JPH02266390A - Method for mixing data with image - Google Patents

Abstract

PURPOSE:To mix much information while compressing data without spoiling image quality by utilizing visual integration effect and determining the arrangement of two picture elements in a dither image according to the data to be mixed. CONSTITUTION:When density information on an original image is quantized by using a dither method to perform artificial gradational expression, the arrangement of the two picture elements of the dither image corresponding to two cells which have a specific difference value between threshold values assigned to respective cells of a dither matrix is determined according to other data to be mixed. In a dither image processing stage as an image light and shade expressing means, the desired data are mixed, so the presence of the mixed data hardly appears in an image signal and a reproduced image to mix an extremely large amount of information with the image while compressing the data without spoiling the image quality.

Description

DETAILED DESCRIPTION OF THE INVENTION (Field of Industrial Application) The present invention relates to a method of synthesizing data into an image for mixing or separating data of a document or other communication into an image signal.

(Prior art) In recent years, the use of OA (office automation) equipment has become more networked as technology advances, and
The information handled is also diversifying to include not only document data but also audio and image data.

For example, facial photographs as information about two people are handled as image data, but in addition to this, names.

Attribute data such as age and address are also indispensable, and are generally treated as document data.

However, since image data and document data differ in their signal processing formats, they are generally transmitted and stored separately. In that case, you will have to manage two files even though there is only one file, which not only complicates the handling.

This method has the drawback of easily causing errors such as mismatch between the two.

When transmitting information between networked OA devices, it is essential to have a means of concealment to prevent leakage to third parties.

In view of these circumstances, various methods have been proposed in the past for mixing information such as document data into image signals.

One of them is C,8,Xydeas,B,Kosti
A data synthesis method for images using 1 model duro masking using C and 1.8teele etc.
ns, Cnmmun, C0M-32NOl, 1
984).

"Improvement of data synthesis method for images by 1 model duro masking using raw material, base materials, Hase-yo, and Chuyo-yo" (IBBE)
'francs, Commun C0M-36゜19
88) has been announced. Proposed in these papers7'
The method using the j-% duro masking method takes advantage of the fact that there is a strong correlation between adjacent pixels in a multivalued image, and the value of the intermediate pixel should be mixed into three adjacent pixels. This is to replace the code with a code corresponding to the data.

By using this method, both an image and a document can be stored at the same time, which is not only extremely convenient, but also makes it possible to conceal and transmit even more burdensome information under the guise of transmitting an image. It can also be used as a type of encrypted communication method.

However, although this modulo-masking method can completely restore the original image as an output image, it is unsuitable for data compression because it necessarily retains multilevel information.

It lacks practicality as an image processing method itself.

On the other hand, 9 inventors etc.
The method proposed in "Embedding attribute information in dithered images using multi-stage quantization method 1" presented at K.K. There is room for improvement since this involves deterioration in image quality.

The applicant of the present application has already used the proposed method Density Butter 7 method to synthesize data into 91 images (Patent Application No. 62-4731)
0), by controlling the density pattern using other data when converting the multi-valued density information of each pixel into binary density information).

Other data is embedded in the image. According to this method, an extremely large amount of data can be synthesized without any image disturbance, and it is very effective as a method for compositing data into one image. However, due to the size of the density pattern (nXn pixels), The disadvantage of expressing information for pixels is that the amount of data increases by n times compared to a systematic dither image.

(Object of the Invention) The present invention has been made to eliminate the drawbacks of the conventional image data composition method as described above, and further reduces the amount of data without compromising the quality of one image. It is an object of the present invention to provide a method for synthesizing data into an image, which can mix an extremely large amount of information while compressing the image.

(Summary of the Invention! Hole 9 to Achieve This Object) The present invention describes a dithering method used to express the density of an original image into multiple gradations, and in particular a dithering method using a systematic dithering method that facilitates data compression. Focusing on a set in which the threshold difference of each element is a predetermined value and the output dithered image corresponding thereto, other data is embedded using the statistical characteristics of the dithered image.

Utilizing the visual integration effect that the overall density remains the same even if two pixels in a dithered image are swapped, the arrangement of the two pixels in the dithered image is determined based on the data to be mixed. In other words, this should be replaced.

This method can significantly compress and reduce the amount of Ii image data compared to conventional methods of combining data into images using the Roman masking method and the density pattern method. Even when compared, there are signs that the image quality deteriorates less.

(Example) The present invention will be described in detail below based on the illustrated example, but before proceeding, a systematic dither method will be briefly explained to facilitate understanding of the present invention.

Systematic dithering is a method for displaying the intermediate tones of an image.
Binary display is often used for percentage. Specifically, Figure 2 (,
As shown in 1), the entire screen of the original image gll consisting of a matrix-like dot arrangement is regarded as a set of partial areas Fn consisting of fiXn dots.

As shown in FIG. 6(b), predetermined ft: , n in the dither matrix corresponding to the partial area Fn
1 and is used as a threshold value for each dot.

That is, an nxn dither matrix Dn, where d-) is an element in p rows and q columns, and ranges from 0 to n! Let the integer at q −11 indicate the ordinal number of the threshold that appears only once.

At this time, the brightness 1 [IuY of the pixel in the Uth row and Vth column on the display screen is .

Threshold value pq created from d(n) in n in the dither matrix defined as p=umodn q=ma mod n T = (d pq(n'+ 1/2) R/ n
”, and if Iuv≧T then l, Iuv<T
If so, 0 is output. However, R is the luminance range of the multivalued image.

If each pixel of the nine-dither image 2 represented in this way is minute, it will be recognized by the human eye as having an intermediate density according to the ratio of black and white,
It becomes possible to express the matrix element n3+1 gradation.

Density gradation expression using the dither method can be applied not only to black and white images but also to color images.

Regarding colors, it is best to perform the density calculation as described above for each of the three primary colors.

Furthermore, in the dither method, the means for encoding an image is O#"1
The technology is being applied not only to 200 million digital signals but also to multilevel digital signals, and research is underway to obtain images close to the original with as little data as possible.

In general, the number of pseudo gradations Leq that can be expressed with an L-value dither matrix of nxn size is Leq = (L-1) n +1 ·-- (1), and in the case of 200 million dithers with L = 2, Leq = n ”-1.

In addition, the size of n is determined by considering the resolution and visual characteristics of the display device that is normally used. , each of the generated dithered images! The gradation is determined only by the proportion of black and white in the matrix, and is unrelated to the order in which they are arranged. That is, there is a certain degree of freedom in selecting dithered images that represent the same gradation.

When creating a dithered image, the present invention utilizes the degree of freedom of the dithered image based on the visual integral effect as described above, and replaces pixels in the image in accordance with data to be mixed.

That is, by focusing on a combination of cells in which the threshold difference between two cells in the dither matrix is a predetermined value, and setting the combination of pixels in the dithered image corresponding to that combination according to the data to be mixed. Synthesize data into image signals.

In order to simplify the explanation, this embodiment exemplifies the case of a binary dither method.

FIG. 1 is a block diagram illustrating a method for synthesizing data into an image according to the present invention.

In the same figure, 1 is the original image, and its partial area Fnt
A jlR extraction quantization block 3 reads the density of each pixel, and a 1st gradation comparison block 4 derives a numerical value indicating the lp1 rank of each pixel and outputs this to a dithering block 5.

Reference numeral 6 is a coding block, and has a dictionary 7 that stores data on characters, symbols, and codes that may be used in advance. When document data such as characters to be mixed in is input, this encoding block 6 receives a code 1 representing the corresponding data in the dictionary, such as a JI8 code or an ASCII code.
It is encoded into a 0''@110 bit sequence according to the I code or the like, or it is further encrypted to generate a numerical sequence Ki, which is input to the dithering block 5.

The dithering block 5 combines both the image signal input from the gradation comparison block 4 and the document data input from the encoding block 6, as will be described in detail later.

Next, the threshold difference of the dither matrix and the output dither image based on it will be described.

Figure 2 (il is an example of quantizing each pixel of the original image, (b
) is an example of a 4×4 dithered IJ matrix, which is a spiral-shaped tither matrix, and as shown in FIG. Add a sign. (d) is the original picture @Fn and Tiza Matrix D
This is the dithered image obtained as a result of comparison with n, and each pixel has 1cyi (i=0°1, . . . ) as shown in (e).
..., 15) Attach a code tube.

Now, the threshold closeness k is defined as follows.

k=x, -x-(k=1, 2, 3, -=, 15
)'' 1 However, xj is the j-th threshold, Xi is the i-th threshold, and x H<X J@For example, i=6. j
=3, in the spiral dither matrix shown in FIG. 2(b), =xj-x i=x, -x6=9-1=8. The threshold margin in this case is 8. In this way, the set of Xt,
)1, in this case it becomes (1,9).

Similarly for =8 in (o, a)lll, (z, to)
.

,(3,11),,(4,lz),,(s,13),
, (6,14)s, ('yet・5), and this set can be expressed as 8 = ((o, 8), (1, 9),...
(?, 15)) To table. There is a similar set 8k for k=1 to 15.

Also, the binary code output by the set of thresholds (Xt,)J)s is expressed as (yi, yj)k.0 At this time, the output is 0
'1-)'j)k is uniquely determined by the input signal, that is, the original image 1 and the threshold difference k, and the combination is (0,0L
There are 04 ways (0,1), (1,0), (1,1).

For example, same as before (k=8, i=6, j=3
Then, the dithered image shown in Fig. 2(d) is (yl, yj)k=(y,,y,),=(x,o) and T
o? ), k = 8, i = 1, j = 14
When (yi4j)k=(y, *y14), =(o
, O).

Next, a method for synthesizing characters with the dithered image described above and a method for restoring the same will be described.

First, the dither matrix Dn to be used and the threshold difference k
Determine. Based on this dither matrix Dn and the threshold difference k, the binary output (
y i = y j) y is (0, 0) as mentioned above
, (0,1), (1,0), (1,1).

Even if the two dots of the output (yi, yj)k applied to the dithered image are interchanged, the density of one dithered image remains the same as a whole. At this time, for (0, 0) and (1, 1), the arrangement of the dithered image does not change even if two dots are exchanged, but for (0, 1) and (1, 0), the arrangement of the dithered image does not change when two cells are exchanged. The arrangement of greatness changes. So this (o, i)
Data is synthesized by focusing on the two sets of (1, 0).

However, it is difficult to determine the threshold k.

The details will be explained later, but in the dither image, many (1,0) patterns are generated (x t * X J) k
Note that the number of pairs differs depending on k.

First, find out what (Yi-7j)1 is, and if y 3
-Yj)1=(o,o) or (yH,yj)1=(i
-1), data is not combined into this set.

Combine data in other cases. For example, if the bit you want to combine is 0 and (yi, yj
)1-(0,1), output 11 of them.

(y i, y 1 )) If c=(1-0), swap the cells and (y i, y 3 ) k=(0, 1)
Output.

- In the unlikely event that the bit that can be combined is 1, and (y 1 - y
j)1=(0,1), then replace the dots and write (y
i, yj)), = (1, O) and +(yl-yj)
)1=(1,O), then output that t.

According to the above procedure] when the bit to be synthesized is 00, it is (0, 1)
, 1, data is mixed as (1, 0).

To restore the document data synthesized using this method, the above synthesis procedure is reversed. That is.

( x s e X J
) The patterns corresponding to y are (0,0), (1,1)
Then skip = (yH-yj)k=co,
1) Then output b = o and -(y i-y
3) If k:H('-0), output b,=1. Document data can be decoded from this bit series.

Next, we will examine the amount of data that can be synthesized using the above method.

For example, if one dither matrix consists of 4x4 pixels,
If O<k=8, the number of pairs with close threshold differences existing in the l dither matrices is 8 or less. Of these, (0, 1) and (1, 0) that can be used for data synthesis are considered to be about half of the total.

Therefore, since 8/2 = 4 bits of data can be synthesized in the dither matrix, the size of one entire image is 256X.
If it is a 258 image, it will be tonashi, and if it is a character expressed in 8 bits, it will be about 2k.
bytes can be synthesized.

However, the same data synthesis method using the density pattern method (about 70 bytes of characters can be synthesized in a 256 x 256 pixel image, and a large amount of data can be mixed in compared to the present invention. In the Density Butter 7 method, if one dither matrix is 4×4, the amount of data required for one image is 16 times that of the systematic dither method, and the amount of original data is extremely large.

Therefore, although the amount of document data that can be synthesized is smaller in the present invention than in the density pattern method, the amount of original data can be much smaller, making it an extremely convenient data processing method.

The above steps can be summarized as follows.

[Synthesis procedure]

8TEP1: Dithering) A partial region corresponding to the risk is extracted, and a set of threshold values (X., X) and a set of corresponding output values (yi, yj) are created.

8TEP2: If (7H-7j)1=(1-0) or (yi, yj)1=(o, i), synthesize the data as follows. That is, if b [=0 then (y H-y j)k”(0-1)*
If br=1, 5TEP2 is executed for all pairs in the (71-7j)k=(1,0) matrix, and 8TBP1 and 8TEP2 are repeated for the entire image.

[Decryption procedure]

8TEP1: Extract the partial region corresponding to the dither matrix, and extract the set of threshold values (xt e X J ) k and the corresponding set of output values (yi,
Make yj)k.

8TBP 2 : ()'i, 7j) 1 = (1, 0) or (y H, y))) If = (0, 1), then the data is decoded as follows. That is, if 0'1-7j), = (0, 1), then br=O1-Koshi is output as composite data.

- If (yl, y3)1=(1, O), then compose br-1 as composite data.

Collect 8TBP2 for all pairs in the matrix and repeat 8TBP1.5TEP2 for all 2 strokes.

As explained above, according to the present invention, it is possible to synthesize a large amount of information while compressing data.

However, it is clear from the procedure that combining two document data into a dithered image disturbs the dithered image. Therefore, in order to obtain good image quality when outputting a single image, it is necessary to restore the single image.

By the way, the output of the dithered image (yi-
Comparing the numbers of (1,0) and (0,1) among yj)), (1,0) is the majority, and (0,1) patterns are extremely rare. This ratio varies depending on the type of dither matrix, but it depends on the threshold difference k, and as k increases, the probability of becoming (1, 0) becomes extremely high.

fi, ttf standard image data page 8IDBAf)
Girl.

Moon, for each image of real (256 x 256 pixels, 8 bits) = 1, 2,...
..., at 15 K (y H= y j) y is (
Among the patterns that are (0,1) and (1,0), (0,1
) is found as shown in Figure 3. (a) Buyer type.

Regardless of whether (b) spiral type or (C) halftone type dither matrix is used, if the appearance ratio of (0, 1) patterns is close to each other, it is due to the local redundancy of one image. 0'1-7j)1=(1, o).

For example, the Bayer type K times 8, 12 and the halftone type -12 have the characteristic that the appearance ratio of the (o, i) pattern is 0.

Then, using this % sign, one stroke of (y 1 * yj)
All parts where k=(0,1) are converted to (1,0). Through this processing, it is possible to restore an image that is extremely close to the dithered image before the single-character information overlap.

It is also possible to apply the present invention to encrypted communication means, and FIG. 4 is a block diagram showing an embodiment in that case.

In the same figure, the pixel signal extracted from the original image signal from 8 companies and the signal obtained by encrypting the data 9 to be mixed in by the scrambler IIK using the encryption key 10 are processed in the encoding circuit 12 as described above. A dithered image 913 is generated and transmitted to the desired user 14.14. After receiving this, the user uses the decryption circuit 15 to separate the original image signal 16 and the encrypted signal based on a predetermined rule, and then de-scrambler 17 converts the encrypted signal into the original image signal using a predetermined encryption key 18. Extract data 19.

Furthermore, 8? :, each block in the system, for example, the scrambler 11. Since the descrambler 17 or the encryption method can be easily realized using existing technology, a detailed explanation will be omitted.

We conducted extensive trials based on the method described above and obtained very good results.

For example, images from the standard image database 8IDBA (25
6 x 256 pixels, x 8 bits), n = 4, 2
When executed on a value image, it is the result of converting to ASCII code and composing approximately 2 kbytes of characters.

The difference between the reproduced image when nothing is synthesized using the conventional systematic dithering method - the next thing you can hardly see.

11 Order 1 By changing the pixel arrangement of the image data generated by the method of the present invention, #'Iit'! Experiments have also confirmed that it can be restored to the same one.

Furthermore, when data is mixed into an image based on the present invention,
When a long input of bit 0 occurs as a data string, the output (y 6 , y 3 ) y at this time is a series of (0, 1), which increases the disturbance of the 91 image. In this case, for example, just before synthesizing the data, an operation of inverting and inputting each bit may be added, and this has also been confirmed through experiments.

This method may be used not only in the case described above but also in a wide range of cases where there is a possibility that a certain regular data string may give an undesirable shadow 41 to the reproduced image.

There are various ways to use this method of secretly transmitting or storing Kiiig II data, but one example is to build a system that combines personal information with an individual's facial photo data and manages it all at once. The data is recognized only as a face photo by a third party, and only those with a matching encryption key can know the next data, making it possible to create a system that ensures complete confidentiality.

Furthermore, if the present invention is applied to facsimiles or other image signal transmissions that are used on a daily basis, data related to one image can be transmitted together with the image, regardless of the presence or absence of encryption means, thereby simplifying the transmission process. It can be achieved. Furthermore, it is clear that the data to be mixed in is not limited to mere characters, but the audio signal ta may be any other data such as an image signal.

Furthermore, although it has been explained that K(0, 1) is used when combining bit 0 and (1, 0) when combining bit 1, the plain is the same even if this is reversed.

In addition, in the embodiments shown in 1 or more, in order to simplify the explanation, 2
Nine mentions the case of value images.

The present invention is not limited to this, and can be applied to multivalued images, color images, etc., and various modifications may be made to the configuration of the apparatus and system used when implementing the present invention.

(Effects of the Invention) As described above, the present invention mixes desired data in the dither image processing process, which is a means of expressing the density of one image, so that the presence of the mixed data appears on the image signal and the reproduced image. This method is extremely effective in providing a means for synthesizing a large amount of information into an image while still compressing the amount of data without compromising the image quality.

By using the present invention, the image signal and other data can be handled by the same means, so the information media can be greatly simplified.Furthermore, since the presence or absence of mixed data is difficult to show on the screen, it is possible to handle the image signal and other data by the same means. It is also effective in transmitting and storing important data in a confidential manner.

[Brief explanation of drawings]

FIG. 1 is a block diagram showing an embodiment of the present invention, FIG. 2 is a diagram explaining a general dithering method, and FIG. 3 is a diagram showing (yi-)'j)k=(0-1 ), and FIG. 4 is a block diagram showing an embodiment in which the present invention is applied to encrypted communication means. 1...Original image, 2...
...Dithered image. 3...Quantization block, 4...
Light: gradation comparison block, 5...
Dithering block, 6... Encoding block. 7・・・・・・・・・Dictionary. Patent Applicant: Toyo Tsushinki Co., Ltd. Koshi 1 Figure 2 (α) Bayer Tatami (b') Uzu Ichi (C) 1st Silk Dot Diagram + Prisoner

Claims (5)

[Claims] (1) When expressing pseudo gradation by quantizing the density information of the original image using the dithering method, the dithering method is applied to two cells such that the difference between the threshold values assigned to each cell of the dithering matrix is a predetermined value. A method for synthesizing data into an image, characterized in that the arrangement of two pixels of the image is determined based on other data to be mixed. (2) When separating the data synthesized using the above method,
2. The method of synthesizing data into an image according to claim 1, wherein other mixed data is identified by the arrangement of two pixels in the dithered image. (3) The values of two pixels in the dithered image are yi and y
j, and if yi = yj, the data is not combined, and in other cases, if the data to be mixed is 1, yi > yj, if it is 0, yi < yj, or the data to be mixed is 1. Claim 1 or 2, characterized in that the pixel arrangement is determined by replacing the pixels as necessary so that yi < yj, and if 0, yi > yj.
Data synthesis method described in section. (4) When separating the data synthesized using the above method,
If yi=yj, the data is not combined; otherwise, if yi>yj, the mixed data is 1, and yi<yj
If so, identify the mixed data as 0, or yi
4. The method of synthesizing data into an image according to claim 3, wherein if <yj, the mixed data is determined to be 1, and if yi>yj, the mixed data is determined to be 0. (5) A patent characterized in that after identifying the other mixed data, pixel placement is determined by replacing pixels as necessary so that yi > yj in order to improve image quality. A method for synthesizing data into an image according to claim 4.