JPH07114637A - Image processor - Google Patents

Abstract

PURPOSE:To prevent texture generated in plural density areas and to reproduce a half-tone image with high quality. CONSTITUTION:Respective density levels of a specific number of pixels including an aimed pixel, e.g. four pixels are temporarily stored in a 2X2 window of an aimed pixel selecting means 1. In a texture generation density range detecting means 3, plural density areas where the texture is generated are previously set and the detecting means detect whether or not the density level of the aimed pixel is in a density area. When that is detected, a noise adding means 7 adds a random noise to the four pixels to vary the respective density levels of those pixels, and improve deviation of the pattern of binarized levels characteristic to the texture.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an image processing apparatus capable of reducing a visual discomfort caused by a texture generated in an image having a plurality of gradations.

[0002]

2. Description of the Related Art An error diffusion method is used to represent a halftone image in an image. In this error diffusion method, binarization is performed with a fixed threshold value, and the difference between the density of the target pixel corresponding to the scanning and the density distribution error from the peripheral pixels is added to the new density. Diffusion to the target pixel as a distribution error is repeated for each scan. According to this error diffusion method, an image having a plurality of gradations can be expressed, and a halftone can be expressed with high quality when output to a printer or the like.

[0003]

However, this error diffusion method has a problem that a peculiar pattern (texture) appears at a predetermined density portion depending on the degree of shading, which is an eyesore. 15 and 16 are views showing this texture. In the figure, 0 (white) gradually increases in the vertical direction.
An image with a density change of 255 (black) is represented,
A texture is generated in each specific density area. It should be noted that these FIGS. 15 and 16 are divided into two figures for the sake of space.

The density range in which the texture is generated in this experiment is A (density range of 80 to 92) and B (140 to 1).
46), C (168 to 178), D (196 to 200)
There were four places. A large number of black pixels suddenly appear in the pattern of black and white pixels (binary data) formed at these four points by the error diffusion process, and it seems that the density change suddenly changes. It cannot be changed. Even in an actual halftone image, the textures at these four places give a visually uncomfortable feeling to the image.

The present invention has been made in order to solve the above problems, and an image processing apparatus capable of preventing a texture generated in a plurality of density regions by an error diffusion method and reproducing a halftone image with high quality. Is intended to provide.

[0006]

In order to achieve the above object, the image processing apparatus according to the first aspect of the present invention is such that a multi-value density signal corresponding to each pixel is inputted and each pixel is subjected to error diffusion processing. In an image processing apparatus for reducing a texture generated in an image composed of a target image, a density level of a target pixel (f _{m, n} ) sequentially input corresponding to a scanning signal is temporarily stored in a predetermined number of pixels including the target pixel. The pixel selection means (1) and the density range in which the texture is generated are set in advance, and the texture generation is detected depending on whether the density level of the input target pixel (f _{m, n} ) is within the texture generation density range. Texture generation density range detection means (3)
A random noise means (9, 11) for randomly outputting density noise corresponding to each pixel having the predetermined number of pixels by detecting the generation of texture by the texture generation density range detection means; A noise adding means (7) for adding the noise output from the random noise means to each pixel with respect to a predetermined number of pixels output from the selection means is provided.

An image processing apparatus according to a second aspect of the invention is an image processing apparatus that receives a binarized signal obtained by binarizing the density of each pixel by error diffusion processing and reduces the texture generated in an image composed of each pixel. , A pixel column storage means (2) for temporarily storing the binarized signals of the target pixel (f _{m, n} ) sequentially input corresponding to the scanning signal in a predetermined number of pixels including the target pixel.
0), a texture pattern setting means (22) in which a binarized signal corresponding to a texture is set in advance as a texture pattern corresponding to the predetermined number of pixels, and the texture element of the texture pattern setting means has the pixel row A pattern recognition means (24) for detecting texture generation depending on whether or not the binarized signals of a predetermined number of pixels of the storage means match, and the texture generation by the pattern recognition means is detected, so that And rearrangement means (26) for randomly inverting the binarized signal of an arbitrary pixel with respect to the binarized signal.

[0008]

According to the image processing apparatus of the present invention, the texture is reduced based on the density signal of the multi-value density level before binarization, and the target pixel selecting means 1 corresponds to the scanning signal and the target pixel f. _{When m and n} are sequentially input, this density level is temporarily stored in a predetermined number of pixels including the target pixel. On the other hand, the texture generation density range detection unit 3 is set with a density range in which a texture is generated in advance, and whether or not the density level of the target pixel f _{m, n} input from the target pixel selection unit 1 is within the texture generation density range. Detects the occurrence of texture. When the texture generation by the texture generation density range detection means 3 is detected, the random noise means 9, 11 are generated.
Outputs randomly the density noise corresponding to each pixel having the predetermined number of pixels. The noise adding means 7 is provided for the predetermined number of pixels output from the target pixel selecting means 1.
The noise output from the random noise means 9 and 11 is added to each pixel. As a result, the pixel of interest having a multi-value density level can be changed by adding random noise in units of a predetermined number of pixels including the pixel of interest to change the density level, Improves the bias of the binarization level and reduces the texture. As a result, the sharp change in the density change is improved to be smooth. In the image processing device according to the second aspect, the binarized signal after binarization of each pixel is input, and the bias of the binarized level is improved to reduce the texture.

[0009]

1 is a functional block diagram showing a first texture processing unit 2 using a random noise method according to a first embodiment of an image processing apparatus of the present invention. The texture reduction process by the random noise method is performed before the binarization process of the pixel signal.

The scanning signal and the density signal of the image are inputted to the target pixel selecting means 1, and the target pixel selecting means 1 outputs the density signal S1 of each of the four peripheral pixels including the target pixel f corresponding to the scanning signal.

Therefore, the target pixel selection means 1 is provided with a window section. This window part is
The density signals of 4 pixels around 2 × 2 including the target pixel f are stored. This 2 × 2 window will be described with reference to FIG. As shown in the figure, m is the main scanning direction and n is the sub scanning direction, and the pixel of interest f is sequentially scanned in the m and n directions in response to the scanning signal.

In the window portion _{, if the} target pixel f on the coordinate axis is f _{m, n} , the target pixel f _{m, n} is 1 in the main scanning direction.
And f _{m + 1, n} of the pixel neighboring these f _{m, n} and f _{m + 1, n} arranged next in the sub-scanning direction with respect to f _m, and _{n + 1, f m + 1} , n + 1 of the total The density levels of 4 pixels are sequentially stored, and the blocks of 4 pixels are sequentially scanned in the scanning direction. Therefore, the window portion corresponds to the scanning signal and the main scanning m as shown by the dotted line in the figure.
By skipping one pixel in each of the n-direction and the sub-scanning direction, the window is constructed in units of pixel blocks F that do not overlap the previous four pixels.

The texture generation density range detecting means 3 includes
It is detected whether or not the density level of the target pixel f _{m, n} corresponding to the scanning signal is in any of the texture generation density ranges A to D, and the detection signal S2 is detected at the time of detection, and the non-detection signal S3 is detected at the time of non-detection. Is output. The detection signal S2 is AD
It is added that the texture generation density range is shown. When the non-detection signal S3 is detected, the original density signal is output without performing the noise addition processing by this apparatus. Therefore, the texture generation density range setting means 5 is preset with the density level ranges A to D in which the textures are generated.

The noise adding means 7 receives the output of the detection signal S2 when the density of the pixel of interest f _{m, n} is within the density density range of the textures A to D, that is, for each pixel of the pixel block F. Add noises respectively and add signal S
4 is output.

This noise is generated by the noise source 9. In the noise source 9, as shown in FIG. 3, noises a and b respectively corresponding to the respective texture generation density ranges A to D are set in pairs of positive and negative values. The noises a and b are both effective values for suppressing each texture, and are obtained in advance by experiments. The noises a and b have the same density level as the density signal.

The noises a and b corresponding to the texture generation density range of the detection signal S2 are paired with the noise signal S5.
Is output to the randomizing means 11. The randomizing means 11 outputs the noise signal S5 in two pairs a1, a2, b1, b.
Output as 2. At the same time, two pairs of noise, which is a total of four noises, are output from the pixel block F by the random number generation.
It is distributed to any one of the two pixels and output.

Therefore, the noise adding means 7 adds the noises a1, a2, b1 and b2 selected by the random numbers to the respective pixels f of the pixel block F, and outputs them as data f'after the noise addition. An example of this state is shown in the following formula (1). f'm _{, n} = f _{m, n} + a1 f'm _{+ 1, n} = f _{m + 1, n} + b1 f'm _{, n + 1} = f _{m, n + 1} + b2 f ' _{m + 1, n + 1} = F _{m + 1, n + 1} + a2

The noises a1, a2, b1 and b2 are replaced by random numbers every time the pixel block F is scanned, and it is uncertain to which pixel they are added. However, it is assumed that these noises a1, a2, b1 and b2 are always added to any of the four pixels. As a result, since the noises a1 and a2 and the paired noises b1 and b2 have positive and negative noise values, the density level of the entire pixel block F does not change even after noise addition.

The density level comparing means 13 compares whether or not the addition signal S4 output by the noise adding means 7 is within a predetermined density level range, and outputs the comparison result signal S6 only when it is within the density level range. To do. Therefore, in the threshold value setting means 15, a density level range of pixels which each pixel can take, that is, a density level (gradation) of 0 to 255 is set as a threshold value and is input to the density level comparison means 13. There is.

Then, the density level comparing means 13 outputs the comparison result signal only when all four pixels of the pixel block F showing the content of the addition signal S4 after the noise addition are within the density level range 0-255. Outputs S6. This is performed because the density level of the addition signal S4 may be less than 0 or 256 or more after adding the noises a and b. Therefore, the addition signal S4 is 0-255.
If it is outside the range of the density level of, the comparison result signal S6
Is not output.

The selection output means 17 includes a detection signal S2 of the texture generation density range detection means 3, a comparison result signal S6,
The non-detection signal S3 is input, and either signal is selectively output as the output signal S7. At this time, when the non-detection signal S3 is output, only the non-detection signal S3 is output, and when the comparison result signal S6 is input, the comparison result signal S6 is output, and the comparison result signal S6 is output. When is not input, the output is prioritized so that the detection signal S2 is output.

This is because the density level after noise addition is within the range of 0 to 255 regardless of the result of the noise addition processing in the preceding stage. Therefore, the output signal S7 of the selective output means 17 has a density level within the range of 0 to 255 regardless of whether the comparison result signal S6 or the detection signal S2 which is the original density signal is output. Of course,
The same applies when the non-detection signal S2 is output.

The density levels of the pixel of interest f input by the above-mentioned respective components are density ranges A to D in which the texture is generated.
In other cases, the noise addition process is not performed on the pixel block F including the target pixel f, and the non-detection signal S3 is output. On the other hand, when the density level of the target pixel f is in the density range A to D in which the texture is generated, noise addition processing is performed on the pixel block F including the target pixel f, and the output signal S7 is output.

The process of image processing by the random noise method having the above configuration is shown in the flowchart of FIG. In the following, according to this flow, each pixel is configured in a predetermined density range (density level is 0 to 255) (SP
1) After the error diffusion process is performed on the aggregate (image) of pixels (SP2), it is determined whether the texture reduction process is performed (SP3). In the texture reduction processing, the range of densities A to D at which the texture is generated is set in advance by the texture generation density range setting means 5 (see FIG. 3), and the density level of the target pixel f _{m, n} is the density A of this texture generation range. It is detected whether or not it is within ~ D (SP4).

When the density level of the target pixel f _{m, n} is within the range of the texture density A (80 to 92) (SP5
−1), a pixel block F (2 including this target pixel f _{m, n}
Noise (a = 30, b = −30) having a value corresponding to the texture A is added to a total of 2 pairs with respect to × 2 (4 pixels in total) (SP6-1). In the noise addition process at this time, random noise is added by the noise addition means 7 as described above, and a total of two pairs of noises a and b are added. Therefore, the density levels of the entire pixel block F before and after the noise addition are added. There is no change in, and the density of the entire image is preserved.

After addition of random noise, the density levels of the individual pixels in the pixel block F are compared again (SP
7) In any case, if the density level deviates from the predetermined density level (0 to 255), the noise addition processing by SP6-1 is not performed (SP8). Specifically, as described above, the selection output operation by the selection output means 17 selects the detection signal S2 before addition of random noise and outputs it as the output signal S7.

The noise addition processing by SP5 and SP6 is similarly performed for each texture generation density range B to C (SP5-2 to SP5-4 and SP6-
2 to SP6-4), as shown in FIG. 3, the noise values a and b corresponding to each texture generation density range are added. 5 and 6 are diagrams showing the state after the texture reduction processing by the random noise method. It should be noted that FIGS. 5 and 6 are divided into two figures for convenience of space. After the texture reduction process by the random noise method described above, the binarization process for obtaining the binarized signal is performed.

Next, the second texture processing unit 21 according to the rearrangement method of the second embodiment of the image processing apparatus of the present invention will be described. In this rearrangement method, after the binarization processing, after the textures shown in the texture generation density ranges A and C shown in FIGS. 15 and 16, the texture reduction processing is performed by recognizing the peculiar pattern. In addition, in this embodiment, a case where a pair of pixels a and b are adjacent to each other with a predetermined area ratio (2: 1) will be described as an example. This different size pixel a,
b is a pixel for the corresponding different size TPH.

* Texture Pattern 1 (Regarding Texture Generation Density Range A1) In the texture generation density range A shown in FIG. 15, a portion A1 where a continuous line is generated in the main scanning direction m is shown in FIG.
The pixel row of (a) is enlarged. FIG. 7A shows the pair of pixels a and b, and the pair of pixels a and b.
Is extracted in the sub-scanning direction n by a predetermined column. In the texture generation pattern of the texture generation density range A1 shown in the figure, the pixel a is W (color arrangement W; Whit) in 15 columns continuously in the sub-scanning direction n.
e) After the binarized data in the state where the pixel b is B (Black) are lined up, the pixel a suddenly becomes B after a predetermined column scanning (the 16th column from the top; f _{m, n} portion). . As a result, FIG.
As shown in, the boundary-like texture that is continuous in the main scanning direction m is generated at a position below the texture generation density range A.

* Texture pattern 2 (Regarding texture generation density range A2) In addition, of the texture generation density range A shown in FIG.
A portion A2 where a continuous line is generated in the sub-scanning direction n is enlarged as a pixel row in FIG. FIG. 8A shows a pair of pixels a and b extracted in a predetermined column in the main scanning direction m and a predetermined column in the sub-scanning direction n. In the example of this figure, the texture generation pattern of the texture generation density range A2 is
The binarized data in which the pixel a is W and the pixel b is B is arranged in the main scanning direction m, and is continuous in the sub scanning direction n. As a result, as shown in FIG. 15, at the position above the texture generation density range A, a continuous linear texture is generated in the sub-scanning direction n.

* Texture Pattern 3 (Regarding Texture Generation Density Range C) Next, the texture generation density range C shown in FIG.
The pixel row of (a) is enlarged. In FIG. 9A, the pixel a column is extracted in a predetermined column in the sub-scanning direction n. The texture generation pattern in the texture generation density range C shown in the drawing is such that after the binarized data of W and B are alternately arranged in the sub-scanning direction n, after a predetermined column scanning (the thirteenth column from the top; f _{m, n} and f
Both of the two pixels a become B in the ( _{m, n + 1} portion). As a result, as shown in FIG. 16, in the lower position of the texture generation density range C, a continuous boundary texture is generated in the main scanning direction m.

The texture having the above-mentioned peculiar pattern is rearranged by means for recognizing the pattern and the texture reduction processing is performed. FIG. 10 is a functional block diagram showing the second texture processing unit 21 based on this rearrangement method.

The image scanning signal and the binarized signal level (BorW) are sequentially input for each pixel and temporarily stored in the pixel column storage means 20. The pixel column storage means 20 is
The storage range of the pixel row is different for each of the three texture patterns A1, A2, and C described above. After the predetermined pixel row is stored, the pattern recognition means 2 is used as the pixel row signal S10.
4 is output.

The texture pattern setting means 22 presets and stores the above-mentioned three texture patterns A1, A2 and C. The pattern recognition means 24 uses the pattern signal S
11 and the pixel column signal S10 are compared, and if they do not match, the pixel column signal S10 is directly output to the outside without performing the texture reduction processing on the pixel column signal S10. If they coincide with each other, the pattern signal S15 is output to the rearrangement means 26 as the occurrence of the texture.

The rearrangement means 26 randomly selects one or more pixels in the pixel row according to each texture pattern A1, A2, C, and inverts the binarization level of the selected pixel. Texture reduction processing.
After the rearrangement, the rearrangement signal S17 is output. The randomizing means 27 randomly selects the pixels, and the random numbers are different for each texture pattern, and outputs a corresponding random number generation signal S16.

The selection output means 28 includes a pixel column signal S10,
The rearrangement signal S17 is input, and these are switched and output as binarized data S19. Here, the pixel column signal S
When 10 is input, only the pixel column signal S10 is output, and when the rearrangement signal S17 is input, this rearrangement signal S17 is selectively output.

The structure and operation of the texture reduction processing in the above structure will be described for each texture pattern. * Regarding Texture Pattern A1 As shown in FIG. 7A, the pixel column storage unit 20 temporarily stores a pair of pixels a and b for 16 lines in the sub-scanning direction, that is, a pixel block F of 32 pixels in total. The pixel of interest at this time is f
_{Let m and n} . The pixel of interest moves one each in the main scanning direction m and the sub-scanning direction n.

In the texture pattern setting means 22, as shown in FIG. 7A, only the target pixel f _{m, n} has a pattern in which the binarization level is different in the pixel row a. Therefore, when the pixel column signal S10 matches the texture pattern shown in FIG. 7A, the pattern recognition means 24 outputs the pattern signal S15 to the rearrangement means 26.

The randomizing means 27 uses the random number generation signal S1.
6 randomly generates one of the random numbers i = 1 to 15 and outputs it to the rearrangement unit 26. The rearrangement means 26 receives the pattern signal S15 and receives the target pixel f _{m, n.}
Then, the binarization level of the pixel f _{m + 1, n} next to the pixel is inverted (B → W). Further, the binarization levels of a pair of adjacent pixels corresponding to the random number i of the random number generation signal S16 are inverted. For example, as shown in FIG. 7B, when the random number i = 5, the binarization level of the pixel f _{m, n-5} is inverted (W → B), and the pixel f _{m + 1, The} binary level of _n-5 is also inverted (B → W).

Here, the rearrangement means 26 determines the pixel f _{m + 1, n.}
And corresponding to the fact that a total of 2 pixels of pixel f _{m + 1, n-5} are inverted (B → W), the binarization level of pixel f _{m, n-5} is inverted (W → B). There is. This is performed in order to save the density of the entire pixel block F.
Based on the fact that = 2: 1, the inversion at the pixel b2 location corresponds to the inversion at the pixel a1 location.

In this way, when the occurrence of texture is recognized, it is possible to reduce the texture in the texture generation density range A1 by inverting the binarization level by generating random numbers while preserving the density of the entire pixel block F '. it can. The state after the texture reduction processing by the above configuration operation is shown in FIG. 12 after the texture reduction.
Note that FIG. 12 and FIG. 13 are continuous and are divided into two views due to space limitations.

* Regarding Texture Pattern A2 The pixel column storage means 20 temporarily stores a total of 30 pairs of pixel blocks F, for a total of 5 pairs of pixels a and b for 3 lines in the sub-scanning direction, as shown in FIG. 8A. To do. The pixel of interest at this time is f
_{Let m and n} . The pixel of interest moves every four pixels in the main scanning direction m and every three lines in the sub-scanning direction n.

As shown in FIG. 8A, the texture pattern setting means 22 includes the target pixel f _{m, n} in the main scanning direction m.
There are a total of 10 alternating binary levels (W, B, W, B, ...).
A pattern in which pixels continue and this state continues for three lines in the sub-scanning direction n is stored and set. Therefore, the pattern recognition means 24 outputs the pattern signal S15 to the rearrangement means 26 when the pixel column signal S10 matches the texture pattern shown in FIG.

The randomizing means 27 uses the random number generation signal S1.
6 randomly generates one of the random numbers i = 1, 2,
It outputs to the rearrangement means 26. In the rearrangement means 26, a total of 4 pixels in the main scanning direction m and 3 pixels in the sub-scanning direction, as shown in FIG. 8B or 8C, corresponding to the random number i of the random number generation signal S16. Rearrangement is performed on the pixel block F ′ of 12 pixels. The pixel block F'corresponds to the movement of the target pixel. When the random number i = 1,
As shown in FIG. 8B, the rearrangement (inversion of the binarization level) is not performed on the pixel block F ′ of 12 pixels in total.

When the random number i = 2, four pixels f _{m, n} , f _{m + 1, n} , f _{m + 2, n} , f from the target pixel in the main scanning direction m
_The binary levels of three random pixels of _{m + 3, n} are inverted. For example, as shown in FIG. 8C _, the binarization level of the target pixel f _{m, n} is inverted and (W →
B), the binarization level of the pixel f _{m + 1, n} is inverted (B →
W), and the binary level of the pixel f _{m + 3, n} is also inverted (B →
W) is done.

Here, the rearrangement means 26 determines the pixel of interest f.
Corresponding to the inversion of the binarization level of _{m, n} (W → B) _, a total of two pixels of pixel f _{m + 1, n} and pixel f _{m + 3, n} are inverted (B → W). The density of the entire pixel block F is saved. This is based on the fact that the pixel a: b = 2: 1 and the inversion at the pixel b2 location corresponds to the inversion at the pixel a1 location.

Thus, this texture pattern A2
When the generation of the texture is recognized, the texture in the texture generation density range A2 can be reduced by inverting the binarization level by generating a random number while preserving the density of the entire pixel block F ′. The state after the texture reduction processing by the above configuration operation is shown in FIG. 12 of FIG. 12 and FIG.

* Regarding Texture Pattern C The pixel column storage means 20 temporarily stores the pixel block F for 25 lines of the pixel a in the sub-scanning direction as shown in FIG. 9A. The pixel of interest at this time is fm _{, n} . The pixel of interest moves one each in the main scanning direction m and the sub-scanning direction n.

In the texture pattern setting means 22, as shown in FIG. 9A _{, all} the pixels f _{m, n + 1} adjacent to the target pixel f _{m, n} by one line have the same binarization level.
A pattern in which the binarization level is alternately generated for a total of 11 pixels in which _{m, n + 2} is the binarization level W and the target pixel f _m, n is the pixel f _{m, n-11 in} the sub scanning direction n I remember. Therefore, in the pattern recognition means 24, the pixel column signal S1
When 0 matches the texture pattern shown in FIG. 9A, the pattern signal S15 is output to the rearrangement unit 26.

The randomizing means 27 uses the random number generation signal S1.
6 randomly generates _one of random numbers i ₁ = 1 to 15 and i ₂ = 1 to 7, and outputs the random number to the rearrangement unit 26. In the rearrangement means 26, as the first rearrangement, the binarization level of the pixel corresponding to the random number i ₁ of the random number generation signal S16 and the binarization level of the pixel corresponding to i ₁ + i ₂ are set to B and B, respectively. To do. For example, as shown in FIG.
When the random numbers i ₁ = 5 and i ₂ = 2, the pixel f _{m, n-i1}
Let B be the binarization level of (f _{m, n-5} ) and the pixel f _{m, n- (i1 + i2)} (f _{m, n-7} ).

As a result, as shown in FIG. 9B, the pixel f
The front-back direction of the sub-scanning direction n of _{m, n-i1} (f _{m, n-5} ), that is, the pixel f _{m, n-i1-1} (f _{m, n-6} ) and the pixel f _{m, n-i1 + 1} (F
_{Each of m, n-4} ) is followed by the same binarization level (B) of 3 pixels. Similarly, the pixel f _{m, n- (i1 + i2)} (f
_{m, n-7} ) in the front-back direction of the sub-scanning direction n, that is, the pixel f
_{m, n- (i1 + i2) -1} ( _{fm , n-8} ) and pixel f
_{In m, n- (i1 + i2) +1} (f _{m, n-6} ), 3 pixels are the same 2
The digitization level (B) will continue. Since the texture reduction cannot be performed only by the rearrangement of the first stage,
Rearrange columns.

By the first rearrangement, the pixel f _{m, n-i1} (f
_{When both} adjacent lines (f _{m, n- 4} , f _{m, n-6} ) including _{m, n-5} ) have the same binarization level (B), FIG.
As shown in (c), the second rearrangement is further performed and the pixel f
_{Let m, n-i1 + 1} (f _{m, n-4} ) be W. Similarly, f
_{m, n- (i1 + i2)} (f _{m, n-7} ) including both adjacent lines (f
_{m, n-8} , fm _{, n-6} ) are all the same binarization level (B)
_Therefore , the pixel f _{m, n- (i1 + i2) +1} (f _{m, n-6} ) is set to W.

9A and FIG. 9C are compared before and after the rearrangement, the rearrangement means 26 shows a total of two pixels of the pixel f _{m, n-5} and the pixel f _{m, n-7.} Is inverted (W → B), the binarization levels of the pixels f _{m, n-4} and f _{m, n-6} are inverted (B → W), and the entire pixel block F is The concentration of is preserved.

In this way, when the occurrence of texture is recognized, it is possible to reduce the texture in the texture occurrence density range C by inverting the binarization level by generating random numbers while preserving the density of the entire pixel block F. The state after the texture reduction processing by the above configuration operation is as shown in FIGS.
13 out of three.

The process steps of image processing by the rearrangement method having the above configuration are shown in the flowchart of FIG. In the following, according to this flow, each pixel is configured in a predetermined density range (density level is 0 to 255) (SP1), and the error diffusion process and the binarization process are performed on the pixel aggregate (image). After that (SP2), the density range in which the texture generation density range A (A1, A2) or C occurs is detected (SP3). This texture generation density range detection is
Visually detect the print status or similar screen display,
The detection range is set in advance.

Thereafter, the binarized signals of the pixels are sequentially input by scanning, and the texture generation density range A (A1, A2)
(SP4), the above-mentioned pixel block F
After pattern recognition corresponding to this texture generation density range A (A1, A2) is performed in units (SP5), rearrangement for texture reduction is performed (SP6). On the other hand, in the case of the texture generation density range C (SP7), after the pattern recognition corresponding to the texture generation density range C is performed for each pixel block F (SP8), the corresponding rearrangement is performed (SP8). SP6). In the texture reduction process at this time, random rearrangement is performed as described above,
In addition, the density level of the entire pixel block F is stored, and the density of the entire image is stored.

Image processing apparatus 1 described in each of the above embodiments
Is applied to the stencil printing apparatus 30 shown in the block diagram of FIG. The stencil printing device 30 reads a document 31 with a scanner unit 33 having an image sensor 32 and outputs it to an image processing unit 35 via an A / D converter 34. In the image processing unit 35, the output of the scanner unit 33 is subjected to data correction such as shading in the preprocessing unit 35a, and then data processing by the error diffusion method is performed in the data processing unit 35b.
It is output after binarization.

When performing the texture processing by the above-mentioned first texture processing section 2, this data processing section 35 is used.
The first texture processing unit 2 is provided in b, and texture processing is performed to output the output signal S7. After that, the binarizing means outputs the binarized data S20. On the other hand, when performing the texture processing by the second texture processing unit 21, the second portion is output to the output portion of the binarized data S20 of the binarizing means.
The texture processing unit 21 is provided to perform texture processing and output the binarized data S19. Binarized data S19 or S20 output from the image processing unit 35 to the outside
Is output to the printing unit 36. This printing means 36
It comprises a plate making section 37 and a printing section 38. The plate making part 37 is
The image of the original 31 is made by heat-sensitive plate making from the stencil sheet.
At this time, based on the binarized signals S19 and S20, the heat-generating means such as a thermal head in which the adjacent heat-generating elements are different in size drive these heat-generating elements corresponding to the binarized levels of the pixels a and b. Control.

The printing section 38 has a rotary drum on which the stencil sheet that has been made is attached. While rotating the rotary drum, the printing paper is fed to the plate cylinder and the ink is ejected from the inside of the rotary drum, so that an image is formed on the printing paper by the ink that has passed through the stencil sheet.

Further, the stencil printing apparatus 30 is provided with a plate discharging section 39 for separating the used stencil sheet from the rotary drum, and a sheet feeding section 40 for feeding the printing sheet. In addition, each of the above-mentioned components is integrally controlled by the controller 41.

The image processing apparatus described above can be applied to other than the stencil printing apparatus. That is, the binarized signal S19,
In S20, the same operation and effect as described above can be obtained in any device having a printing section in which the ratio of adjacent heating elements is different, such as a configuration in which the printer is simply provided with a thermal head. In the above embodiment, the ratio of the pixels (heating elements) is set to 2: 1, but the ratio is not limited to this and may be another ratio.

[0062]

According to the image processing apparatus of the first aspect, the texture is reduced based on the density signal of the multi-value density level before the image data is binarized, and the target pixel selection means selects the target pixel. The density level having a predetermined number of pixels to be stored is temporarily stored, and the texture generation density range detection means detects whether or not the density level of this pixel of interest is within the texture range, and when the texture generation is detected, noise addition is performed. Since the random density noise is added to each pixel by the means, each pixel in a specific density range in which the density level is changed in units of a predetermined number of pixels including the pixel of interest to generate a texture The bias of the pattern at the binarization level of is improved, and the texture can be reduced.

According to the image processing apparatus of the second aspect,
After the binarized signal after binarization of each pixel is temporarily stored in the pixel column storage means for a predetermined number of pixels, the pattern recognition means determines in advance whether the pattern corresponds to the texture, and if they match, an arbitrary value is determined. By inverting the binarized signal of the pixel, it is possible to improve the bias of the pattern of the binarized level and reduce the texture.

[Brief description of drawings]

FIG. 1 is a block diagram showing an image processing device by a random noise method according to a first embodiment of the present invention.

FIG. 2 is a diagram showing a scanning window.

FIG. 3 is a diagram showing noise values respectively corresponding to texture generation density ranges A to D.

FIG. 4 is a flowchart showing the operation of the first embodiment.

FIG. 5 is a diagram showing a state after texture reduction according to the embodiment.

FIG. 6 is a diagram showing a state after the texture is reduced.

FIG. 7 is a diagram for explaining the rearrangement method of the second embodiment, in which (a) is a diagram showing a texture pattern in a texture generation density range A1. FIG. 6B is a diagram showing a rearrangement pattern after the texture processing.

FIG. 8A is a diagram showing a texture pattern in a texture generation density range A2. (B), (c) is a figure which shows the rearrangement pattern after texture processing, respectively.

9A is a diagram showing a texture pattern in a texture generation density range C. FIG. (B) is a figure which shows the rearrangement pattern of a texture processing process. FIG. 6C is a diagram showing a rearrangement pattern after texture processing.

FIG. 10 is a block diagram showing an image processing device by a rearrangement method according to a second embodiment of the present invention.

FIG. 11 is a flowchart showing the operation of the second embodiment.

FIG. 12 is a diagram showing a state after texture reduction according to the embodiment.

FIG. 13 is a diagram showing a state after the texture is reduced.

FIG. 14 is a diagram showing an example of application of this apparatus to a stencil printing apparatus.

FIG. 15 is a diagram showing texture generation.

FIG. 16 is a diagram showing the same texture generation.

[Explanation of symbols]

DESCRIPTION OF SYMBOLS 1 ... Target pixel selection means, 3 ... Texture generation density range detection means, 5 ... Texture generation density range setting means, 7 ... Noise addition means, 9 ... Noise source, 11 ... Randomization means,
13 ... Concentration level comparing means, 15 ... Threshold setting means,
Reference numeral 17 ... Selection output means, 20 ... Pixel column storage means, 22 ... Texture pattern setting means, 24 ... Pattern recognition means,
26 ... Rearrangement means, 27 ... Randomization means, 28 ... Selection output means.

─────────────────────────────────────────────────── ───

[Procedure amendment]

[Submission date] May 19, 1994

[Procedure Amendment 1]

[Document name to be amended] Statement

[Name of item to be corrected] Figure 5

[Correction method] Change

[Correction content]

FIG. 5 is a diagram showing a halftone image after texture reduction according to the embodiment. (Drawing substitute photograph)

[Procedure Amendment 2]

[Document name to be amended] Statement

[Name of item to be corrected] Figure 6

[Correction method] Change

[Correction content]

FIG. 6 is a diagram showing a halftone image after the texture reduction.
(Drawing substitute photograph)

[Procedure 3]

[Document name to be amended] Statement

[Name of item to be corrected] Fig. 12

[Correction method] Change

[Correction content]

FIG. 12 is a diagram showing a halftone image after texture reduction according to the embodiment. (Drawing substitute photograph)

[Procedure amendment 4]

[Document name to be amended] Statement

[Name of item to be corrected] Fig. 13

[Correction method] Change

[Correction content]

FIG. 13 is a diagram showing a halftone image after the texture reduction.
(Drawing substitute photograph)

[Procedure Amendment 5]

[Document name to be amended] Statement

[Correction target item name] Figure 15

[Correction method] Change

[Correction content]

FIG. 15 is a diagram showing a halftone image in a texture occurrence state.
(Drawing substitute photograph)

[Procedure correction 6]

[Document name to be amended] Statement

[Correction target item name] Fig. 16

[Correction method] Change

[Correction content]

FIG. 16 is a diagram showing a halftone image in the same texture generation state. (Drawing substitute photograph)

Claims (2)

[Claims] 1. An image processing apparatus for reducing a texture generated in an image composed of each pixel by inputting a density signal of a multi-value density level corresponding to each pixel, and a target pixel sequentially input corresponding to a scanning signal. (F _{m, n} )
The target pixel selecting means (1) for temporarily storing the density level of the target pixel with a predetermined number of pixels including the target pixel, and the density level of the target pixel (f _{m, n} ) to which the density range in which the texture is generated is set in advance. Has a predetermined number of pixels by detecting texture generation density range detection means (3) for detecting texture generation depending on whether or not is a texture generation density range, and detecting texture generation by the texture generation density range detection means. Random noise means (9, 11) for randomly outputting density noise corresponding to each pixel, and for a predetermined number of pixels output from the target pixel selecting means, noise output from the random noise means is applied to each pixel. An image processing apparatus comprising: a noise adding means (7) for adding to the. 2. An image processing apparatus, which receives a binarized signal obtained by binarizing the density of each pixel and reduces the texture generated in an image composed of each pixel, is sequentially input corresponding to a scanning signal. Pixel (f _{m, n} )
And a pixel column storage means (20) for temporarily storing the binarized signal of No. 2 in a predetermined number of pixels including the target pixel, and the binarized signal corresponding to the texture is preset as a texture pattern corresponding to the predetermined number of pixels. A texture pattern setting means (22) and a pattern recognition means (22) for detecting texture occurrence depending on whether or not a binary signal of a predetermined number of pixels of the pixel row storage means matches the texture pattern of the texture pattern setting means ( 24), and rearrangement means (26) for randomly inverting the binarized signal of an arbitrary pixel with respect to the binarized signal of the predetermined number of pixels by detecting the occurrence of texture by the pattern recognition means. An image processing apparatus comprising: