JP2001285628A - Image forming method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a half-tone image forming method that can improve the accuracy of gradation reproduction for an original image. SOLUTION: Before transmission of image data (S205), an image consisting of intermingles pixel with each of N levels is formed from multi-gradation distribution image A (S201). After the image is transmitted, an m-value half-tone image (S209) can be obtained by simply applying threshold processing to the image data.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an image forming method, and more particularly to a halftone image forming method capable of improving image quality when performing two-stage halftoning.

[0002]

2. Description of the Related Art When outputting image data, it is necessary to execute a gradation reduction process (halftoning) of the image data in accordance with the number of gradation levels that can be output by the apparatus. That is, even when one image is output by a plurality of different output devices, it is necessary to perform a different halftoning process for each of the devices. Therefore, each output device needs to separately have the function of the halftoning process.

FIG. 37 is a diagram showing a system for outputting a specific image by a plurality of output devices. Referring to the figure, the system includes a host 501, a network 503, and a plurality of terminals 505a to 505d that receive image data. An image A to be distributed (distribution image) A is input to the host 501 and sent to each of the terminals 505a to 505d via the network 503.

[0004] Output devices b and c are connected to a terminal 505a, and an output device d is connected to the terminal 505b. The output device e is output to the terminal 505c.

[0005] The host 501 performs an image conversion process before transmitting the distribution image A. This conversion processing is mainly compression processing for reducing the capacity of image data. Each of the terminals 505a to 505d performs an image conversion process for restoring an image. This processing is mainly a compression decompression processing.

The output devices b to e perform a gradation level reduction process (for example, a binarization process or a multi-value process) of image data for output.

[0007] By such processing, output images BE can be obtained from the output devices be.

No matter how simple the output devices be are, if the number of output levels differs among them, it is essential to perform the halftoning process in each output device. Therefore, in each output device, a load is imposed on a memory and resources, and a calculation time is required.

FIGS. 38 to 41 are flowcharts showing each of the four modes of the image distribution process.

In the embodiment shown in FIG. 38, a distribution image A of n gradations (n levels) is compressed in step S601 without being subjected to halftoning processing, and
603. On the receiving side of the image data, decompression processing of the received data is performed in step S605. Further, in step S607, a halftoning process for reducing the gradation of the image data from the n → m level is performed. Thus, the receiving side can obtain an m-level output image B.

In the example of FIG. 39, an n-level distribution image A is first subjected to an n → m level halftoning process on the data transmission side (S611). And
In step S613, compression processing is performed.
At 15, transmission takes place. On the receiving side, the received image data is decompressed in step S617 to obtain an m-level output image B.

In the example shown in FIG. 40, an n-level distribution image A is converted into an nn-level image on the image data transmitting side (S621). Then, the converted image is compressed in step S623 and transmitted in step S625.

On the image data receiving side, the image data is decompressed in step S627, and the process proceeds to step S627.
At 629, halftoning processing of the nn → m level is performed. As a result, an output image B having m-level gradation can be obtained.

In the example of FIG. 41, an n-level distribution image A
Is converted to an nn level image (S631). Then, compression processing is performed in step S633, and transmission is performed in step S635.

On the receiving side, decompression processing of image data is performed in step S637, and nn is determined in step S639.
A gradation restoration process from a level to an n level is performed. next,
In step S641, halftoning processing for reducing the gradation from n level to m level is performed, and an m-level output image B is obtained.

Considering the cost of image transmission, it is desirable to reduce the image capacity as much as possible. In addition, when an image is finally output by an output device, a half-toning process is performed.
At that point, the image capacity decreases. Therefore, depending on the purpose, as shown in FIG. 39, it is efficient to perform the halftoning process from the beginning on the image transmission side and transmit the image.

However, for that purpose, the output devices used on the receiving side need to have the same number of output gradation levels. If they are not the same, it is necessary to perform different halftoning processes on the receiving side as shown in FIG.

In such a case, if the number of gradation levels of an image is reduced beforehand on the transmitting side to some extent and transmitted, and the output device is further reduced to the individual number of gradation levels, the halftoning processing can be performed. There is also the idea that it can be made a little lighter and that the data capacity during transmission can be somewhat reduced (see FIG. 40).

However, in the conventional half-toning method, direct conversion is difficult when converting an image having a reduced number of gradation levels to an image having a further reduced gradation level. Therefore, as shown in FIG. 41, it is necessary to once restore a multi-tone image (S639), and the processing on the receiving side becomes heavy. In addition, such processing is not preferable in terms of gradation reproduction accuracy.

[0020]

SUMMARY OF THE INVENTION The present invention has been made to solve the above-mentioned problems, and has as its object to provide an image forming method capable of improving the accuracy of gradation reproduction.

[0021]

According to one aspect of the present invention, there is provided an image forming method, comprising: forming a first halftone image in which pixels of each level of N values are mixed from a multi-tone image; The method includes a first step of creating, and a second step of creating a second halftone image having an m value by thresholding the first halftone image created in the first step.

According to another aspect of the present invention, an image forming method includes a first step of creating a first halftone image in which pixels of each level of N values are mixed from a multi-tone image, and a first step. Receiving the created first halftone image by a plurality of terminals having different output gradation numbers.

According to still another aspect of the present invention, an image forming method includes: a first step of forming a first halftone image in which pixels of each level of N values are mixed from a multi-tone image; A second step of distributing the first halftone image created in the above on a network on the assumption that the first halftone image is received by a plurality of terminals having different output gradation numbers.

[0024]

BEST MODE FOR CARRYING OUT THE INVENTION Hereinafter, one embodiment of the present invention will be described.
The halftone image forming method will be described. In the present embodiment, an N-valued multi-valued halftone image is created so that pixels of each level of the N-value are mixed (halftoning A), and the N-valued multi-valued halftone image is created. By thresholding with an appropriate number of thresholds (halftoning B), an arbitrary m value (m
<N) to form a halftone image.

Once an output image is formed by the halftoning A described above, the image is received on the receiving side via a network, and a simple process called halftoning B is performed to obtain a different number of gradation levels. An image can be output by each of the output devices.

FIG. 1 is a flowchart showing the flow of processing of the halftone image forming method according to the present embodiment.

Referring to the figure, in the present embodiment, an image A to be distributed (image having an n-level gradation number) is subjected to a halftoning A process in step S201. The halftoning A process is a process of reducing the distribution image A having the n-level gradation number to the N-level gradation number. Further, in the halftoning A, an image in which pixels of each level of N values are mixed from a multi-tone image (distribution image A) is created. Half Toning A is
This is a process of outputting one gradation level of an input image using three or more output states.

In step S203, the N-level image subjected to the halftoning A is compressed. In step S205, the compressed image is distributed via the network. In step S207, the compressed data is decompressed on the image data receiving side.

In step S209, the received N
Halftoning B for images with level gradation
Is performed. Thus, an image having N-level gradation is converted into an m-level (n>N> m) image. This halftoning B is a conventional simple gradation process using a threshold value.

Thus, an output image B having m-level gradation can be obtained. FIG. 2 is a diagram for explaining a specific example of an image distribution method using the halftone image forming method according to the present embodiment.

On the transmitting side, the sample image A having 256 gradations is converted into an image having 12 gradations by halftoning A (S201). Then, the image is received by a plurality of terminals via the network. In each terminal, half toning B
Is used, an image having a gradation corresponding to the characteristics of the output device can be obtained.

Here, specifically, in step S301, halftoning processing of 12 → 6 levels is performed, and a six-valued output image B1 is obtained. In step S303, halftoning of 12 → 4 levels is performed, and a quaternary output image B2 is obtained.
In step S305, halftoning of 12 → 3 levels is performed, and a ternary output image B3 is obtained. In step S307, 12 → 2
Level halftoning is performed, and a binary output image B4 is obtained.

The processing performed in each of the half-toning A and the half-toning B will be described below.

[First Example of Half Toning A] FIG. 3
FIG. 3 is a diagram schematically illustrating a process of an image processing apparatus that executes halftoning A. This image processing apparatus converts an input signal of multiple gradations into a plurality of gradations of three or more gradations (hereinafter, referred to as an output state). Referring to the figure, in step S101, a pixel value (input signal) of one pixel in image data to be processed is input.

In step S103, based on the input signal, the appearance probability of each of a plurality of states that the output pixel can take is set. The setting of the appearance probability may be performed by calculation or by referring to table data stored in advance. Alternatively, both may be adopted. What is important is that the appearance probability is set based on the input signal before the output state is determined. Hereinafter, a specific example of setting the appearance probability will be described.

Assuming that each output state having the appearance probabilities of d1, d2, and d3 represents output values (densities) of 0, 0.5, and 1.0, respectively, the average output value is the output value of each output. The value is multiplied by each occurrence probability, and z = 0 × d1 (x) + 0.5 × d2 (x) + 1.0 × d
3 (x).

When the output value is set to reflect the input value, the output value z = input value x, and x = 0 × d1 (x) + 0.5 × d2 (x) + 1.0 × d
It is necessary to set each of d1, d2 and d3 so as to be 3 (x).

On the other hand, the sum of the respective appearance probabilities should be 1, and the relationship represented by the following equation must be satisfied: d1 (x) + d2 (x) + d3 (x) = 1.

The method of setting d1, d2, and d3 can be freely selected. For example, the peak value is obtained at each representative output value (0, 0.5, 1.0) as shown in FIG. It is also possible to set a bell-shaped function with.

That is, referring to FIG.
Assuming that d1, d2, and d3 are in the range of 1.0 and the appearance probabilities for the outputs of 0, 0.5, and 1.0, respectively, the function d1 (x) has a peak at the input value x = 0. Is a bell-shaped function, where d2 (x) is x = 0.5
Is a bell-shaped function with a peak at d3 (x) is x
= 1.0 A bell-shaped function with a peak at 1.0.

In the example of FIG. 4, the sum of the appearance probabilities with respect to the input value x (d1 (x) + d2 (x) + d3
(X)) is not always 1. Then, d1 (x)
+ D2 (x) + d3 (x) = 1, so that d1 (x) + d2 (x) + d3 as indicated by the arrow in the figure.
The value of (x) is compressed. That is, d1: d2: d
The sum is normalized to be 1 so that the ratio of 3 does not change.

Further, as shown in FIG. 5, the parameters of the function are adjusted so that the output value z becomes substantially equal to the input value x.

That is, as shown in FIG.
The parameters of the function are adjusted so that a linear relationship is established between and the input value x.

In step S105, one state is determined from the plurality of states based on the appearance probabilities and thresholds (parameters) of each of the plurality of states.

In step S107, step S1
A signal indicating the state determined in 05 is output. Also,
In step S109, a value obtained by correcting a predetermined initial threshold based on the presence or absence of each of the plurality of states determined in step S105 is output.

FIG. 6 is a block diagram showing the configuration of the image processing apparatus. Referring to the drawing, the image processing apparatus outputs any one of N types 1 to N as an output state of one pixel. The image processing apparatus includes an appearance probability setting unit 151 that calculates appearance probabilities d1 to dN of each of a plurality of states based on an input signal, and subtracts a corresponding correction threshold value from each occurrence probability to obtain a subtraction value. A selection unit 153 that selects a state corresponding to the maximum value among the above, an output unit 155 that outputs a signal indicating the state selected by the selection unit 153,
A calculation unit 157 that calculates a feedback value for each state of 1 to N, and a β multiplication unit 159 that multiplies each of the calculated feedback values by coefficients β1 to βN.
And an initial threshold value output unit 161 that outputs initial threshold values t1 to tN for each state, and a subtractor 163 that subtracts a feedback value multiplied by a coefficient from the initial threshold value. The output of the subtractor 163 becomes the correction threshold value and is output to the selection section 153.

FIG. 7 is a block diagram showing a more detailed configuration of the block diagram of FIG. In this block diagram, there are three types of pixel states output by the image processing device (N =
3)

Referring to FIG. 7, the image processing apparatus includes an input unit 171 for inputting a pixel value as an input signal, an appearance probability setting unit 173 for setting appearance probabilities d1 to d3 of each state based on the input signal, A subtractor 175a for subtracting the modified threshold values t'1 to t'3 from the appearance probabilities d1 to d3 in each state.
Selecting unit 1 for selecting the largest one of the outputs of subtractors 175a to 175c and the state corresponding to the largest one
77 and an output unit 179 for outputting the selected state Ss
And an inverting unit 181 for inverting the values S1 to S3 based on the selection result of the selecting unit 177, and subtracters 183a to 183a for subtracting the modified threshold values t'1 to t'3 from the output of the inverting unit 181.
83c and outputs fb1-f of subtracters 183a-183c.
β multiplying unit 185 for multiplying b3 by coefficients β1 to β3
And an initial threshold value generating section 187 for outputting initial threshold values t1 to t3, and a β multiplication section 1 based on the initial threshold values t1 to t3.
And 85 subtracters 189a to 189c.

The outputs of the subtracters 189a to 189c become the modified threshold values t'1 to t'3. Further, the selection unit 177 instructs the inversion unit 181 to set “1” in accordance with the selected state.
And outputs “0” corresponding to other states.

As shown in the figure, an appearance probability and an initial threshold value are set for the number N of states to be output, and feedbacks fb1 to fb3 for correcting the threshold value are set.
Exist as many as the number N of output states. Since the sum of the appearance probabilities (sum of d1 to d3) is set to “1”, the number of appearance probabilities may be N−1. This is because the remaining one occurrence probability can be calculated from the N-1 occurrence probabilities.

If the sum of the initial threshold values t1 to tN or the coefficients β1 to βN is constant, the total number of feedbacks can be reduced to N−1, and the amount of calculation can be reduced. Can be.

The initial threshold value output from the initial threshold value generating section 187 may be changed for each output state.
It may be a constant value (t1 = t2 = t3). Further, the initial threshold value may be a pattern like a dither matrix. Such a pattern may be different for each output state or may be the same.

FIG. 7 shows an example of a method for determining the output state to appear. However, if the output state can be determined from the relationship between the threshold value and the appearance probability,
Other methods may be used.

Further, β multiplied by β multiplication section 185
May be changed for each output state or the same. Further, β may be changed according to a purpose such as changing the γ characteristic.

FIGS. 8 and 9 are diagrams showing how the image data is ternarized by the image processing apparatus. In the figure, processing of an input signal is performed in each of steps (1) to (4). Here, it is assumed that the value of the input signal is the same in all steps.

Referring to FIG. 8A, an appearance probability is set based on an input signal. As described above, table data for the input signal may be used for this setting,
The appearance probability may be set by a function based on the input signal. Here, it is assumed that the appearance probabilities are set to 0.3, 0.5, and 0.2 in each of the states 1 to 3.

The initial threshold value generator 187 generates an initial threshold value of 0.5 corresponding to all states. The subtracters 189a to 189c subtract the output of the β multiplication unit 185 calculated last time from the initial threshold value.
Since this is the first process, the previous β multiplication unit 185
Is 0.

The subtractors 175a to 175c subtract the correction threshold from the appearance probability in each state. Thus, the subtractor 175a outputs -0.2, the subtractor 175b outputs 0, and the subtractor 175c outputs -0.3. Then, since the output of the subtractor 175b is the maximum, the selector 177 selects the state 2 (S2). Then, 1 is output to the inverting unit 181 as a value corresponding to the selected state 2, and 0 is output as a value corresponding to the other states.

The inverting section 181 inverts the output result from the selecting section 177 and outputs the result. Each of the subtractors 183a to 183c subtracts a correction threshold from the output of the inverting unit 181. Accordingly, the subtracters 183a to 183c output values of 0.5, -0.5, and 0.5, respectively.

The β multiplication unit 185 includes subtractors 183a to 183
The respective outputs of c are multiplied by coefficients β1 to β3. Here, since β1 = β2 = β3 = 0.5, the β multiplication unit 185 outputs values of 0.25, −0.25, and 0.25.

Subsequently, in the next step (2), the output of the β multiplication unit 185 calculated in the previous step (1) is input to the subtracters 189a to 189c. And
The same processing as described in (1) is also performed in step (2). The same applies to steps (3) and (4) in FIG.

In the processing shown in FIGS. 8 and 9 and FIG. 10 to be described later, feedback to a single unprocessed pixel is performed for simplicity. The weighted feedback is performed on the pixels.

FIG. 10 is different from FIGS. 8 and 9 when the number of output states is N = 3, the input signal is constant, and the appearance probabilities of the three states are 0.4, 0.5, 0.
It is the figure which showed the calculation example in case of 1 in each of the steps (1)-(12). Here, the initial threshold values are all constant at 0.5, and the coefficients β are all constant at 0.5.

According to the example of the halftoning, a plurality of output states can be output so as to have an average property corresponding to the input signal. In addition, the output states can be arranged such that the plurality of output states are uniformly distributed.

For example, when the gradation of an image is expressed by a plurality of output states, a visually uniform texture can be provided while always mixing the plurality of output states.

Further, depending on the example of the present halftoning, there is no problem of texture shift or pseudo contour which is caused by the error diffusion method. In the case where the type of color material is changed corresponding to a plurality of output states, any combination of color materials can be arranged according to the number of output states, and the combination of color materials can be changed. They can be arranged evenly and without gaps. By these effects, the quality of the output image can be improved in the present halftoning.

Next, the reason why the halftoning A produces such an effect will be described. For example, when a method such as a conventional error diffusion method is adopted, processing is performed so as to minimize the average error. This is the same regardless of the simple binarization or the error diffusion processing in the vector space. That is, the function is such that the target pixel is in an average state with the peripheral pixels, and the output state of the target pixel has a strong correlation with the output state of the adjacent pixel.

On the other hand, according to the halftoning, the appearance probability of each output state is specified in advance, so that the achievement of the appearance probability is prioritized. That is, the output does not depend on the output state of the adjacent pixel. Then, it is prioritized that the output state appears with the set probability.

For example, assume that the input signal was constant at 0.5. At this time, the output state is A = 0, B = 0.
5 and C = 1 (N = 3). According to the concept of the average error diffusion method, only the state B appears as the output state.

Assuming that the output of A is made, the probability that the state of C appears conversely increases thereafter. Thus, the processing is performed so that the output is close to 0.5 on average. On the other hand, in the method of giving priority to the appearance probability as in the halftoning A, there is no such dependency.

For example, if the appearance probability is A = 25%, B = 5
It is assumed that 0% and C = 25% are set. Assuming that A is output, the probability that A is not output thereafter is high, but the probability that C appears is not high. Still, the appearance probability of B is 50%.
That is, the output does not depend on the output of A.

This difference does not substantially occur when there are two output states, that is, when binarization is performed. The difference occurs when there are three or more output states. In the case of two types of output states, if one arrangement is determined, the other arrangement is automatically determined. Therefore, different ideas actually give the same result.

When there are three or more output states, different results are obtained, but the property of reproducing the input on average is the same for both. The difference is the arrangement of dots in a micro output state.

That is, the fine structure of the output state of an adjacent pixel is different from that of a pixel in a certain output state. In the halftoning A, there is no strong correlation between the output states of the adjacent pixels, that is, the correlation is extremely small in the arrangement of the output states. Then, each output state is uniformly dispersed in the two-dimensional pixel array structure, and arrangement can be performed so as not to give a visually unnatural feeling.

[Second Example of Half Toning A] FIG. 11
FIG. 3 is a block diagram illustrating a configuration of an image processing apparatus that executes a second example of halftoning A. In the image processing apparatus shown in FIG. 6, a feedback value is calculated by determining whether or not to invert a parameter (correction threshold value) and the sign thereof based on the determined output state. The parameter was to be modified based on the feedback value. However, as shown in FIG. 11, the appearance probability may be corrected based on the calculated feedback value.

That is, in FIG. 11, the initial appearance probabilities d1 to dN are obtained by a data table or the like based on the input signal.
Is calculated by the subtractor 213 based on the feedback value to obtain corrected appearance probabilities d′ 1 to d′ N,
The selector 203 selects a state in which d'i-ti is maximized.

Even if such a configuration is adopted, the same effect as in the first example of the halftoning A can be obtained.

[Third Example of Half Toning A] FIG.
FIG. 9 is a block diagram illustrating a configuration of an image processing apparatus that executes a third example of halftoning A. In this example, a feedback value is calculated by determining whether or not to reverse the appearance probability and its sign based on the determined output state, and the appearance probability is corrected based on the calculated feedback value.

That is, the calculation unit 307 sets d′ i as a feedback value corresponding to the selected appearance state.
Is adopted, and -d'i is adopted as the other feedback value.

By performing such processing,
The same effect as in the first example can be obtained.

[Fourth Example of Half Toning A] FIG.
FIG. 9 is a block diagram illustrating a configuration of an image processing apparatus that executes a fourth example of halftoning A.

In this example, the feedback value is calculated by determining whether or not to reverse the appearance probability and the sign thereof based on the determined output state, and the parameters are corrected based on the calculated feedback value. Is what you do.

The same effect as in the first example can be obtained by configuring the apparatus in this manner.

[Sample of Output by Half Toning A] FIG. 14 is a diagram showing a sample of output in half toning A. Input image to be processed is 10
It is assumed that the image has a uniform density of 0 × 100 pixels. This was converted into image data having five states with the number of output states N = 5. In addition, the settings were made so that the appearance probabilities of the five output states were equal.

FIGS. 14 (D) to 14 (H) show states 1 to 4, respectively.
FIG. 5 is a diagram in which the pixels for which 5 is output are indicated by black dots. (A) shows the pixels to which the statuses 1 to 4 are output in black.

(B) shows the pixels to which the output states 1 to 3 are output in black. (C) shows the pixels to which the output states 1 and 2 are output in black.

As shown in the figure, it can be seen that the distribution of pixels in which each output state appears is uniform. Further, there is no gap between a certain output state and a specific output state in an adjacent pixel.

That is, in this output sample,
In an area having the same input state (an area having the same gradation), an image is output in which three or more output states are mixed and the output states are evenly dispersed.

An example in which a natural image composed of 512 × 512 pixels is processed and half-toning A image processing is performed will be described below. Assuming that the number of output states N = 5,
The appearance probabilities of the respective output states for the input signals 0 to 1 are set so as to be a combination satisfying the following expression (1).

Input value = 0 × d1 + 0.25 × d2 + 0.5 × d3 + 0.75 × d4 + 1 × d 5 (d1 + d2 + d3 + d4 + d5 = 1) (1) It is always necessary that the sum of the appearance probabilities of each output state becomes 1. These conditions can be freely determined, but other conditions can be arbitrarily determined.

FIGS. 15 to 22 are diagrams in which pixels appearing in each state are indicated by black dots. 15 to 19 show output states 1 to 5, respectively.

FIG. 20 shows a combination of output states 1 and 2, FIG. 21 shows a combination of output states 1 to 3, and FIG. 22 shows a combination of output states 1 to 4.

Although not shown in the sample, if all the output states 1 to 5 are combined, all the pixels become black. 15 to 22, it can be seen that the distribution of the appearance pixels in each output state is uniform, and that the appearance probability effectively contributes to the reproduction of the natural image.

[Example of Half Toning B] Next, a specific example of half toning B will be described. FIG. 23 is a diagram for explaining processing of converting an N-value image output by halftoning A into a quaternary image in halftoning B.

As shown in the figure, 4-1 = 3 threshold values (T1, T2, T3) are set, and processing using these threshold values is performed to obtain an N-valued input image (0 From 255
Is determined in four areas, and output as four-level output values.

The processing in the halftoning B must be performed on each output device side. However, since the determination processing using a fixed threshold value is sufficient, the processing load can be reduced. Also, the processing speed is fast. The processing of the halftoning A is relatively heavy, but need only be performed once on the host side. Also, by using halftoning A, the capacity of the image to be transmitted can be significantly reduced.

FIG. 24 is a diagram for explaining more detailed processing of the halftoning B. First, step S4
In step 01, a threshold level is set in advance in accordance with the gradation characteristics of the output device. In step S403,
The set threshold level is stored.

If an N-level halftone image has been input via the network, in step S405 each threshold value is compared with the pixel value of one pixel to determine an output level. Processing is performed. As a result, an m-level halftone image B can be obtained.

As shown in the lower right part of FIG. 24, for example, when m = 4, three of t1, t2, and t3
It is good to set three thresholds.

As described above, in the present embodiment, a two-stage halftoning process is performed in which halftoning A is performed before image data transmission and halftoning B is performed after image data transmission is completed. FIG.
In the prior art shown in FIG. 0, two-stage halftoning is performed. However, halftoning in the prior art is not a process that can mix a large number of output states like halftoning A. However, there is a problem that the image quality is degraded when two-stage halftoning is performed. The reason will be described below.

In the prior art, in order to represent a certain input gray level x as m levels, the nearest gray level x is expressed by 2
Only two levels (L _k , L _{k + 1} ) were used.

That is, referring to FIG. 25, in the prior art, when there is a uniform pattern of input x, only two levels (L _k , L _{k + 1} ) sandwiching x are included in the pattern. Used as output state. That is, the value weighted and averaged by the ratio in which these two levels exist is the input value x
The processing was performed to express.

However, in the present embodiment, for a uniform pattern such as an input x by halftoning A, two levels (L
_k , L _{k + 1} ), by specifying the appearance ratio, other levels (for example, L _k−1 , L _{k + 2} ) can be freely mixed, and the input x is expressed on average. I'm going to.

More specifically, description will be made with reference to the histograms shown in FIGS. 27 and 28. In the case where a uniform pattern of the input x is input, in the related art, FIG.
As shown in FIG. 7, the input value is reproduced by pixels of two levels sandwiching x.

However, as shown in FIG. 28, in the present invention, the input value is reproduced using pixels other than the two-level pixels sandwiching the input x.

As described above, in the halftoning A, the input value x can be expressed by mixing a large number of levels. Therefore, even if the gradation is reduced by using a simple threshold processing (half-toning B) later, the image quality can be kept good.

In the case of the prior art, although it is said that multi-value conversion is performed by half-toning, when considering only the original input value x, the image is expressed only in two levels. It is impossible to convert the number of gradation levels to the above. Therefore, in the prior art, once the gradation reduction process is performed, the gradation level cannot be converted again unless the image is restored to a state close to the next continuous gradation.

There is another important condition for outputting a good image. That is, if the input value x is expressed by mixing a large number of output levels, it is not merely necessary to reproduce the input value x on average, but the pixels of each output level are evenly arranged as an image. It must be.

In FIG. 28, only a histogram is shown. However, if pixels having these output levels are arranged unevenly on an image, the unevenness will visually appear as noise. Therefore, although there are pixels of various types of output levels, they need to be mixed and arranged so as to be evenly dispersed with each other. Half Toning A
Thus, it is possible to output an image satisfying such a condition.

It is to be noted that the present invention is not limited to the above example of the half-toning A, and any half-toning capable of producing an image satisfying the above-mentioned conditions can be adopted as the half-toning A.

[Sample of Image Output] Next, the process of converting the distribution image A into the output image B according to the present embodiment will be described with reference to specific examples of images. Delivery image A
Is a grayscale image of uniform density of 28 × 25 pixels. FIG. 29 shows an example in which the number of output states is set to N = 12, halftoning A is performed, and the image is converted into an N-level image.

In FIG. 29, there is a possibility that 12 types of gradation levels may be mixed, but the appearance probability of a distant level (a level close to black or white) is small with respect to a level almost close to the median value. Therefore, within the range of 28 × 25 pixels, 1
Only less than two levels of pixels actually appear.

The images obtained by subjecting the image shown in FIG. 29 to the processing corresponding to the halftoning B are shown in FIGS.
3 is shown. The number of output levels of each image is 6 in FIG.
There are four levels in FIG. 31, three levels in FIG. 32, and two levels in FIG.

As shown in FIGS. 30 to 33, it can be seen that output images of various levels smaller than 12 can be easily obtained from one 12-valued image (FIG. 29).

More specifically, an example in which image processing is performed on a 256-gradation distribution image will be described below.

FIG. 34 is a diagram showing an image after halftoning A and halftoning B have been performed on the distribution image as shown in the present embodiment. In this example, the conversion of 256 → 12 levels is performed by halftoning A, and the conversion of 12 →
Conversion to two levels has been performed.

As shown in this image, in the present embodiment, it is possible to provide an image with high reproduction accuracy of the original image despite performing half-toning twice.

FIG. 35 is a diagram showing an image after halftoning the original image of 256 gradations to 12 gradations using the error diffusion method and further performing halftoning of binarization. is there. As shown in this figure, if two-stage halftoning is performed by the conventional technique, it becomes difficult to reproduce the gradation of the original image.

FIG. 36 is a diagram showing a state where an original image of 256 gradations has been binarized only by half-toning B. As described above, if only the halftoning B is performed, it is difficult to reproduce the gradation of the original image. However, in this embodiment, first, the output states of the pixels of each level are diffused by the halftoning A. By binarizing, it is possible to increase the accuracy of tone reproduction of the original image.

[Effects of Embodiment] As shown in the present embodiment, after the number of gradations is once reduced by halftoning A, the image is again output to the output device by simple processing (halftoning B). There are the following advantages by matching the number of levels.

First, since the number of gradations can be reduced in advance, the capacity of an image can be reduced, and the load of handling such as image transmission can be reduced. Next, the number of output levels can be arbitrarily converted according to the output device. In addition, since the conversion process simply needs to perform threshold processing, no load is imposed on the device.

Therefore, as described in the embodiment, the efficiency of the system is improved when distributing images through a network, or when displaying or printing a specific image on a number of different output devices. This has advantageous effects in terms of processing speed and resources.

The embodiments disclosed this time are to be considered in all respects as illustrative and not restrictive. The scope of the present invention is defined by the terms of the claims, rather than the description above, and is intended to include any modifications within the scope and meaning equivalent to the terms of the claims.

[Brief description of the drawings]

FIG. 1 is a flowchart illustrating processing of a halftone image forming method according to one of the embodiments of the present invention.

FIG. 2 is a diagram illustrating a specific method of using a halftone image forming method.

FIG. 3 is a diagram illustrating an operation of an image processing apparatus that executes a first example of halftoning A.

FIG. 4 is a diagram showing a setting example of a distribution of appearance probabilities with respect to input values.

FIG. 5 is a diagram illustrating a setting example of a distribution of appearance probabilities with respect to input values.

FIG. 6 is a block diagram illustrating a configuration of an image processing apparatus that executes a first example of halftoning A.

FIG. 7 is a more detailed block diagram of FIG. 6;

FIG. 8 is a diagram for explaining processing in a first example of halftoning A;

FIG. 9 is a view following FIG. 8;

FIG. 10 is a diagram illustrating a specific example of a process in a first example of halftoning A;

FIG. 11 is a block diagram illustrating a configuration of an image processing apparatus that executes a second example of halftoning A.

FIG. 12 is a block diagram illustrating a configuration of an image processing apparatus that executes a third example of halftoning A.

FIG. 13 is a block diagram illustrating a configuration of an image processing apparatus that executes a fourth example of halftoning A.

FIG. 14 is a diagram showing an output sample of an image.

FIG. 15 is a first diagram illustrating an output result of a natural image.

FIG. 16 is a second diagram illustrating an output result of a natural image.

FIG. 17 is a third diagram illustrating an output result of a natural image.

FIG. 18 is a fourth diagram illustrating an output result of a natural image.

FIG. 19 is a fifth diagram illustrating an output result of a natural image.

FIG. 20 is a sixth diagram illustrating an output result of a natural image.

FIG. 21 is a seventh diagram illustrating an output result of a natural image.

FIG. 22 is an eighth diagram illustrating an output result of a natural image.

FIG. 23 is a diagram for explaining a halftoning B process.

FIG. 24 is a diagram for explaining a specific processing example of halftoning B;

FIG. 25 is a diagram for explaining a problem of the conventional technique.

FIG. 26 is a view for explaining effects of the present invention.

FIG. 27 is a diagram for explaining halftoning processing according to the related art.

FIG. 28 is a diagram for describing an output result by halftoning A.

FIG. 29 is a diagram illustrating a result of processing an image by halftoning A.

30 is a diagram showing an image obtained by binarizing the image of FIG. 29 by halftoning B. FIG.

31 is a diagram showing an image obtained by quaternizing the image of FIG. 29 by half-toning B. FIG.

32 is a diagram showing an image obtained by ternarizing the image of FIG. 29 by halftoning B. FIG.

FIG. 33 is a diagram showing an image obtained by binarizing the image of FIG. 29 by half-toning B.

FIG. 34 is a diagram illustrating an image output by the halftoning image forming method according to the present embodiment.

FIG. 35 is a diagram showing an output result of an image according to the related art.

FIG. 36 is a diagram illustrating a state where a multi-valued image is simply binarized.

FIG. 37 is a diagram illustrating a method of distributing an image via a network.

FIG. 38 is a diagram illustrating a first transmission example of an image.

FIG. 39 is a diagram illustrating a second example of image transmission.

FIG. 40 is a diagram illustrating a third transmission example of an image.

FIG. 41 is a diagram illustrating a fourth transmission example of an image.

[Description of Signs] 151 appearance probability determination unit, 153 selection unit, 155 output unit, 157 calculation unit, 159 β multiplication unit, 161 initial threshold value generation unit, 163 subtraction unit, 171 input unit, 1
73 appearance probability determination unit, 175a to 175c subtractor,
177 selection section, 179 output section, 181 inversion section, 1
83a-183c Subtractor, 185β Multiplying Unit, 187
Initial threshold generator, 189a to 189c Subtractor.

Continued on the front page F term (reference) 2C262 AA24 AA30 CA08 EA12 FA20 GA57 GA59 5B057 CA08 CA12 CA16 CB08 CB12 CB16 CE11 5C077 MP01 NN02 PP42 PQ12 PQ23 RR06 RR14 RR21

Claims (3)

[Claims] 1. A first step of creating a first halftone image in which pixels of each level of N value are mixed from a multi-tone image, and a threshold value of the first halftone image created in the first step Forming a second halftone image having an m value by processing. 2. A first step of creating a first halftone image in which pixels of respective levels of N values are mixed from a multi-tone image, and outputting the first halftone image created in the first step as an output tone Receiving at a plurality of terminals having different numbers. 3. A first step of creating a first halftone image in which pixels of each level of N values are mixed from a multi-tone image, and outputting the first halftone image created in the first step as an output tone A second step of distributing on a network on the assumption that the information is received by a plurality of terminals having different numbers.