JPH11164143A - Binary multi-valued image conversion method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To prevent periodic moires from appearing in a multi-valued image by temporarily generating a dummy binary image, having black pixels dispersed at random from a binary image and converting it into a multi-valued image by using a scan matrix which has a different size from or the same size with a threshold matrix. SOLUTION: A dummy binary image is generated by having coloring materials (or white pixel) of an inputted binary image moved and dispersed, and a multi-valued image which has a different scan matrix size from or the same size with a threshold matrix is obtained from the binary image. This converting method determines a colored image of interest in the binary image in a first step and determines the movement position of the colored image of interest in the binary image in a second step. In a third step, the colored image is dispersed to the above determined movement position and in a next step, the multi-valued image is obtained according to the binary image obtained from the result of the 3rd step.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

[0001] 1. Field of the Invention [0002] The present invention relates to a binary / multi-value conversion method for converting a binary color image into a multi-valued color image.

[0002]

2. Description of the Related Art In order to perform offset printing of a multi-valued color image, a binary image which is halftone-dotted is used. In order to correct the color tone from the binary image after printing, it is necessary to return to the multi-valued image. Further, in recent years, there has been a demand for output to an inexpensive multi-value color proofer at a distributed factory such as a local newspaper factory to which only a binary image is sent from the head office.

Conventionally, when converting a binary image to a multi-valued image, the image is divided into scan matrices (V × W = m × n units) having the same size as the threshold matrix (m × n units), and There is known a method of obtaining a multi-valued image corresponding to black and white. In this method, by setting the threshold matrix and the scan matrix to have the same size, it is possible to reduce the moire of the periodicity that was not present in the original image before binarization (Japanese Patent Laid-Open No. 60-140349).
JP-A-6-70138).

However, in this method, since one pixel of a multi-valued image is formed from a scan matrix having the same size as the threshold value matrix, the resolution (image quality) and the maximum number of gradations of the multi-valued image are determined by the threshold value matrix. In the case of a binary image in which the size of the threshold matrix is large, the maximum number of tones of the obtained multivalued image is large, but there is a problem that the resolution is reduced, and the size of the threshold matrix is small. In the case of a binary image, the resolution of the obtained multivalued image is high, but there is a problem that the maximum number of gradations is reduced.

Accordingly, it is possible to reduce the maximum number of gradations by using a scan matrix having a size smaller than the threshold matrix, so that the resolution should be increased. Further, the resolution should be reduced by using a scan matrix having a size larger than the threshold matrix. Therefore, there is a demand for a method of obtaining a multi-valued image suitable for use such as increasing the maximum number of gradations.

However, when obtaining a multi-valued image with a scan matrix size different from the size of the conventional threshold matrix, since the binary image is processed without any modification, the periodicity which has not been provided in the original image is obtained. There is a drawback that moire occurs.

[0007]

SUMMARY OF THE INVENTION An object of the present invention is to solve such a problem of the prior art, and is the same even if the size of the threshold matrix and the size of the scan matrix are different. Also, the periodic matrix moire does not occur in the multi-valued image, and the difference between the size of the threshold matrix and the size of the scan matrix makes it possible to increase the resolution while maintaining the maximum number of gradations or to maintain the resolution. An object of the present invention is to increase the maximum number of gradations and to provide a binary / multi-value conversion method suitable for the application.

[0008]

Means for Solving the Problems As a result of repeated studies to solve the above problems, the present inventor has moved and dispersed colored pixels (or white pixels) of an input binary image in a desired manner. In this way, a pseudo-binary image is created, and a multi-valued image is obtained from this binary image with a size different from the threshold matrix or with the same scan matrix size.

Therefore, no periodic moire occurs in the multi-valued image, and the size of the threshold matrix and the size of the scan matrix are different.
By enlarging or reducing the magnification twice, it is possible to increase the resolution while maintaining the maximum number of gradations, increase the maximum number of gradations while maintaining the resolution, and obtain a binary image suitable for the application.

That is, the present invention provides a method for converting a binary image obtained by using a threshold matrix into a multivalued image using a scan matrix having a different size or the same size as the threshold matrix. A first step of determining a colored pixel of interest in the binary image, a second step of determining a moving position of the colored pixel determined in the first step in the binary image, Second
A third step of dispersing the colored pixels at the movement position determined in the step, and a step of obtaining a multi-valued image based on a binary image obtained from the result of the third step. The present invention relates to a binary / multi-valued image conversion method which is a feature.

Further, the present invention provides (1) M × N (dots) formed by binarized halftone dots obtained using a threshold matrix of m × n units (where m and n are integers). However, in a binary image composed of (M and N are integers) pixels, the binary image is enlarged, reduced, or magnified by k times to obtain M ′
× N ′ (where M ′ = k × M, N ′ = k × N, M,
A means for obtaining a binary image consisting of:
Means for randomly distributing the colored pixel portions of the binary image composed of two pieces, and creating a pseudo-binary image from the result. For this pseudo-binary image, a V × W unit (where V and W are integers) ) Is divided for each scan matrix, and (M ′ / V) × corresponding to the ratio of the number of white and colored portions in the scan matrix
(N '/ W) pieces (however, M' / V and N '/ W are integers)
And a means for obtaining a multi-valued image composed of pixels.

Further, according to the present invention, (2) means for generating a pseudo binary image by randomly dispersing the colored pixel portions of the M ′ × N ′ binary image described in (1) above, '
Means for assigning a uniform number P (1, 2, 3,..., T) while counting the colored pixel portions in a binary image area composed of × N ′ pieces; .., T) means for randomly selecting the colored pixel portion, means for randomly selecting the moving direction D of the selected colored pixel portion (up, down, left and right directions of the colored pixel portion), and direction D for moving the colored pixel portion. And a means for moving only when the pixel is white, and repeating this means about [[{(m + n) / 2} / 2] × k] ² × T times. About.

Further, according to the present invention, (3) means for generating a pseudo binary image by randomly dispersing the colored pixel portions of the M ′ × N ′ binary image described in (1) above, '
Means for assigning a uniform number P (1, 2, 3,..., T) while counting the colored pixel portions in the binary image area consisting of × N ′ pieces, the uniform number P (1, 2, 3,. ... Means for selecting the colored pixel portions of T) at random or in the order of the number P

[Outside 1] Means for determining the destination angle θ (0 ≦ θ <2π) of the selected colored pixel portion, Δx
= Int [rcosθ], Δy = int [rsinθ]
(However, int [] is defined as taking the integer part of [].) A method for converting a binary / multi-valued image is characterized in that this means is repeated about T times. .

Further, according to the present invention, there is provided (4) means for generating a pseudo binary image by randomly dispersing the colored pixel portions of the M ′ × N ′ binary image described in the above (1). '
Means for assigning a uniform number P (1, 2, 3,..., T) while counting the colored pixel portions in a binary image area composed of × N ′ pieces; .., T) means for selecting the colored pixel portions randomly or in the order of the number P, and the moving distance r of the selected colored pixel portion is given by r = [k (m + n) / 20] · (G1-2 × | G2 | +
10) Here, G1 and G2 are normal random numbers with mean 0 and variance 1 obtained by the Box-Muller method, and G1 = (− 2 × log (R1)) ^1/2 × cos (2π ×
R2) G2 = (− 2 × log (R1)) ^1/2 × sin (2π ×
R2) Here, R1 and R2 are means for determining from [0,1] uniform random numbers, the destination angle θ of the selected colored pixel portion.
Means for determining (0 ≦ θ <2π), Δx = int [rcosθ], Δy = in at the original position of the selected colored pixel portion
The present invention relates to a binary / multi-valued image conversion method characterized by comprising means for adding and moving t [rsinθ] and repeating this means about T times (total number of colored pixels).

The pixel of interest in the present invention is not limited to a colored pixel, but may be a white pixel.

[0016]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS An embodiment of the present invention will be described below with reference to the drawings. <First Embodiment> FIG. 1 shows a flowchart of the first embodiment of the present invention.

FIG. 2 shows a binary image used in the present embodiment.

In the pixel interpolation step 1 in FIG. 1, the binary image 101 shown in FIG. 2 is input.

The binary pixel 101 has a size of 10 × 10 units (mx
area ratio 5 obtained using a threshold matrix of (n units)
An image formed by binarized halftone dots representing 0%, with 80 pixels vertically (M), 80 pixels horizontally (N), and a total number of pixels of 6
It consists of 400 pieces.

The pixel interpolation step 1 multiplies the binary image 101 by k = 2. The binary image 102 pixel-interpolated using the nearest neighbor method has 160 vertical pixels (M ′), 160 horizontal pixels (N ′), and a total of 25600 pixels. FIG. 3 shows the interpolated binary image.

In the interpolated binary image 102, black pixels in the binary image area are counted in a black pixel counting step 2. The black pixel counting step 2 sequentially assigns a uniform number, which is a number for identifying a black pixel, to each black pixel.

This counting process will be described with reference to a binary image 103 of one unit of the matrix shown in FIG. In the black pixel counting step 2, the uniform number P = 1 is assigned to the black pixels of X = 9 and Y = 1, and the unique number P = 2 is assigned to the black pixels of X = 10 and Y = 1.
Is assigned a uniform number to each black pixel. When all the black pixels in the binary image area have been assigned a uniform number, the total number of black pixels T = 12800 is obtained.

In the pixel of interest determination step 3, black pixels in the binary image area are randomly noted. In the embodiment of the present invention, "random" means selecting a black pixel with a probability of 1/1800 (1 / T).

Specifically, P is an integer part obtained by adding 1 to a value obtained by multiplying 12800 times by using a [0, 1] uniform random number R. However, when the uniform random number R = 1, P = 12800.

P = int [T × R + 1] (1) (However, 0 ≦ R <1, int [] is defined to take an integer part of [].) P = T = 12800 where R = 1) (2) From the T (12800 pixels), one black pixel (pattern number P) can be randomly noted from the above equation.

Next, in a reference pixel determination step 5, pixels around the target pixel are randomly referred to. That is,
In the target pixel determination step 3, one of the pixels adjacent to the randomly selected target pixel is randomly selected,
The selected pixel is set as a reference pixel. If the reference pixel is a white pixel, the pixel of interest is moved to the position of the reference pixel in the next step. If the reference pixel is outside the black pixel or the binary image area, the target pixel is not moved to the reference pixel position. This is to prevent the number of black pixels from decreasing when moved.

This reference processing will be described in detail with reference to FIG.

In FIG. 5, F is a target pixel randomly selected in the target pixel determination step 3, and the target pixel F
One of the four pixels W ₁ , W ₂ , W ₃ , W ₄ adjacent to is randomly selected by the reference pixel determination step 5. In FIG. 5, adjacent pixels W ₁ , W ₂ , W ₃ and W ₄ are all
The case in the binary image area is shown.

Specifically, the direction D is an integer part obtained by adding 1 to a value obtained by multiplying the quadruple number by using the [0, 1] uniform random number R.
However, when the uniform random number R = 1, D = 4.

D = int (4 × R + 1) (where 0 ≦ R <1) (3) D = 4 (where R = 1) (4) In FIG. = 1, 2, 3, and 4 correspond to the right, left, upper, and lower, respectively, the position (x ₁ , y ₁ ) of the reference pixel adjacent to the target pixel is the target pixel position F (x ₀ , y ₀ ) Using,
The following equation (5): W ₁ (x ₁ , y ₁ ) = (x ₀ +1, y ₀ ) → D = 1 W ₂ (x ₁ , y ₁ ) = (x ₀ −1, y ₀ ) → D = 2 W ₃ (x ₁ , y ₁ ) = (x ₀ , y ₀ +1) → D = 3 W ₄ (x ₁ , y ₁ ) = (x ₀ , y ₀ −1) → D = 4 .

One of the adjacent pixels is randomly referred to as a reference pixel by the above equation (3) or (4), and the position of one selected reference pixel is determined by equation (5).

Next, in the white pixel determination step 6, it is determined whether the reference pixel at the reference pixel position is a white pixel or a black pixel. If the reference pixel is a white pixel, Yes is determined, and if the reference pixel is a black pixel, No is determined. .

Further, as shown in FIG. 6, when the reference pixel is outside the binary image area, the white pixel determination step 6
Is determined. If the determination is No, the process returns to the pixel-of-interest determination step 3 and is repeated until the determination becomes Yes.

[0034] In this example, the pixel W ₁ is selected as the reference pixel.

The position moving step 7, the reference pixels W ₁ it is determined "Yes" in the white pixel determination step 6, the target pixel position a pixel of interest F (uniform number P) F (x, y) =
From (x ₀ , y ₀ ), the reference pixel position B ₁ (x, y) = (x ₁ ,
y ₁ ). Thus, the position of the target pixel P is dispersed as (x ₀ , y ₀ ) → (x ₁ , y ₁ ).

As described above, in the present invention, the target pixel P is randomly selected, and further, the reference pixel is randomly selected from the adjacent pixels with reference to the randomly selected target pixel P. If the reference pixel is a white pixel in the binary image area, the pixel of interest P is moved to the position of the reference pixel, the binary image is randomly (randomly) dispersed, and the distribution of the black pixels is repeated. This is one of the features of the present invention.

Next, in the distribution number determination step 8,
The number of repetitions of the distributed processing of “determining the target pixel → determining the reference pixel → moving the black of the target pixel only when the reference pixel is white” is determined.

In the distribution number determination step 8, the number of repetitions is [[{(m + n) / 2} / 2] × k] ² × T (6) (where m and n are If the size of the threshold matrix, k is the enlargement coefficient, and T is the total number of pixels, it is determined as Yes, and if less than No, it is determined as No. More specifically, in the present embodiment, the number-of-times-of-dispersion determination step 8 is performed when the number of times of dispersion is [[{(m + n) / 2} / 2] × k] ² × T = [[{(10 + 10) / 2} / 2. ] × 2] ² × 12800 = 12800000 If it is equal to or more than Yes, it is determined to be No, and if less than No, it is determined to be No.

If the determination is No, the process returns to the target pixel determination step 3, and the distribution of black pixels is repeated until the determination is Yes. Pseudo binary image 104 as shown in FIG. 7 in which black pixels are randomly distributed by repeating the distribution.
Get.

Here, the number of times of dispersion is an average number of times of movement of a pixel of interest in order to obtain a desired multi-valued image in which data obtained in advance by trial is included in an allowable range.
It is preset in a range from 100 times to at most 100 · T times.

The above steps up to “creating a pseudo binary image” are performed in order to obtain a multi-valued image having a desired high image quality.
It is repeated a preset number of times. In one embodiment of the present invention, the number of repetitions A is set to three times.

That is, by the averaging step 10, the pseudo 2
The number of repetitions up to the creation of the value image is determined. In the averaging step 10, if the number of repetitions is A = 3 or more, Y
If it is less than es, it is determined as No.

If the determination is No, the number of black pixels B _A (x, y) at each position of the pseudo binary image is stored in the storage step 12, and the pseudo binary image is initialized to the binary image 102. Then, the process returns to the target pixel determination step 3. Until the number of repetitions is determined to be Yes in the averaging step 10, the above steps of randomly distributing the pixel of interest and creating a pseudo binary image are repeated as described above.

The averaging step 10 calculates the number of repetitions A =
When 3 is determined, the average pseudo binary image step 13 obtains an average pseudo binary image based on the number of black pixels at each position and the number of repetitions.

The average pseudo binary image step 13 is a pseudo binary
The number of black pixels at each position of the value image is B _A , pseudo 2
Assuming that the number of creations of the value image is A,

(Equation 1) To obtain the average number of blacks in one pixel, and obtain an average pseudo binary image indicating the average number of blacks at each position.

Next, the multi-value conversion step 15 divides the average pseudo-binary image into 8 × 8 (V × W) scan matrices according to the ratio of black and white in each scan matrix. A multi-valued image 105 shown in FIG. 8 including (M ′ / V) × (N ′ / W) pixels is obtained. However, if the sum of black in the scan matrix exceeds V × W = 8 × 8 = 64, the value is 64. The multi-valued image 105 is composed of 20 pixels vertically, 20 pixels horizontally and 400 pixels in total.

Although the input binary image is enlarged (or reduced) in the above embodiment, the input binary image may be used as it is.

Also, a case has been shown in which a pseudo binary image is created a plurality of times and the pseudo binary image is averaged to obtain an average pseudo binary image. A multi-value image may be created based on the value image.

The target pixel may be determined in order. <Embodiment 2> FIG. 9 is a flowchart of a second embodiment of the present invention.

The reference pixel determining step is different from that of the first embodiment shown in FIG. Further, the second embodiment does not require a white pixel determination step. The other points are the same as those of the first embodiment, and the same reference numerals as those in FIG. 1 indicate the same parts.

In the second embodiment, as in the first embodiment, the binary image 101 shown in FIG.

The binary pixel 101 has a size of 10 × 10 units (mx
area ratio 5 obtained using a threshold matrix of (n units)
An image formed by binarized halftone dots representing 0%, with 80 pixels vertically (M), 80 pixels horizontally (N), and a total number of pixels of 6
It consists of 400 pieces.

In the pixel interpolation step 1, the binary image 101 is multiplied by k = 2. The binary image 102 pixel-interpolated using the nearest neighbor method has 160 vertical pixels (M ′), 160 horizontal pixels (N ′), and a total of 25600 pixels. FIG. 3 shows the interpolated binary image.

In the interpolated binary image 102, black pixels in the binary image area are counted in a black pixel counting step 2. The black pixel counting step 2 sequentially assigns a uniform number, which is a number for identifying a black pixel, to each black pixel.

This counting process will be described with reference to a binary image 103 of one unit of the matrix shown in FIG. In the black pixel counting step 2, the uniform number P = 1 is assigned to the black pixels of X = 9 and Y = 1, and the unique number P = 2 is assigned to the black pixels of X = 10 and Y = 1.
Is assigned a uniform number to each black pixel. When all the black pixels in the binary image area have been assigned a uniform number, the total number of black pixels T = 12800 is obtained. That is, the uniform number P of the last black pixel is P = 12800.

Next, in the target pixel determination step 3, 2
Attention is paid to the black pixels in the value image area at random. Random means selecting a black pixel with a probability of 1/1800 (1 / T).

More specifically, an integer part obtained by adding 1 to a value obtained by multiplying the value by 12800 using the [0, 1] uniform random number R is defined as P. However, when the uniform random number R = 1, P = 12800.

P = int [T × R + 1] (where 0 ≦ R <1, int [] is defined as taking an integer part of [].) P = T (where R = 1) , One black pixel out of T
Can be noticed at random.

Next, in the reference pixel determination step 21,
One pixel is randomly (randomly) selected from the pixels around the target pixel, and the selected pixel is determined as a reference pixel.

Here, as described above, in the first embodiment, one reference pixel is randomly selected from the pixels adjacent to the target pixel, and the reference pixel is a white pixel in the binary image area. The target pixel (black pixel) was moved to the reference pixel position, and this processing step was repeated a predetermined number of times to obtain an optimal multi-valued image.

However, in the second embodiment, trials for obtaining an optimal multi-valued image are repeated in advance, and in order to obtain an optimal multi-valued image, each distance r of each image of interest (each black pixel) has moved. Is statistically processed to calculate each pixel of interest (that is, each black pixel).
Is measured for each moving distance r (hereinafter, simply referred to as “frequency”), and a reference pixel is selected from pixels around the target pixel based on the frequency.

That is, the reference pixel determination step 21 is the same as that in FIG.
A reference pixel is determined by randomly referencing pixels around the pixel of interest based on the histogram 106 shown in FIG.

Here, FIG. 10 shows a case where the trial is repeated in advance to obtain the best multi-valued image, and the moving distance r at which each black pixel has moved at this time is measured in units of one pixel. r is a histogram showing the frequency of appearance of black pixels in a columnar manner, where the vertical axis indicates frequency (frequency of appearance) and the horizontal axis is 1
Indicates the size of a pixel.

The reference pixel determination step will be described with reference to the explanatory diagram of the reference pixel determination processing shown in FIG.

As shown in FIG. 11, the reference pixel determination step 21 refers to a pixel located at a position apart from the target pixel F by a moving distance r corresponding to the histogram and a random angle θ among the peripheral pixels, as shown in FIG. The pixel W is assumed.

In the reference pixel determination step 21, the total distance of black pixels is determined by the moving distance r at the frequency represented as a whole in the histogram.
Is determined so as to move the target pixel.

When the [0, 1] uniform random number R is used, the angle θ is defined as θ = 2π × R (8), and an arbitrary direction is determined with equal probability. You.

The polar coordinate system (r, θ) is changed to a rectangular coordinate system (x,
y), and considering that the pixel position is an integer, the reference pixel position W (x, y) = (x ₁ , y ₁ ) is the target pixel position (x, y) = (x ₀ , y ₀₎ ), W (x ₁ , y ₁ ) = (x ₀ + Δx, y ₀ + Δy) = (x ₀ + int [r × cos (θ)], y ₀ + int [r × sin (θ)] .

Thus, the reference pixel determination step 21
Determines a reference pixel by randomly selecting pixels around the target pixel (which are separated by a distance r and an angle θ) based on a histogram stored in advance.

Next, in the position moving step 7, the black pixel (back number P) is set to the target pixel position F (x, y) = (x ₀ ,
(y ₀ ) to the reference pixel position W (x, y) = (x ₁ , y ₁ ).

F (x ₀ , y ₀ ) → (x ₁ , y ₁ ) However, as shown in FIG. 12, when the position of the reference pixel W ₁ is out of the binary image area where the pixel interpolation is performed, the pixel of interest F Is not moved and keeps its original position.

F (x ₀ , y ₀ ) → (x ₀ , y ₀ ) In the second embodiment, even if the reference pixel is a white pixel or a black pixel, the target pixel moves. as shown in 12, (even if there is a black pixel in the reference pixel position, the pixel of interest is moved.), sometimes there are a plurality of black pixels in the same position of the reference pixel W ₂ Next, dispersion times judging step 8 By
It is determined how many times the variance of “the target pixel moves regardless of the black and white of the reference pixel and the same position is maintained only when the reference pixel is outside the pixel-interpolated binary image area” is repeated.

In the second embodiment, the target pixel is dispersed at the reference pixel position by the actually measured histogram.
The number-of-dispersion step 8 is performed when the number of dispersions is equal to the total number of target pixels T = 1.
If it is 2800 or more, it is determined as Yes, and if it is less than 2,800 is determined as No. If the determination is No, the process returns to the pixel-of-interest determination step 20, and the dispersing operation is repeated until the result of the dispersal number determination step 8 is Yes.

When the number-of-dispersion determination step 8 determines that the dispersion is 12,800 times, a pseudo binary image in which all the black pixels in the pixel-interpolated binary image area are randomly dispersed by one movement is converted into a pseudo binary image. It is created in the value image step 9.

The steps up to “creating a pseudo-binary image” are performed in order to obtain a multi-valued image having a desired high image quality.
It is repeated a preset number of times. In the second embodiment, the number of repetitions A is set to three times.

That is, by the averaging step 10, the pseudo 2
The number of repetitions up to the creation of the value image is determined. In the averaging step 10, if the number of repetitions is A = 3 or more, Y
If it is less than es, it is determined as No.

If No, the number of black pixels B _A (x, y) at each position of the pseudo binary image is stored in the storage step 12, and the pseudo binary image is initialized to the binary image 102. Then, the process returns to the target pixel determination step 3. Until the number of repetitions is determined to be Yes in the averaging step 10, the above steps of randomly distributing the pixel of interest and creating a pseudo binary image are repeated as described above.

The averaging step 10 calculates the number of repetitions A =
When 3 is determined, the average pseudo binary image step 13 obtains an average pseudo binary image based on the number of black pixels at each position and the number of repetitions.

The average pseudo binary image step 13 is a pseudo binary
The number of black pixels at each position of the value image is B _A , pseudo 2
Assuming that the number of creations of the value image is A,

(Equation 2) Then, the average number of blacks within one pixel is obtained, and an average pseudo binary image indicating the average number of blacks at each position is obtained.

Next, in the multi-value conversion step 15, the average pseudo binary image is divided into scan matrices of 8 × 8 units (V × W units), and according to the ratio of black and white in each scan matrix. A multi-valued image 105 shown in FIG. 8 including (M ′ / V) × (N ′ / W) pixels is obtained. However, if the sum of black in the scan matrix exceeds V × W = 8 × 8 = 64, the value is 64. The multi-valued image 105 is composed of 20 pixels vertically, 20 pixels horizontally and 400 pixels in total.

In the above embodiment, the case where the input binary image is enlarged (or reduced) is shown, but the input binary image may be used as it is.

Also, a case has been shown in which a pseudo binary image is created a plurality of times, and the pseudo binary image is averaged to obtain an average pseudo binary image. A multi-value image may be created based on the value image.

Although the number of times of dispersion of black pixels is 12600 times, it is not necessarily 12600 times and can be increased or decreased as necessary.

In this embodiment, the target pixel may be determined in order. <Third Embodiment> FIG. 13 is a flowchart of a third embodiment of the present invention.

As compared with the first embodiment of the present invention shown in FIG. 1, the target pixel determining step and the reference pixel determining step are different. Further, the third embodiment does not require a white pixel determination step. The other points are the same as those of the first embodiment, and the same reference numerals as those in FIG. 1 indicate the same parts.

In the third embodiment, the binary image 101 shown in FIG. 2 is input to the pixel interpolation step 1 as in the first embodiment.

The binary pixel 101 has a size of 10 × 10 units (mx
area ratio 5 obtained using a threshold matrix of (n units)
An image formed by binarized halftone dots representing 0%, with 80 pixels vertically (M), 80 pixels horizontally (N), and a total number of pixels of 6
It consists of 400 pieces.

In the pixel interpolation step 1, k = 2 times the binary image 101. The binary image 102 pixel-interpolated using the nearest neighbor method has 160 vertical pixels (M ′), 160 horizontal pixels (N ′), and a total of 25600 pixels. FIG. 3 shows the interpolated binary image.

In the interpolated binary image 102, black pixels in the binary image area are counted in a black pixel counting step 2. The black pixel counting step 2 sequentially assigns a uniform number, which is a number for identifying a black pixel, to each black pixel.

The counting process will be described with reference to the binary image 103 of one unit of the matrix shown in FIG. In the black pixel counting step 2, the uniform number P = 1 is assigned to the black pixels of X = 9 and Y = 1, and the unique number P = 2 is assigned to the black pixels of X = 10 and Y = 1.
Is assigned a uniform number to each black pixel. When all the black pixels in the binary image area have been assigned a uniform number, the total number of black pixels T = 12800 is obtained. That is, the uniform number P of the last black pixel is P = 12800.

Next, in the pixel of interest determination step 23,
Attention is paid to the black pixels in the binary image area in which the pixels have been interpolated in the order of the numbers. In the order of the numbers, the next to 1 is 2, the next to 2 is 3, etc.
.. and black pixels are selected in the order of their numbers.

As described above, the pixel-of-interest determination step 23 is as follows.
One black pixel among the T black pixels is assigned a uniform number 1, 2, 3,
... and the numbers are selected in the order and attention is paid. This is one of the features of the third embodiment.

Next, in the reference pixel determination step 24,
One pixel is randomly (randomly) selected from the pixels around the target pixel, and the selected pixel is determined as a reference pixel.

The third embodiment is characterized in that the reference pixel determination step 24 distributes all black pixels at a desired distribution frequency using random numbers.

The reference pixel determination step will be described with reference to the explanatory diagram of the reference pixel determination processing shown in FIG.

As shown in FIG. 11, the reference pixel determination step 21 is performed at a moving distance r from the target pixel F in the peripheral pixels.
A pixel located at a position separated by a random angle θ is defined as a reference pixel W.

For example, the moving distance r is first determined by
By the Mueller method, two normal random numbers G1 and G2 (mean 0, variance 1) are generated from the two [0,1] uniform random numbers R1 and R2, and three terms of G1, | G2 | From Equation (9) below.

R = [k (m + n) / 20] · (G1-2 × | G2 | +10) = [2 (10 + 10) / 20] · (G1-2 × | G2 | +10) = 2 (G1-2) × | G2 | +10) (9) where G1 = (− 2 × log (R1)) ^1/2 × cos (2π ×
R2) G2 = (− 2 × log (R1)) ^1/2 × sin (2π ×
R2). .

That is, in the reference pixel determination step 24, the moving distance r is determined by the above equation (9) based on the frequency of the histogram 107 shown in FIG.

Here, in FIG. 14, the vertical axis indicates the frequency of occurrence of each black pixel at each moving distance, and the horizontal axis indicates the moving distance of each black pixel expressed in units of one pixel.

The angle θ is defined as θ = 2π × R (10) using the [0, 1] uniform random number R, and an arbitrary direction is determined with equal probability. You.

The polar coordinate system (r, θ) is changed to a rectangular coordinate system (x,
y), and considering that the pixel position is an integer, the reference pixel position W (x, y) = (x ₁ , y ₁ ) is the target pixel position (x, y) = (x ₀ , y ₀₎ ), W (x ₁ , y ₁ ) = (x ₀ + Δx, y ₀ + Δy) (x ₀ + i
nt [r × cos (θ)], y ₀ + int [r × sin
(Θ)].

As described above, pixels around the target pixel (a distance r and an angle θ apart) can be randomly referred to based on the preset histogram.

Next, in the position moving step 7, the black pixel (back number P) is set to the target pixel position F (x, y) = (x ₀ ,
(y ₀ ) to the reference pixel position W (x, y) = (x ₁ , y ₁ ).

F (x ₀ , y ₀ ) → (x ₁ , y ₁ ) However, if the position of the reference pixel W ₁ is outside the pixel-interpolated binary image area as shown in FIG. Is not moved and keeps its original position.

F (x ₀ , y ₀ ) → (x ₀ , y ₀ ) In the third embodiment, even if the reference pixel is a white pixel or a black pixel, the target pixel moves. as shown in 12, (even if there is a black pixel in the reference pixel position, the pixel of interest is moved.), sometimes there are a plurality of black pixels in the same position of the reference pixel W ₂ Next, dispersion times judging step 8 By
It is determined how many times the variance of “the target pixel moves regardless of the black and white of the reference pixel and the same position is maintained only when the reference pixel is outside the pixel-interpolated binary image area” is repeated.

In the third embodiment, the target pixel is dispersed at the reference pixel position in the order of the spine number.
Is determined as Yes if the number of times of distribution is equal to or greater than the total number of pixels of interest T = 12800, and is determined as No if less than T800. If the determination is No, the process returns to the pixel-of-interest determination step 20, and the dispersing operation is repeated until the result of the dispersal number determination step 8 is Yes.

When the number-of-dispersion determination step 8 determines that the dispersion is 12,800 times, a pseudo binary image in which all the black pixels in the pixel-interpolated binary image area are randomly dispersed by one movement is converted into a pseudo binary image. It is created in the value image step 9.

The steps up to “creating a pseudo-binary image” are as follows in order to obtain a multi-valued image having a desired high image quality.
It is repeated a preset number of times. In the third embodiment, the number of repetitions A is set to three times.

That is, by the averaging step 10, the pseudo 2
The number of repetitions up to creation of the value image is determined.

In the averaging step 10, when the number of repetitions is A
If it is equal to or more than three times, it is determined as Yes, and if it is less than No, it is determined as No.

If No, the number of black pixels B _A (x, y) at each position of the pseudo binary image is stored in the storage step 12, and the pseudo binary image is initialized to the binary image 102. Then, the process returns to the target pixel determination step 3. Until the number of repetitions is determined to be Yes in the averaging step 10, the above steps of randomly distributing the pixel of interest and creating a pseudo binary image are repeated as described above.

The averaging step 10 calculates the number of repetitions A =
When 3 is determined, the average pseudo binary image step 13 obtains an average pseudo binary image based on the number of black pixels at each position and the number of repetitions.

The average pseudo binary image step 13 is a pseudo binary
The number of black pixels at each position of the value image is B _A , pseudo 2
Assuming that the number of creations of the value image is A,

(Equation 3) Then, the average number of blacks within one pixel is obtained, and an average pseudo binary image indicating the average number of blacks at each position is obtained.

Next, the multi-value conversion step 15 divides the average pseudo binary image into scan matrices of 8 × 8 units (V × W units) and according to the ratio of black and white in each scan matrix. A multi-valued image 105 shown in FIG. 8 including (M ′ / V) × (N ′ / W) pixels is obtained. However, if the sum of black in the scan matrix exceeds V × W = 8 × 8 = 64, the value is 64. The multi-valued image 105 is composed of 20 pixels vertically, 20 pixels horizontally and 400 pixels in total.

Here, the moire according to the present invention and the moire according to the conventional method will be described in detail with reference to FIG.

In FIG. 15, a multi-valued image 111 is obtained by converting a binary image 110 having a threshold matrix size of 10.times.10 into a scan matrix having a size of 8.times.8. Since the size of the threshold matrix is different from the size of the scan matrix, periodic moire occurs in the multi-valued image 111. FIG. 112 shows a multi-valued image 111.
Is represented by a numerical value. The Z-axis direction in FIG. 112 indicates the number of black pixels in one scan matrix area, and means a gray scale. From FIG. 112, it can be seen that moiré having periodicity occurs.

On the other hand, a multi-valued image 122 is obtained by forming a pseudo-binary image 121 in which the binary image 110 is once dispersed at random, and converting the binary value to a scan matrix size of 8 × 8. FIG. 123 illustrates the multi-value image 122 by numerical values. From the multi-valued image 122 and FIG. 123, it can be seen that the moire with periodicity disappears and approaches the original drawing (50%).

In the above embodiment, the case where the target pixel is a black pixel has been described. However, the present invention is not limited to this, and a white pixel may be used as the target pixel.

Further, the input binary image is enlarged (or reduced).
However, the input binary image may be used as it is.

Also, a case has been shown in which a pseudo binary image is created a plurality of times, and the pseudo binary image is averaged to obtain an average pseudo binary image. A multi-value image may be created based on the value image.

The pixel of interest may be randomly selected as in the first and second embodiments.

Each step of the first to third embodiments is preferably stored in a computer-readable storage medium.

[0124]

As described above, according to the present invention, when converting a binary image to a multi-valued image, the multi-valued image is not directly created from the binary image, but is converted from the binary image to black once. A pseudo binary image in which pixels are randomly dispersed is created, and the pseudo binary image is converted into a multi-valued image using a scan matrix having a size different from or the same as the threshold matrix.

Therefore, there is no occurrence of periodic moire in the converted multi-valued image. Further, since the pseudo binary image has no periodicity for each threshold matrix (m × n), there is an effect that a multi-valued image of a scan matrix suitable for the application can be obtained.

[Brief description of the drawings]

FIG. 1 is a flowchart of a first embodiment.

FIG. 2 is an explanatory diagram of an input binary image (50% area ratio);

FIG. 3 is an explanatory diagram of an interpolated binary image.

FIG. 4 is a binary image of a threshold matrix unit (an enlarged view of FIG. 3);

FIG. 5 is an explanatory diagram of a reference pixel determination step.

FIG. 6 is an explanatory diagram of a reference pixel determination step.

FIG. 7 is an explanatory diagram of a pseudo binary image.

FIG. 8 is an explanatory diagram of a multi-value image.

FIG. 9 is a flowchart of the second embodiment.

FIG. 10 is a histogram of a moving distance.

FIG. 11 is an explanatory diagram of a reference pixel determination step.

FIG. 12 is an explanatory diagram of a reference pixel determination step.

FIG. 13 is a flowchart of the third embodiment.

FIG. 14 is a histogram of a moving distance.

FIG. 15 is a photograph illustrating the reduction of moire due to dispersion according to the present invention.

Claims (8)

[Claims] 1. A binary multi-valued image conversion method for converting a binary image obtained using a threshold value matrix into a multi-valued image by using a scan matrix having a size different from or the same as the threshold value matrix, A first step of determining a target pixel by focusing on one of a colored pixel and a white pixel; and a second step of determining a moving position of the target pixel determined in the first step in the binary image. And a third step of dispersing the pixel of interest at the movement position determined in the second step; and a step of obtaining a multi-valued image based on the binary image obtained from the result of the third step A method for converting a binary / multi-valued image, comprising: 2. The method according to claim 1, wherein in the first step, the pixel of interest is determined by a probability of 1 / total number of pixels of interest. 3. The method according to claim 1, wherein the first step determines the pixels of interest in order.
Value multi-value image conversion method. 4. The method according to claim 1, wherein the second step selects one pixel from the pixels adjacent to the target pixel determined in the first step. The binary / multivalued image conversion method as described in the above. 5. The method according to claim 2, wherein the second step determines a moving distance of the target pixel based on a frequency of a moving distance of the target pixel which is measured in advance from a result obtained by repeating the trial. 4. The method according to claim 1, further comprising the steps of: 6. The method according to claim 1, wherein the second step includes determining a moving distance of each pixel of interest such that the frequency of movement of the pixel of interest has a desired histogram with respect to a distance to be moved. The binary / multi-value image conversion method according to any one of claims 1 to 3. 7. In the second step, the moving distance r of the pixel portion of interest is calculated as follows: r = [(m + n) / 20] · (G1-2 × | G2 | +1
0) where m and n are integers and the size of the threshold matrix, G1 and G2 are normal random numbers with mean 0 and variance 1 obtained by the Box-Muller method, and G1 = (− 2 × log (R1)) ) ^1/2 × cos (2π ×
R2) G2 = (− 2 × log (R1)) ^1/2 × sin (2π ×
R2) The method according to any one of claims 1 to 3, wherein R1 and R2 are determined from [0,1] uniform random numbers. 8. The method according to claim 3, wherein the third step disperses the colored pixel as the target pixel only when the pixel existing at the movement position is a white pixel, and only when the pixel existing at the movement position is a colored pixel. 4. The method according to claim 1, wherein the white pixels as the target pixels are dispersed.