (5 × 5) pixels that is vertically and horizontally symmetric as shown in FIG. For example, as shown in FIGS. 4B, 4C, and 4D, the reference area may be an asymmetric rectangle, a one-dimensional rectangle, or a non-rectangular shape. Furthermore, weighted smoothing as shown in FIG. In particular, when this processing is performed by pipeline-type hardware described later, the circuit cost can be reduced by performing smoothing by a method such as that shown in FIGS. 4B and 4C that can be configured with a small-capacity line memory.

[Third step (pattern superposition)]
Next, in the third step S13, the pattern modulation image Q ′ (x, y) is superimposed on the embedded image. In the present embodiment, a simple addition operation is performed as the superimposition process. As described above, since the pattern image Q (x, y) is a gain given to the color difference amounts (Vr, Vg, Vb), the embedded image is Pi (x, y) (i = r, g, b), and the pattern After the modulated image Q ′ (x, y) is multiplied by the color difference amounts (Vr, Vg, Vb), the embedded image is added to Pi (x, y). Here, the color difference amounts (Vr, Vg, Vb) are set so that the brightness is 0 or approximately 0, and the intensity is approximately below the human visual limit. For example, a value of (Vr, Vg, Vb) = (0.2, 0.2, −0.4) is used. This will be described in detail later. When the addition result exceeds the density value definition range (0, 1), clipping is performed to the minimum value or the maximum value of the definition range.

  Expression (5) shows the superimposition process in the third step S13. However, the pattern superimposition image which is a superimposition result is represented as Oi (x, y) (i = r, g, b).

Or (x, y) = Pr (x, y) + Q ′ (x, y) · Vr
Og (x, y) = Pg (x, y) + Q ′ (x, y) · Vg
Ob (x, y) = Pb (x, y) + Q ′ (x, y) · Vb
If Or (x, y) ≧ 1, Or (x, y) = 1
If Or (x, y) <0, Or (x, y) = 0
Same for Og and Ob (5)
[Fourth Step (Color Correction)]
Next, in the fourth step S14, the ink amount signal used for controlling the ink amounts of the C, M, Y3 colors by the image recording unit 105 using the pattern superimposed image Oi (x, y) expressed by the RGB color components. Convert to OC, OM, OY. This conversion has been widely known as a color correction technique. Here, color correction processing is performed in accordance with equations (6-1) and (6-2). The matrix Acr, Acg, Acb, Amr, Amg, Amb, Ayr, Ayg, Ayb in the equation (6-1) is a value depending on the chromaticity of each ink used in the image recording unit 105, and is an image recording unit. A value suitable for 105 is selected.

  If the process is performed on a YMC basis from the beginning, the process of the fourth step S14 can be omitted.

  The above is a series of image composition processing in this embodiment. After this processing, the image recording unit 105 records a color image as a hard copy in accordance with the ink amount signals OC, OM, OY. As a result, a color image having substantially the same RGB components as the pattern superimposed images Or ′, Og ′, Ob ′ is recorded on a predetermined recording material (recording paper). As the image recording unit 105, for example, a sublimation type thermal transfer printer is used. The sublimation thermal transfer method can control the density to 100 gradations or more for each pixel, and can easily perform full-color recording.

  In addition to this, for example, a silver salt photographic system may be used for the image recording unit 105, or a recording system suitable for binary recording such as an ink jet system or a melt type thermal transfer system may be used. However, when a binary recording printer is used, it is necessary to perform pseudo gradation processing such as an error diffusion method or a tissue dither method in order to express gradation. When these processes are performed, image information with a high spatial frequency close to the recording density of the printer is disturbed or lost, so it is necessary to use a recording system having a recording density sufficiently higher than the spatial frequency of the pattern image.

  In the above description, a series of image composition processing is realized by software processing, but it can also be realized by hardware.

  FIG. 7 shows a configuration of an image composition recording apparatus that implements the above-described series of image composition processing by hardware. The image processing unit includes two image memories 701 and 702 for storing an embedded image and an embedded image, a pattern generating unit 703, a pattern modulating unit 704, and a pattern superimposing unit 705, and an image recording unit 706 is combined with these images. A composite recording apparatus is configured.

  First, for example, a color difference pattern image signal 753 is generated from the pattern generation unit 703 as a pattern image signal. Then, according to the embedded image signal 751 output from the first image memory 701 in the pattern modulation unit 704, for example, phase modulation represented by the equation (4-1) or (4-2) is performed on the color difference pattern image 753. As a result, a pattern-modulated image signal 754 is generated. Next, the pattern superimposing unit 705 superimposes the pattern-modulated image signal 754 on the embedded image signal 752 output from the second image memory 702 to generate a pattern superimposed image signal 755. The pattern superimposed image signal 755 is converted into an ink amount signal 756 by the color correction unit 706, and then input to the image recording unit 707 to output a hard copy image.

  The operation of this image composition recording apparatus will be described in more detail. First, an embedded image R (x, y) and an embedded image P (x, y) are stored in the first and second image memories 701 and 702, respectively. Yes. The embedded image R (x, y) stored in the first image memory 701 is a binary image and is represented by 1 bit per pixel. The embedded image P (x, y) stored in the second image memory 701 is a full-color image of 8 bits for each of RGB, and is represented by a total of 24 bits per pixel.

  The pattern generator 703 generates a pattern image signal 753. The pattern image is a color difference pattern image signal as described above, and is generated according to the equation (1).

  The pattern modulation unit 704 includes, for example, three line memories 711, 15 latch groups 712, an adder 713, and two multipliers 714 and 715. The embedded image signal 751 output from the first image memory 701 is delayed by the line memory 711 and the latch group 712. Each latch output of the latch group 712 is a signal in a rectangular area of (5 × 3) pixels. These signals are added by an adder 713 and then input to a first multiplier 714 to be multiplied by a pattern image signal 753. It is done. The multiplication output of the first multiplier 714 is further multiplied by a set of three parameters Vr, Vg, and Vb by the second multiplier 715. Here, multiplication is performed three times for the signal of one pixel, and the result of these three multiplications is output to the pattern superimposing unit 706 as a pattern modulation image signal 754 composed of time-series RGB signals.

  The pattern superimposing unit 706 includes an adder 716 and a clipping circuit 717. First, the adder 716 adds the embedded image signal 752 from the second image memory 702 to the pattern modulation image signal 754 from the pattern modulation unit 704. The pattern modulation image signal 754 and the embedded image signal 752 are both RGB time-series signals, and the adder 716 adds the same components to each other. The addition result of the adder 716 is clipped by the clipping circuit 717 and output as a pattern superimposed image signal 755. That is, the clipping circuit 717 performs an operation of setting the output to 0 when the addition result of the adder 716 is smaller than 0, and setting the output to 255 when larger than 255.

  The pattern superimposed image signal 755 output from the pattern superimposing unit 705 is a signal having RGB three-color components, and is converted into an ink amount signal 756 representing the ink amount in the image recording unit 707 by the color correcting unit 706. The color correction unit 706 uses a lookup table, for example. This table is calculated in advance according to equations (6-1) and (6-2) and stored in the memory.

  As described above, the image composition processing can be easily realized by hardware. When hardware is used, signal processing can be performed at a relatively high speed, which is particularly effective when creating a large number of images in a short time.

  Next, the property of the composite image (pattern superimposed image) of the embedded image and the embedded image recorded by the above processing will be described. This synthesized image is an image that visually looks almost the same as the embedded image, and the information of the embedded image is information that is completely or hardly visible.

  The RGB color components of the embedded image are Pr (x, y), Pg (x, y), and Pb (x, y), the embedded image is R (x, y), and the smoothed embedded image is R ′ (x, y). If y) (= R (x, y) * LPF (x, y)), the RGB components Or (x, y), Og (x, y), Ob (x, y) of the composite image are expressed by the formula ( 7).

Or (x, y) = Pr (x, y)
+ Q (x, y) .Vr. (-2.R '(x, y) +1)
Og (x, y) = Pg (x, y)
+ Q (x, y) .Vg. (-2.R '(x, y) +1)
Ob (x, y) = Pb (x, y)
+ Q (x, y) .Vb. (-2.R '(x, y) +1)
(7)
Next, consider the brightness component Oy and the color difference component Oc of the composite image recorded in this way. Here, the brightness component Oy and the color difference component Oc are defined by equation (8).

Oy (x, y) = Kr.Or (x, y) + Kg.Og (x, y) + Kb.Ob (x, y)
Oc (x, y) = Or (x, y) −Og (x, y) (8)
The lightness component Oy represents brightness, and the color difference component Oc represents color intensity. The color difference component Oc includes two independent components, but only one component is handled here. (Kr, Kg, Kb) is a coefficient representing the lightness of each RGB component, and (Kr, Kg, Kb) = (0.18, 0.81, 0.01).

  The spectra Foy (fx, fy) and Foc (fx, fy) of the lightness component Oy and the color difference component Oc are expressed by the equations (9-1) and (9-2).

Foy (fx, fy)
= Fpy (fx, fy) + Vy · Fq (fx, fy) * (− 2Fr ′ (fx, fy) + δ (fx, fy)) (9-1)
Foc (fx, fy)
= Fpc (fx, fy) + Vc · Fq (fx, fy) * (− 2Fr ′ (fx, fy) + δ (fx, fy)) (9-2)
However, fx and fy represent the spatial frequencies in the x-axis direction and the y-axis direction, respectively. Fr ′, Fq, Fpy, and Fpc are Fourier transforms of the brightness component and the color difference component of the smoothed embedded image R ′, the pattern image Q, and the embedded image P, respectively. Furthermore, Vy and Vc are the brightness component and the color difference component of the color difference amount described above.

  Here, since the lightness component Vy of the color difference amount V is set to be 0 or sufficiently close to 0, the second term of the equation (9-1) is 0 or substantially 0.

  8A, 8B, 8C, and 8D are diagrams schematically showing Fpc, Fr ′, Fq, and Foc in the equation (9-2). In the spectrum Fpc (fx, fy) of the color difference component of a normal image, the power is concentrated on the low frequency component as shown in FIG. 8A, and the high frequency component is extremely low. On the other hand, as shown by the solid line in FIG. 8B, in the smoothed embedded image R ′, the high frequency component of the embedded image R is cut as shown by the broken line. The pattern image Q has only a high frequency component as shown in FIG. Therefore, as shown in FIG. 8D, the spectrum of the color difference component of the composite image represented by Expression (9-2) has a first term centered on the low frequency component and a second term centered on the high frequency component. Separated.

  Further, as shown in FIG. 9 (quoted from FIGS. 1 and 76 in Chapter 1 of Chapter 3 of the Television Image Engineering Handbook), it is known that human visual characteristics have low sensitivity especially for high-frequency color difference components. . Therefore, the second term of Equation (9-2) is hardly visible to humans. For this reason, only the first term of Expression (9) is observed in the lightness component and the color difference component in the composite image. That is, the composite image looks equal to the embedded image.

  Thus, the image synthesized and recorded by the present embodiment is visually the same image as the embedded image, and the embedded image component becomes a high-spatial-frequency color difference component, so that it is almost invisible. Further, since the high frequency component of the color difference component Fp1 (fx, fy) of the normal image is reduced by the smoothing process of the embedded image, there is no component that shifts to a low frequency due to the convolution of the embedded image and the pattern image.

  Next, a reproduction apparatus that reproduces an embedded image from the composite image recorded as described above will be specifically described.

  FIG. 10 is a diagram illustrating a configuration example of a playback device. The recorded material (recording paper) 1100 on which the composite image is recorded is loaded and fixed to the reproducing apparatus main body 1000 so that the upper end and the right end of the recorded material 1100 are in contact with the upper loading portion 1001 and the right end 1002 of the reproducing apparatus main body 1000. . As a result, the reproduction sheet 1003 and the image on the recorded material 1100 are held in a predetermined positional relationship. Then, a reproduction sheet 1003 connected to the reproduction apparatus main body 1000 is overlaid on the recorded material 1100, and an image on the recorded material 1100 is observed through the sheet 1003, so that the embedded image is superimposed on the embedded image. It is configured to be visible.

  Note that the reproducing apparatus is not limited to the configuration shown in FIG. 10, and may have any structure as long as the relative position between the synthesized image on the recorded material 1100 and the reproducing sheet 1003 can be fixed. In addition, the reproduction sheet 1003 is not fixed to the recorded material 1100, and can be freely moved by one hand in one or two dimensions, so that the sheet 1003 is aligned with the position where the embedded image on the recorded material 1100 is to be reproduced. It may be a simple structure. Furthermore, if the gap between the reproduction sheet 1003 and the recorded material 1100 is large, the reproduction contrast of the embedded image is lowered. Therefore, the reproduction sheet 1003 is pressed from above with a rigid transparent plate so that the gap is within 1 mm, for example. It may be structured.

  Next, the configuration of the playback sheet 1003 and the principle of playback of the embedded image will be described. The reproduction sheet 1003 is made of a transparent thin film medium such as plastic, and a predetermined pattern is formed on the medium.

  As a pattern on the reproduction sheet 1003, a pattern at the time of creating a composite image, that is, a pattern image pattern generated in the first step S11 in FIG. 2 (a pattern image pattern generated in the pattern generation unit 703 in FIG. 7). ) Using an appropriate transmittance distribution. The RGB transmittance distributions Tr (x, y), Tg (x, y), and Tb (x, y) of the reproduction sheet 1003 are shown in Expression (10). Here, (Wr0, Wg0, Wb0) and (Wr1, Wg1, Wb1) represent the RGB transmittances of the pixels whose corresponding pattern image Q (x, y) has a value of 1 and −1, respectively. In this case, two colors of white (transparent) and black are used as shown in Expression (11-1).

Tr (x, y) = 0.5 (Wr0−Wr1) · Q (x, y) +0.5 (Wr0 + Wr1)
Tg (x, y) = 0.5 (Wr0−Wg1) · Q (x, y) +0.5 (Wr0 + Wg1)
Tb (x, y) = 0.5 (Wr0−Wb1) · Q (x, y) +0.5 (Wr0 + Wb1)
(10)
(Wr0, Wg0, Wb0) = (0, 0, 0)
(Wr1, Wg1, Wb1) = (1,1,1) (11-1)
FIG. 11 shows the transmittance distribution pattern of the reproduction sheet 1003. In the figure, W represents a transparent portion, K represents an opaque portion, and W and K portions correspond to gain-1 and gain-1 pixels of the pattern image Q (x, y) shown in FIG. . By overlapping the reproduction sheet 1003 having such a transmittance pattern on the recorded material 1100, interference between the pattern on the sheet 1003 and the component of the pattern image Q (x, y) of the composite image on the recorded material 1100. The embedded image is observed as a yellow / blue color difference image superimposed on the embedded image.

  Further, as another example of the reproduction sheet 1003, a configuration in which the portions W and K in FIG. 11 are replaced with portions that transmit the colors Y and B, respectively, as shown in Expression (11-2) is very good. In this case, the embedded image is observed as a monochrome grayscale image superimposed on the embedded image.

(Wr0, Wg0, Wb0) = (1, 1, 0)
(Wr1, Wg1, Wb1) = (0, 0, 1) (11-2)
Such a reproduction sheet 1003 may be created by the recording system of the above-described image composition recording apparatus, or may be created by a completely independent recording system. However, since the recording density may vary depending on the recording system, it is easier to obtain accuracy when the reproducing sheet 1003 is created in the same recording system as the image composition recording apparatus.

  When the reproduction sheet 1003 described above is overlaid on the recorded material 1100, the high-frequency color difference information embedded as an embedded image is shifted to the low-frequency region and becomes visible to the human eye. Hereinafter, this reason will be specifically described.

  The RGB reflectances Or, Og, Ob of the composite image are expressed by the above-described equation (7). Accordingly, assuming that the RGB reflectances of the image observed by superimposing the reproduction sheet 1003 on the recorded material 1100 are Sr, Sg, and Sb, these are expressed by Expression (12). However, since the G component and the B component are the same as the R component, they are omitted.

Sr = Or · Tr
= Pr (x, y) .Tr (x, y) + Q (x, y) .Vr. (-2.R '(x, y) +1) .Tr (x, y)
= Pr (x, y) .Tr (x, y) + Q (x, y) .Vr. (-2.R '(x, y) +1). (0.5 (Wr0-Wr1) .Q (x , Y) + (0.5 (Wr0 + Wr1))
= Pr (x, y) .Tr (x, y) + 0.5.Vr. (Wr0-Wr1) .Q (x, y) ^2. (-2.R '(x, y) +1) +0.5 * Vr * (Wr0 + Wr1) * Q (x, y) * (-2 * R '(x, y) +1)
= Pr (x, y) .Tr (x, y) + 0.5.Vr. (Wr0-Wr1). (-2.R '(x, y) +1) + 0.5.Vr. (Wr0 + Wr1) .Q (X, y) · (−2 · R ′ (x, y) +1)
Same for g and b components (12)
Here, FIGS. 12A, 12B, and 12C show schematic diagrams of the spectrum distributions of the first term, the second term, and the third term of the formula (12), respectively. The first term is an image equivalent to an image in which the reproduction sheet 1003 is superimposed on the embedded image itself, and the third term is a high-frequency color difference component, so it cannot be visually seen. On the other hand, the second term is obtained by multiplying the embedded image by chromaticity (Vr · (Wr0−Wr1), Vg · (Wg0−Wg1), Vb (Wb0−Wb1)). In the present embodiment, this is a color difference component, but since it is demodulated to the same frequency as the embedded image, it becomes a visible image. Therefore, an image obtained by adding the image to be embedded, which is the first term, to the image that has been chromatically modulated by the embedded image is observed.

  Further, when a sheet having a pattern represented by the expression (11-2) is used as the reproduction sheet 1003, (Vr · (Wr0−Wr1), Vg · (Wg0−Wg1), Vb · (Wb0−) Since Wb1)) is a monochrome component, an image obtained by adding a pattern-modulated image density-modulated by the embedded image to the embedded image is observed.

  As described above, according to the image composition recording apparatus of the present embodiment, it is possible to record a composite image that is visually the same as the embedded image and in which another image (embedded image) is embedded without deterioration in image quality. In addition, by overlaying a predetermined playback sheet on the recorded material on which the composite image is recorded, it is possible to reproduce the embedded image so that it can be easily visually recognized without the need for complicated signal processing. Is possible.

(Second Embodiment)
Next, a second embodiment of the present invention will be described. In the first embodiment, the regular pattern is used as the pattern image and the reproduction sheet, but the present embodiment is different in that the irregular pattern is used. The basic configuration of the image composition recording apparatus according to this embodiment is the same as that of the first embodiment, but the processing procedure is slightly different.

  Hereinafter, the processing procedure of the image composition recording apparatus according to the second embodiment will be described with reference to the flowchart shown in FIG. 13 with a focus on differences from the first embodiment.

[First step (pattern generation)]
First, a pattern image Q (x, y) is generated in the first step S21. In the first embodiment, the pattern image Q (x, y) is a pattern image having a regular structure, but in this embodiment, an irregular pattern image or an image obtained by enlarging the irregular pattern image is used.

  The irregular pattern image has the following two conditions: (1) DC component and low frequency spectrum are 0, high frequency component is strong, and (2) the pattern image structure cannot be estimated easily from the embedded image. It is desirable to satisfy In this embodiment, as shown below, an irregular pattern image is generated by a two-dimensional Markov probability process. That is, the transition probability Prob is defined as a function f of the pixel value Q (x + axi, y + ayi) already determined around the pixel of interest (x, y), and the value of the pattern image Q (x, y) is set to 1 with this probability Prob. Or decide to -1. This is shown in equation (13). However, Prob [Q (x, y), a] is a probability that Q (x, y) = a.

Prob [Q (x, y), 1] = f (Q (x + axi, y + ayi))
Prob [Q (x, y),-1] = 1-f (Q (x + axi, y + ayi))
(13)
By repeating this process while appropriately scanning (x, y), all images of the binary irregular pattern image Q (x, y) are generated. The function f used in this embodiment is shown in Expression (14).

  FIGS. 14A and 14B show the autocorrelation function and power spectrum of the binary irregular pattern image Q (x, y) generated in this way. However, in FIG. 14, only the component in the x-axis direction is shown for simplicity. As shown in the figure, the pattern image Q (x, y) has an almost zero power at a low spatial frequency and is an image in which the power is concentrated at a high frequency.

  Here, in order to generate a probability image on a computer, a pseudo-random number such as an M series is used. That is, one pseudo-random number D that takes a value in the range from 0 to N−1 with a uniform probability is generated for each pixel, and the value of the pattern image Q (x, y) is determined by equation (15). That's fine.

If D / N ≧ Prob, Q (x, y) = 1
If D / N <Prob, Q (x, y) = − 1 (15)
In this case, if only the seed of the pseudo random number and the function f are stored, the same pattern image can be generated uniquely, and it is not necessary to store the pattern image Q (x, y) itself.

  In the above description, since the pattern image Q (x, y) is generated by the Markov process, the amount of calculation increases. Therefore, the pixel value of the pattern image Q (x, y) may be determined using a random number directly. A pattern generated by error diffusion processing may be used. The former is white noise, and the direct current component and low frequency components are not necessarily small values, but the calculation of generating the pattern image Q (x, y) becomes very simple. In the latter case, the low frequency component is reduced and the image can be calculated deterministically, so that random number generation processing is not required.

[Second step (pattern modulation)]
[Third step (pattern superposition)]
[Fourth Step (Color Correction)
Next, in the second step S22, the third step S23, and the fourth step S24, pattern modulation, pattern superimposition, and color correction are sequentially performed, and the finally obtained composite image is recorded. Since this is the same as the embodiment, the description is omitted. However, since the pattern is irregular and does not have periodicity, the phase modulation based on the periodic pattern as exemplified in the first embodiment is not used in the pattern modulation processing.

  Through the above processing, a combined image obtained by combining the embedded image and the embedded image is recorded. Similar to the first embodiment, this synthesized image is visually an image that looks almost the same as the embedded image, and the information of the embedded image is information that is completely or hardly visible.

Next, the properties of the composite image obtained in the present embodiment will be specifically described.
The RGB color components Or (x, y), Og (x, y), Ob (x, y), the lightness component Oy (x, y) and the color difference component Oc (x, y) of the composite image are the first Similar to the embodiment, it is expressed by equations (7) and (8). The spectra Foy (fx, fy) and Foc (fx, fy) of the lightness component Oy (x, y) and the color difference component Oc (x, y) are expressed by equations (9-1) and (9-2). . Only the content of the spectrum Fq of the pattern image Q (x, y) is different from that of the first embodiment.

  Similar to the first embodiment, the lightness component VY of the color difference amount V is set to be 0 or sufficiently close to 0, so the second term of the equation (9-1) is 0. Further, as shown in FIG. 14, Fq is also designed to have a low low frequency component, so the second term of the equation (9-2), which is a convolution of Fq and Fr + δ, is a very weak signal with a low frequency component. That is, the contribution of the second term of Equation (9-2) to the low frequency component of the color difference component of the composite image is very small. As described above, since the high-frequency color difference component has low visual sensitivity, only the component of the first term, that is, the embedded image is observed in the synthesized image.

  FIGS. 15A, 15B, 15C, and 15D show spectral distributions of Fpc, Fr ′, Fq, and Foc in the formula (9-2).

  In this embodiment, since the low-frequency component of Fq is not completely zero, the contribution of the embedded image R (x, y) to the low-frequency component is larger than that in the first embodiment, but the pattern image Q ( Since the low-frequency component of the irregular pattern x, y) can be controlled by the function f, it can be designed so as not to be seen sufficiently by setting it appropriately. On the other hand, since disorder of the regular pattern is easily detected, this influence is reduced by using an irregular pattern in the pattern image Q (x, y) in this aspect.

  Next, a method for reproducing an embedded image from the composite image recorded in the second embodiment will be described. Also in this embodiment, embedded information is reproduced by the same method as in the first embodiment. However, unlike the first embodiment, the reproduction sheet 1003 is a sheet having the same structure as the pattern image used in the image composition recording apparatus of the second embodiment.

  Expression (16) shows the RGB transmittance distributions Tr (x, y), Tg (x, y), and Tb (x, y) of the reproduction sheet 1003 used in this embodiment. This is the same form as Expression (10) shown in the first embodiment, but the pattern image Q (x, y) is different, so the contents are different.

Tr (x, y) = 0.5 (Wr0−Wr1) · Q (x, y) +0.5 (Wr0 + Wr1)
Tg (x, y) = 0.5 (Wg0-Wg1) .Q (x, y) +0.5 (Wg0 + Wg1)
Tb (x, y) = 0.5 (Wb0−Wb1) · Q (x, y) +0.5 (Wb0 + Wb1)
(16)
Also, the values of parameters Wr0, Wr1, Wg0, Wg1, Wb0, Wb1 and their modifications are shown in equations (17-1) and (17-2).

(Wr0, Wg0, Wb0) = (0, 0, 0)
(Wr1, Wg1, Wb1) = (1,1,1) (17-1)
(Wr0, Wg0, Wb0) = (1, 1, 0)
(Wr1, Wg1, Wb1) = (0, 0, 1) (17-2)
Similar to the first embodiment, such a reproduction sheet 1003 is superimposed on the recorded material 1100 as shown in FIG. 10 to cause interference between patterns, so that the embedded image is superimposed on the embedded image. Observed as a blue color difference image.

Next, the reproduction principle of the embedded image in the present embodiment will be described. Assuming that the combined RGB reflectance when the reproduction sheet 1003 is superimposed on the recorded material on which the composite image is recorded is Sr, Sg, Sb, these are expressed in the same manner as in the first embodiment. Also in this embodiment, since Q (x, y) ² is always 1, Sr (x, y) is expressed by equation (18) as in equation (12).

Sr = Or · Tr
= Pr (x, y) · Tr (x, y)
+ 0.5 · Vr · (Wr0-Wr1)
・ (−2 ・ R ′ (x, y) +1)
+ 0.5 · Vr · (Wr0 + Wr1)
・ Q (x, y) ・ (−2 ・ R ′ (x, y) +1) (18)
FIGS. 16A, 16B, and 16C show the spectral distributions of the first term, the second term, and the third term in Expression (18). As in the first embodiment, the first term is an image equivalent to an image in which the reproduction sheet 1003 is superimposed on the embedded image itself, and the third term is a high-frequency color difference component, so it cannot be visually seen. . On the other hand, the second term is obtained by multiplying the embedded image by chromaticity (Vr · (Wr0−Wr1), Vg · (Wg0−Wg1), Vb · (Wb0−Wb1)), and becomes a visible image. Therefore, an image obtained by adding an image that has been chromatically modulated with the embedded image to the embedded image that is the first term is observed.

  Further, when a sheet having a pattern represented by the expression (11-2) is used as the reproduction sheet 1003, (Vr · (Wr0−Wr1), Vg · (Wg0−Wg1), Vb · (Wb0−Wb1) Since ()) is a monochrome component, an image obtained by adding an image whose density is modulated with an embedded image to the embedded image is observed.

  As described above, in this embodiment as well, as in the first embodiment, a synthesized image in which another image (embedded image) is embedded without being deteriorated in image quality is visually similar to the embedded image. Can do. Further, by superimposing a predetermined reproduction sheet on the recorded material on which the composite image is recorded, it is possible to reproduce the image so that it can be easily recognized visually without requiring complicated signal processing.

  This embodiment further has the following advantages over the first embodiment. Since an irregular pattern image is used as the pattern image, it is difficult to estimate the pattern of the embedded image from the synthesized image. It becomes difficult for a person to estimate pattern information from a composite image and create a composite image or reproduce an embedded image. Therefore, the present embodiment is suitable for an application in which image synthesis and reproduction are allowed only for a specific person.

(Third embodiment)
Next, a third embodiment according to the present invention will be described. In the first and second embodiments, the embedding of the pattern modulation image is performed by addition processing, whereas the present embodiment is different in that embedding is performed by pseudo gradation processing. In this embodiment, an ink jet printer that is a binary recording system is used as the image recording system.

  The image composition recording apparatus of the present embodiment is basically the same as that of the first embodiment, and only the processing procedure is different. Hereinafter, the processing procedure will be described in detail with reference to the flowchart shown in FIG.

[First step (pattern generation)]
First, a pattern image Q (x, y) is created in the first step S31. As this pattern image Q (x, y), an irregular pattern similar to that of the second embodiment is created.

[Second step (pattern modulation)
Next, in the second step S32, the pattern image is modulated by the embedded image. Since this point is the same as that of the second embodiment, details are omitted.

[Third step (color correction)]
In the present embodiment, the embedded images Pr, Pg, and Pb expressed in RGB in the third step S33 represent control amounts of three color inks of C (cyan), M (magenta), and Y (yellow). Color correction processing for converting the ink density signals Pc, Pm, and Py is performed. The conversion of the color correction process is the same as the color correction process of the first embodiment, and is performed according to equations (19-1) and (19-2).

[Fourth step (pattern superposition)]
In the next fourth step S34, the pattern-modulated image is superimposed by the error diffusion method based on the ink density signals Pc, Pm, Py obtained by the color correction process. Since this pattern superimposition process is significantly different from the first and second embodiments, it will be described in detail. This process is similar to the pseudo gradation method represented by the conventional error diffusion method. However, control is performed so that the pattern structure peculiar to the pseudo gradation is close to the image pattern.

  In the fourth step S34, the same processing is performed for the Y component Py, the M component Pm, and the C component Pc of the ink density signal, so only the Y component Py will be described as a representative here. The fourth step S34 includes the following four substeps S34-1, S34-2, S34-3, and S34-4.

[Substep S34-1]
First, as shown in Expression (20), the accumulated error signal E′Y (x, y) is added to the embedded image Py (x, y). The accumulated error signal E′Y (x, y) is for correcting a quantization error due to binarization, and a generation method thereof will be described later.

P′Y (x, y) = PY (x, y) + E′Y (x, y) (20)
[Substep S34-2]
Next, the addition result PY ′ is binarized according to the equation (21).
if PY ′ (x, y) + (2 · P (x, y) −1) · Vy
≧ 0.5 OY (x, y) = 1
if PY ′ (x, y) + (2 · P (x, y) −1) · Vy
<0.5 OY (x, y) = 0 (21)
Here, Vy is a parameter for determining the color difference intensity to be superimposed, and Vm and Vc are used for the M component and C component, respectively. In the present embodiment, values (Vm, Vc, Vc) = (+ 0.2, −0.12, −0.12) are used. The pattern superimposition process by the error diffusion method in the fourth step S34 is different from the conventional error diffusion method in the sub-step S34-2.

[Sub-step S34-3]
Next, the error calculation of Expression (22) is performed.
EY (x, y) = PY ′ (x, y) −Oy (x, y) (22)
EY (x, y) represents a quantization error due to binarization, and the error is compensated by feeding back this error component to the input signal.

[Sub-step S34-4]
Next, the cumulative error is calculated by equation (23).

Here, a (xi, yi) is an error distribution coefficient, and the value shown in Table 1 is used.

  By repeating the above processing while scanning the pixels, the processing is performed for all the images. Further, by performing the same calculation for the M component and the C component, binarized images O′y, O′m, and O′c are obtained by error diffusion.

  Although the error diffusion method is used here, a dither method or the like may be used instead of the error diffusion method.

  The above is a series of image composition processing. Finally, the image recording unit records an image according to the binarized output images O′Y, O′M, and O′C. That is, when Oy (x, y) = 1, Y color ink is printed on the pixel (x, y), and when Oy (x, y) = 0, printing is not performed. As a result, an image in which the density averaged in the macro area is almost the same as the density represented by O′y, O′m, and O′c is reproduced.

  At this time, the black printing process may be performed. Ink reprinting is a process that replaces all YMC pixels that record colors with K (black) ink, reducing printing costs by reducing the amount of ink, reducing ink bleeding, and improving density by using black ink. There are effects such as. Various methods have been proposed as specific processing. For example, the following processing may be performed and recording may be performed in accordance with O ″ Y, O ″ M, O ″ C, and O ″ K of the blackened image. . In this case, the image recording unit needs to be able to record four colors of YMCK.

if (O′Y = 1 and O′M = 1 and O′C = 1)
OY ″ = 0 OM ″ = 0 OC ″ = 0 OK ″ = 1
else OY ″ = OY OM ″ = OM OC ″ = OC OK ″ = 0
(24)
Through the series of processes described above, a pattern-modulated image is embedded as an error diffusion pattern. A recorded image (synthesized image) obtained by this processing has the following properties. That is, by compensating the density by error diffusion processing, an image having substantially the same chromaticity as that of the embedded image is recorded as a synthesized image in a macro manner. Similarly to the second embodiment, the modulation pattern image component is a color difference signal having a high frequency component with low visual sensitivity. Further, the low-frequency component that is slightly present is further reduced by the density compensation effect by error diffusion. For this reason, the component of the pattern modulation image modulated by the embedded image can hardly be visually observed.

  Since the synthesized image is binarized after the intensity of the pattern modulation image is added in the error diffusion process, the binarized image has a very high correlation with the pattern modulation image. That is, according to the value of the modulation parameter (Vy, Vm, Vc), a positive correlation is obtained when the parameter is positive, and a negative correlation is obtained when the parameter is negative. Further, the correlation increases as the absolute value of the parameter increases. Here, since Vy ≧ 0, Vm <0, and Vc <0, the Y component has a high positive correlation with the pattern modulation image, and the M and C components have a high negative correlation with the pattern modulation image.

  On the other hand, the pattern modulation image is an inversion of the pattern image by the embedded image. For this reason, the positive / negative of this correlation is reversed at a pixel whose image value is 1 in the embedded image. In other words, in the region where the pixel value of the embedded image is 0, the pattern-modulated image has a positive correlation with the Y component of the composite image and a negative correlation with the M and C components. In the region where the pixel value of the embedded image is 1, the pattern-modulated image has a negative correlation with the Y component of the composite image, and has a positive correlation with the M and C components.

  Next, a method for reproducing an embedded image from the composite image recorded according to the third embodiment will be described. Also in this embodiment, as in the second embodiment, a composite image is recorded on a transparent reproduction sheet 1003 having a transmittance distribution corresponding to the irregular pattern image Q (x, y) as shown in FIG. The embedded image is reproduced by being superimposed on the recorded material 1100. In this embodiment, the same sheet as that shown in the second embodiment is used as the reproduction sheet 1003. That is, the transmittance distributions Tr (x, y), Tg (x, y), and Tb (x, y) of the reproduction sheet 1003 are expressed by the above equation (16). By superimposing the reproduction sheet 1003 on the recorded material 1100, the embedded image is reproduced as a Y-B color difference signal.

  The principle of reproducing the embedded image by superimposing such a reproduction sheet 1003 on the recorded material 1100 will be described. As described above, in the region where the pixel value of the embedded image is 0, the pattern modulation image has a positive correlation with the Y component of the composite image and a negative correlation with the M and C components. In the region where the pixel value of the embedded image is 1, the pattern-modulated image has a negative correlation with the Y component of the composite image, and has a positive correlation with the M and C components.

  Therefore, by superimposing the above-described reproduction sheet 1003 on the recorded material 1100, the pixel on which Y ink is printed easily overlaps the black pixel of the reproduction sheet 1003 in the area where the pixel value of the embedded image is 0, Pixels printed with M ink and C ink tend to overlap with white (transparent) pixels of the reproduction sheet 1003. That is, when viewed in a macro manner, in a region where the pixel value of the embedded image is 0, the color is shifted toward B, which is a composite color of M and C. On the other hand, in the area where the pixel value of the embedded image is 1, it shifts to Y for the same reason. Therefore, since the chromaticity of the image is shifted in the Y or B direction according to the pixel value of the embedded image by overlapping the reproduction sheet 1003, the embedded image is reproduced as information modulated by the Y-B color difference. .

  As described above, in the third embodiment, as in the first and second embodiments, it is visually the same as the embedded image, and another image (embedded image) is embedded without deterioration in image quality. A composite image can be recorded. Further, by superimposing a predetermined reproduction sheet on the recorded material on which the composite image is recorded, it is possible to reproduce the image so that it can be easily recognized visually without requiring complicated signal processing. In addition, since the irregular pattern is used in the pattern image in the embodiment, it is difficult to estimate the embedded pattern from the synthesized image, as in the second embodiment.

  Furthermore, in this embodiment, as a feature not in the first and second embodiments, since a composite image is recorded by binary recording, a recording method in which control of multi-value density for each pixel is not easy as in an ink jet recording method. This printer has the advantage that it can be easily applied even when the printer is used.

(Fourth embodiment)
Next, a fourth embodiment according to the present invention will be described. In the first to third embodiments, an embedded image is reproduced by superimposing a reproduction sheet having a transmittance distribution on a recorded material, whereas this embodiment is an optical element having a thickness distribution. The difference is that playback can be performed.

  First, the image composition recording apparatus according to the present embodiment will be described. The configuration and processing procedure of the image composition recording apparatus in this embodiment are basically the same as those in the first embodiment, and only the configuration of the pattern image is slightly different. Hereinafter, the processing procedure in the present embodiment will be described with reference to the flowchart shown in FIG.

[First step (pattern generation)]
First, a pattern image Q (x, y) is generated in the first step S41. In the present embodiment, a stripe pattern image as shown in FIG. 19 is used as the pattern image Q (x, y). This pattern image Q (x, y) is obtained by arranging pixels having a gain of −1 and pixels having a gain of −1 to a color difference amount (Vr, Vg, Vb) in a stripe shape. In the example of FIG. The pixels are arranged in two rows in the y-axis direction and the pixels in which the gain 1 pixels are arranged in two columns in the y-axis direction are alternately arranged, that is, arranged in a cycle of 4 pixels in the x-axis direction. A formula for generating the pattern image Q (x, y) is shown in Formula (25).

If (int (x / 2) mod 2 = 0, Q (x, y) = − 1
If (int (x / 2) mod 2 = 1, Q (x, y) = 1
(25)
[Second step (pattern modulation)]
[Third step (pattern superposition)]
[Fourth Step (Color Correction)
Next, in the second step S42, the third step S43, and the fourth step S44, pattern modulation, pattern superimposition, and color correction are sequentially performed, and the finally obtained composite image is recorded. Since it is exactly the same as the embodiment, the description is omitted.

  Also in the present embodiment, a composite image obtained by combining the embedded image and the embedded image is recorded as in the first embodiment. This synthesized image is an image that looks visually the same as the embedded image, and the information of the embedded image is information that is completely or hardly visible.

  Next, a method for reproducing an embedded image from a composite image recorded in the present embodiment will be described. In this embodiment, an optical system in which a cylindrical lens array, a so-called lenticular lens, is formed into a sheet shape is used for reproducing an embedded image.

  FIG. 20 shows the configuration of the lenticular lens 2000, which has a structure in which a plurality of cylindrical lenses are arranged in parallel. The focal point of each cylindrical lens is on the bottom surface 2001. Further, the pitch of the cylindrical lenses is equal to the cycle (4 pixels in the present embodiment) in the x-axis direction of the pattern image Q (x, y).

  FIG. 21 is a schematic diagram when the lenticular lens 2000 is overlaid on the composite image on the recorded material 2002. On the composite image, the cylindrical axis of the cylindrical lens (perpendicular to the paper surface in FIG. 21) and the x-axis direction of the composite image, that is, the periodic direction of the pattern image Q (x, y) are overlapped, and further corresponding patterns The positions are adjusted so that the center of the Q (x, y) = 1 portion of the image is on the central axis of each cylindrical lens. Then, by observing from the upper surface of the lenticular lens 2000, the embedded image is reproduced.

  The reproduction principle of the embedded image in this embodiment will be described with reference to FIG. Since the Q (x, y) = 1 portion of the pattern image is aligned with the central axis of the cylindrical lens, when the image is observed from the direction perpendicular to the cylindrical lens surface, all the light is Q (x, y) of the pattern image. ) = 1. For this reason, the image of the part of Q (x, y) = 1 of the pattern image is seen, and the image of the part of Q (x, y) = − 1 does not contribute to the observation image at all. For this reason, an image in which (Vr, Vg, Vb) is added can be seen in the region where the pixel value R (x, y) of the embedded image is 0, and (Vr, Vg, Vb) in the region where R (x, y) = 1. ) Is subtracted. Therefore, an image in which the color difference is shifted according to the embedded image is observed.

  Here, in the first embodiment, the portion of R (x, y) = 1 in the embedded image did not overlap with the black portion of the reproduction sheet, whereas in this embodiment, R (x, y) = Since the portion 1 also appears as a color-shifted color, twice the color-difference contrast can be obtained compared to the first embodiment. Further, since the embedded image component is not blocked by the black pixels of the sheet, the embedded image component is observed with the original brightness.

  In the first embodiment, the embedded image cannot be reproduced unless the positions of the reproduction sheet and the composite image match a predetermined relationship. In the present embodiment, the lenticular, which is a reproduction optical element, cannot be reproduced. Even if the lens 2000 deviates from the predetermined positional relationship with respect to the composite image, the embedded image can be reproduced by moving the viewpoint. That is, for example, as shown in FIG. 22, consider a case where the pattern image Q (x, y) = 1 portion is slightly shifted to the right from the central axis 2201. In such a case, by shifting the viewpoint so as to observe from the direction of the arrow 2202, the focal position is shifted to a position of Q (x, y) = 1, and the embedded image can be correctly reproduced.

  Also in this embodiment, it is visually almost the same as the embedded image as in the first embodiment, and the embedded image can be recorded. In addition, by overlaying a predetermined sheet-shaped reproducing optical element such as a lenticular lens, the embedded image can be easily reproduced as a superimposed image on the embedded image and visually confirmed.

  In this embodiment, since a lenticular lens is used for the reproducing optical element, (1) the reproduction contrast is about twice that of the first embodiment. (2) the reproducing optical element and the composite image. Even if the phase is shifted, the position where the reproduction contrast is maximum can be obtained by shifting the viewpoint, and (3) the embedded image can be observed with the same brightness.

1 is a block diagram showing the configuration of an image composition recording / reproducing system according to a first embodiment. The flowchart which shows the image processing procedure in 1st Embodiment. The figure which shows the structure of the checkered pattern image used in 1st Embodiment. The figure which shows the kernel of the smoothing filter for smoothing the embedding image in 1st Embodiment. The figure which shows an example of the relationship of the embedded image in 1st Embodiment, the smoothing embedded image, a pattern image, and a pattern modulation image The figure which shows the other example of the relationship of the relationship between the embedding image in 1st Embodiment, a smoothing embedding image, a pattern image, and a pattern modulation image. 1 is a block diagram showing an example in which an image processing system of an image composition recording apparatus according to a first embodiment is realized by hardware. The figure which shows the frequency spectrum of the color difference component of the to-be-embedded image, the smooth | blunt embedded image, a pattern image, and a synthesized image in 1st Embodiment. Diagram showing visual chromaticity spatial frequency characteristics The perspective view which shows the structure of the reproducing | regenerating apparatus concerning 1st Embodiment. The figure which shows the pattern structure of the sheet | seat for reproduction | regeneration in FIG. The figure which shows the frequency spectrum of the image which piled up the sheet | seat for reproduction | regeneration on the recorded matter in 1st Embodiment The flowchart which shows the image processing procedure in 2nd Embodiment. The figure which shows the autocorrelation coefficient and power spectrum of the irregular pattern image used by 2nd Embodiment The figure which shows the frequency spectrum of the color difference component of the to-be-embedded image in the 2nd Embodiment, a smoothing embedded image, a pattern image, and a synthesized image. The figure which shows the frequency spectrum of the image which piled up the sheet | seat for reproduction | regeneration on the recorded material in 2nd Embodiment A flowchart showing an image processing procedure in the third embodiment. The flowchart which shows the image processing procedure in 4th Embodiment The figure which shows the pattern structure of the pattern image used in 4th Embodiment. The figure which shows the structure of the lenticular lens which is a sheet-like reproduction | regeneration optical element in 4th Embodiment. The figure which shows the reproduction | regeneration principle of the embedded image in 4th Embodiment The figure which shows the reproduction | regeneration state when the phase of the pattern image in the synthetic | combination image on the reproduction | regeneration optical element and a recorded material has shifted | deviated in 4th Embodiment.

Explanation of symbols

  DESCRIPTION OF SYMBOLS 101 ... CPU, 102 ... Image memory, 103 ... Image input part, 104 ... Program memory, 105 ... Image recording part, 106 ... Bus, 107 ... Image processing part, 701 ... Image memory, 702 ... Image memory, 703 ... Pattern generation 704... Pattern modulating unit 705... Pattern superimposing unit 706... Color correcting unit 707... Image recording unit 1000 1000 Image reproducing device main body 1003. Reproducing optical element), 2002 ... recorded matter.

Claims (4)

Gain information given to a predetermined color difference amount is multiplied by a predetermined color difference amount by a pattern modulated image signal obtained by modulating a color difference pattern image signal having a pattern expressed for each image unit with a first image signal . after the image signal superimposed on the second image signal is an image reproducing apparatus for reproducing the first image signal as a visible image from the recorded matter recorded as an image,
An image reproducing apparatus comprising: a sheet having a transmittance distribution of the same pattern as the pattern of the color difference pattern image signal and disposed on the recorded matter. Gain information given to a predetermined color difference amount is multiplied by a predetermined color difference amount by a pattern modulated image signal obtained by modulating a color difference pattern image signal having a pattern expressed for each image unit with a first image signal . after the image signal superimposed on the second image signal is an image reproducing apparatus for reproducing the first image signal as a visible image from the recorded matter recorded as an image,
An image reproducing apparatus comprising: an optical element having a thickness distribution of the same pattern as the pattern of the color difference pattern image signal and disposed on the recorded matter. Gain information given to a predetermined color difference amount is multiplied by a predetermined color difference amount by a pattern modulated image signal obtained by modulating the color difference pattern image signal having a pattern expressed for each image unit with the first image signal. An image reproduction method for reproducing the first image signal as a visible image from a recorded matter in which an image signal superimposed on the second image signal is recorded as an image later,
By arranging a sheet having a transmittance distribution of the same pattern as the pattern of the color difference pattern image signal on the recorded material, the first image signal can be reproduced as a visible image from the recorded material through the sheet. An image reproduction method characterized by the above. Gain information given to a predetermined color difference amount is multiplied by a predetermined color difference amount by a pattern modulated image signal obtained by modulating a color difference pattern image signal having a pattern expressed for each image unit with a first image signal. An image reproduction method for reproducing the first image signal as a visible image from a recorded material in which an image signal superimposed on the second image signal is recorded as an image,
By arranging an optical element having the same thickness distribution as the pattern of the color difference pattern image signal on the recorded material, the first image signal can be reproduced as a visible image from the recorded material through the optical element. An image reproduction method characterized by that.

Images