JP2003101776A - Threshold matrix forming method, image outputting system, program and recording medium - Google Patents

Abstract

PROBLEM TO BE SOLVED: To form a threshold matrix for expressing half tone in high quality by using on-off of a plurality of kinds of dots. SOLUTION: Gradation levels on interlaced start points side and end points side are picked up (S1, S2). Dot arrangement at both of the gradation levels is optimized (S3). Dot arrangement between both the levels is determined by interpolation (S4). From dot arrangement of the whole gradation levels which is determined by repeating the above processes, basic layout of a binary threshold matrix is formed (S5) The basic layout is made a common layout, and a plurality of threshold matrixes for half tone expression of at least ternary are formed (S6, S7).

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to the field of technology for expressing halftones by turning dots on and off, and in particular, expressing halftones by turning on and off a plurality of types of dots that differ in size, density and / or brightness. To this end, the present invention relates to a technique of creating a threshold matrix for quantizing multi-valued image data into three or more values.

[0002]

2. Description of the Related Art The technical background of the present invention will be described below. Various researches have been made so far on a method of creating a halftone image in digital image processing,
It is roughly divided into two types: the "dither method" and the "error diffusion method".
The dither method is lightweight and can perform high-speed processing.
It is suitable for low to medium quality applications, but measures against moire are essential. The error diffusion method has a heavy processing load and is not easy to speed up the processing as compared with the dither method, but is suitable for high image quality applications. As described above, since both methods have advantages and disadvantages, they are used properly according to the configuration of the image output apparatus to be applied and the type of the target image.

In the latter half of the 1900's, a halftone processing method was proposed which combines the advantages of these two methods. It is a method generally called “FM mask method”, “FM screen method”, “blue noise mask method” or the like. These are taxonomically dithered, and more specifically Bayer
This is a method of distributed matrix comparison method similar to the type.

In the FM mask method, frequency characteristics (Blue Noise (B
By using a threshold matrix with (N) characteristics),
Even though it is a mask comparison method, it can obscure the low frequency periodicity found in Bayer type dither and dot concentration type dither, increasing the resistance to moire and providing resolution characteristics close to error diffusion. I was able to. Therefore, various approaches have been attempted with attention from various fields including the printing field.

As an early attempt, Japanese Patent No. 2622429
As described in No. "Method and apparatus for halftoning a grayscale image using a blue noise mask," a completely random (white noise) dot pattern is Fourier transformed and filtered by a filter having a BN characteristic. Then, a method of trying to construct an ideal FM mask by an inverse Fourier transform is proposed.

In addition, SPIE Vol.1913 "The void-and-clu
ster method for dither array generation "(Robert U
lichney), there are sparse dots (voi
A method of optimizing a mask by comparing d) with a dense portion (cluster) of dots and exchanging dots is also proposed.

In any of the above methods, since the random pattern is used as the initial pattern, the finally obtained calculation results have different values each time, and the irregular distribution state is reconstructed to the homogeneous distribution state. Since it has to be applied, there is a problem that a huge calculation time is required depending on the size of the mask.

In order to avoid this problem, Japanese Patent Laid-Open No. 8-8
As seen in Japanese Patent Publication No. 0641, a method has been proposed in which an initial pattern created by an error diffusion process is used as a starting point. In this method, the designer can arbitrarily control the starting mask that is the starting point, so it is possible to eliminate the luck element and ensure the recursiveness of the calculation result, and to obtain a uniform distribution in the initial state. Not only improves the quality of the final mask created, but also
It has also succeeded in shortening the calculation time.

However, in the sequential decision method in which the gradation is optimized one by one from the initial pattern as the starting point,
The dot arrangement pattern determined immediately before limits the dot arrangement that can be selected next time. what if,
Even if the initial pattern can form a perfect dot arrangement in terms of anti-graininess and anti-texture, it is only specialized for the gradation level that is the starting point, and several dots are required for that dot arrangement. At the next gradation level that was added (or removed), although the quality was high, it was no longer a perfect dot arrangement. As the gradation level progresses, this deviation is accumulated one after another, and at the final gradation level, it is not unusual that the dot arrangement is far from the optimum dot arrangement. FIG. 15 schematically shows such a quality deterioration of the dot arrangement.

In order to suppress such deterioration of dot arrangement quality, various optimization functions and dot arrangement search methods have been proposed. However, at a specific gray level (1/255
128/255) and the dot matrix is sequentially obtained dot by dot to construct a threshold value matrix, and the problem that the restrictions due to the immediately preceding gray level are accumulated remains. None of them can be a decisive remedy.

Further, in order to add the ideal BN characteristic to the threshold matrix, it is necessary to make the threshold matrix of a very large size according to the resolution. Looking at the research results from various directions, at least for a 600 dpi image,
It is necessary to make a matrix of a size of 256 × 256, and at a higher resolution, a larger matrix size is needed. This is an enormous size, considering that the commonly used existing distributed dither matrix has a size of about 4 × 4 to 16 × 16. In the dithering method, which has the advantage of high speed, if the matrix is too large, implementation becomes difficult depending on the configuration of the image forming apparatus. Above all, the calculation load associated with mask creation becomes too large.

[0012]

In recent years, a method of expressing a halftone by turning on / off two or more types of dots having different sizes, densities, brightness, or both has been adopted in an image forming apparatus such as a printer. In such a halftone expression method, it is necessary to prepare threshold matrixes corresponding to respective dot types and use them to quantize multi-valued image data into three or more values.

An object of the present invention is to provide means for efficiently creating a threshold matrix for expressing a halftone with high quality by turning on / off a plurality of types of dots. Another object of the present invention is to provide means for realizing a high quality halftone representation using such a threshold matrix.

[0014]

According to the present invention, a plurality of threshold value matrices for quantizing multivalued image data into three or more values in order to express a halftone by turning on and off a plurality of types of dots are created. A method is provided, the main feature of which is, as described in claim 1, a threshold value of a threshold value matrix for binarizing multi-valued image data for expressing a halftone by turning on and off one kind of dot. It comprises a first step of determining the arrangement order, and a second step of creating each of a plurality of threshold value matrices for quantizing into three or more values in accordance with the threshold value arrangement order determined in the first step. Is.

In this way, by sharing the basic layout (threshold arrangement order) of a plurality of threshold matrices,
Even if the threshold matrix is switched according to the gradation level in the quantization of multi-valued image data, it is possible to give consistent dispersion characteristics to dots of different types with different dot diameters. It is possible to avoid the inconvenience of interfering with each other and generating a texture having a different period / tone. Further, since the arrangement order of the dots of each type is constant, it is possible to prevent the generation of unnecessary isolated dots due to the arrangement of large dots in the gaps of the matrix pattern that happened to occur. In addition, the threshold value arrangement order of the plurality of threshold value matrices can be optimized by two values. In other words, this means that the optimization is performed with the maximum dots, and thus the maximum value that is most likely to affect the granularity is the maximum. By improving the dispersion characteristics of dots, the graininess of small and medium-sized dots, which are more difficult to see, is also improved,
The graininess of the entire image can be raised.

Another method of creating a threshold matrix according to the present invention
As one of the main features, as described in claim 2, in the first stage, a gradation level which is discontinuous in all expressible gradation levels is selected, and the dot arrangement in the selected gradation level is optimized. The dot arrangement at each gradation level between the selected start point side and end point side gradation levels is optimized based on the optimized dot arrangement of the start point side and end point side gradation levels. The threshold arrangement order of the threshold matrix for binarization is determined based on the determined dot arrangement of each gradation level.

As described above, by optimizing the dot arrangement at the selected gradation level intervals, it is possible to suppress the accumulation of the deviation from the optimum dot arrangement, so that the general F
It is possible to avoid the inconvenience that the quality of dot arrangement gradually deteriorates as the gradation level changes, as seen on the M screen.

According to a preferred mode, as the initial dot arrangement pattern at the gradation levels on the start point side and the end point side, a dot pattern created by existing distributed dither or error diffusion processing is used. Since dot dispersibility is ensured in such a dot pattern, the calculation load for optimization can be reduced compared to the case where a completely random pattern such as a white noise pattern is used as the initial dot arrangement pattern. There are advantages.

Another method of creating a threshold matrix according to the present invention
One of the main characteristics is, as described in claim 3, in the first step, matching for including the optimized dot arrangement of the starting point gradation level in the initial dot arrangement pattern of the ending point gradation level. Processing is performed, and based on the initial dot arrangement pattern after this matching processing, the dot arrangement on the starting point side is saved while the dot arrangement on the end point side gradation level is determined according to the optimization conditions. With such a method, continuity between gradations can be secured, and it is effective in preventing the occurrence of pseudo contours and textures.

According to a preferred embodiment, as an optimization condition, an evaluation value obtained by applying a weighting filter in a predetermined range is calculated for each ON dot, and the ON / OFF of the dot for which the sum of the evaluation values is the minimum is calculated. The combination is used as the optimization condition. The method using such an optimization condition has an effect that the calculation load for the optimization can be reduced as compared with the optimization method that directly manipulates the frequency component using the Fourier transform like the blue noise mask creating method. .

[0021] Preferably, as the weighted filter,
A weighted value obtained by correcting human visual characteristics is arranged on a concentric circle, and the weighted value arranged in a specific angular direction from the center is emphasized. As described above, by using a weighted distribution that reflects human visual characteristics instead of a simple inclined arrangement or a normal distribution arrangement, it is possible to achieve dot dispersion with better graininess.

More preferably, the specific angle for emphasizing the weighting in the weighting filter is in the range of ± 14 ° in the vicinity of 0 ° and 90 °, and in the angular directions of 45 ° and 135 ° (second and third). For the fourth quadrant, the angle is extended on both sides of the central axis). By simply arranging the weights on the concentric circles, it is not possible to suppress the formation of dot arrangements aligned in the vertical and horizontal directions that are easily noticeable as a texture, but by emphasizing the weights in the angular directions near 0 ° and 90 °, It becomes difficult to select the dot arrangement that is lined up in the vertical and horizontal directions. This has the effect of suppressing the occurrence of banding due to interference with the main / sub scanning operation of the image forming engine. Also, when an error diffusion or Bayer type dither matrix dot pattern is used as the initial dot arrangement pattern, a texture aligned at 45 ° is likely to be generated due to the directionality of the processing, but the weighting in the 45 ° direction should be emphasized. This also has the effect of suppressing the occurrence of such texture.

Further, according to a preferred aspect, for example, the basic layout of a plurality of threshold value matrices corresponding to respective dot diameters is common, but the individual threshold values and the threshold value differences to be arranged are in the respective corresponding dot diameter types. 100% possible
It is determined by the difference in solid density and the threshold matrix size. In this way, the density width that can be represented by the same dot area ratio differs depending on the dot diameter, but by adjusting the threshold interval to be arranged according to the dot diameter, the density change per gradation level Even if different, it is possible to obtain an effect that it can be kept constant.

According to another aspect, when the threshold value arrangement order of the binary threshold value matrix is determined, a specific gradation section starting from the 0th level is replaced with an existing distributed dither pattern. By doing this,
It is possible to improve the graininess, particularly the graininess of a highlight portion where the dot arrangement is conspicuous, when the image recording resolution that causes irregularity of irregular dot arrangement such as error diffusion and deteriorates the graininess is rather low.

The threshold value matrix creating method of the present invention described above can be executed on a computer.
A program therefor (Claim 4) and a computer-readable recording medium (Claim 5) in which the program is recorded are also included in the present invention. Further, as described in claim 6, an image forming system having means for quantizing multi-valued image data using the threshold value matrix created by the method of the present invention is also included in the present invention. The image forming system may be composed of a computer such as a personal computer and an image forming apparatus such as a printer, or may be composed of an image forming apparatus such as a printer alone. Also included in the present invention.

[0026]

DETAILED DESCRIPTION OF THE INVENTION Embodiments of the present invention will be described below with reference to the accompanying drawings. FIG. 1 is a flowchart showing a simplified procedure for creating a threshold matrix according to the present invention. FIG. 2 shows an example of the threshold matrix created by this procedure.

In FIG. 1, steps S1 to S5 are processing blocks for creating the basic layout 1 shown in FIG. 2, and steps S6 and S7 are the threshold matrix 2 shown in FIG. 2 according to this common basic layout. It is a processing block for creating 3 and 4. Such processing can be executed by software using a computer having a general configuration, but a program therefor and various recording media recording the program are also included in the present invention. Note that, here, description will be made on the assumption that a halftone is expressed by turning on / off three types of dots having different diameters (small dot, medium dot, and large dot). However, the method of the present invention is basically similarly applicable to the creation of a threshold matrix for expressing a halftone by using two or more types of dots having different sizes, densities, or brightness, or both. is there. Hereinafter, the content of the procedure shown in FIG. 1 will be described with a specific example.

First, the processing blocks of steps S1 to S5 will be described. In this processing block, it is assumed that halftones are represented by turning on and off one type of dot, that is, binary halftones are represented.

In steps S1 and S2, a gradation section that can be expressed (for example, gradation level 0 to gradation level 255)
(256 gradations up to), a set of a gradation level on the starting point side and a gradation level on the ending point side is picked up. The gradation levels on the start point side and the end point side are not continuous, and a plurality of gradation levels must exist between them. Further, the gradation level on the starting point side is assumed to be a level on the highlight side of the gradation level on the ending point side. The pickup of this gradation level is performed from the highlight side to the shadow side. In the next step S3, the dot arrangement at the picked-up start point side and end point side gradation levels is determined (only one type of dot). The above processing contents are shown in FIGS.
It will be specifically described with reference to.

In steps S1 and S2, first, as shown in FIG. 3, it is assumed that the gradation level m is picked up as the starting point side gradation level and the gradation level n as the ending point side gradation level.

In step S3, first, an initial dot arrangement pattern for the tone level m on the starting point side is prepared, and the dot arrangement is evaluated by using an evaluation function, and the optimized dot arrangement is determined. .

As the initial dot arrangement pattern at this time, a dot pattern obtained by performing an error diffusion process on a solid image having a gradation level m or a dot pattern obtained by using an existing distributed dither can be used. Although it depends on the resolution, in the highlight area (the gradation section starting from gradation level 0 to a specific gradation level), the dot pattern processed by the existing distributed dither such as Bayer type is more granular. From the viewpoint of, it is preferable as the initial dot arrangement pattern. However, it is not preferable to use the white noise pattern. Addition of dispersiveness, the consistency when threshold matrices are stuck together, and the calculation load when making adjustments to maintain continuity between gradations are very large, and the degree of optimization is low. This is because it is easy to fall into a local solution.

Since the initial dot arrangement pattern also reflects the problem peculiar to the processing method of the creation source, it is necessary to perform the optimization processing. FIG. 7 shows an example of a dot pattern created by the error diffusion process, but a gap is formed in the upper part due to “sweep” which is a problem peculiar to the error diffusion. Vertical / horizontal alignment of dots as seen in a dot pattern by Bayer type dither is also easily recognized as a texture, and banding may occur due to interference with the main / sub scanning operation of the image forming apparatus. After all, optimization is necessary.

In step S3, next, a process of optimizing the dot arrangement according to the optimization condition is performed on the initial dot arrangement pattern on the starting point side. Specifically, as schematically shown in FIG. 6, the pattern matrix 11 in which the dots of the initial dot arrangement pattern are ON “1” and OFF “0”
On the other hand, the evaluation value is calculated by multiplying the optimization filter 12 having a predetermined size by using each ON dot as a target dot.
This is performed by rearranging the ON / OFF of the dots so that the total of the evaluation values 13 calculated for each target dot becomes the minimum. This method can achieve optimization with simple filtering compared to the Blue Noise method using Fourier transform.

In FIG. 6, for simplification, the optimization filter has the same size and size as the matrix of the initial dot arrangement pattern, but in reality, it is smaller than the initial dot arrangement pattern and has a circular filter as shown in FIG. Become. The filter size (applicable radius r) at this time is determined by the equation (1) and changes according to the number of dots that are turned on.

[Equation 1]

In the weighted distribution of the optimization filter, a value obtained by correcting the human visual characteristic (VTF) is arranged concentrically with the dot of interest as the center. Human visual characteristics (2)
Although it is approximated by the formula, there is a characteristic that the sensitivity sharply decreases at 1 cycle / mm or less (f <5.0 in the formula), and if reflected in the filter as it is, even if the dot arrangement in the highlight part There is a risk that the threshold value will be set so that is fixed in one place. Therefore, in the area of 1 cycle / mm or less,
Equation (3) is used, which is a regression calculation assuming that the sensitivity of VTF increases as it is. Since it is based on the visual characteristics, it becomes possible to form a dot pattern more suitable for human eyes than the conventional normal distribution weighting.

[Equation 2]

[Equation 3]

As described above, it is extremely undesirable for the dots to be aligned in the horizontal and vertical directions as a result of the optimization. In order to avoid such an unfavorable situation, the vertical and horizontal directions of 0 ° and 9 with respect to the target dot position
The weight placed near 0 ° is set to a larger value. Here, the purpose of describing the neighborhood is 0, as shown in FIG.
This is because even if dots are arranged at positions shifted by 1 or 2 dots from the 90 ° line, it can be considered that they are aligned in the vertical / horizontal direction depending on the size of the threshold matrix to be created. It is considered that at least a matrix size of 8 × 8 or more is required to design a distributed threshold matrix with irregularity (4 × 4 cannot be removed from Bayer type), and as shown in FIG. 2 from 0 °, 90 ° line
A viewing angle of dots, that is, a range of ± 14 ° is treated as a neighborhood. Further, from the experimental data, it was found that when the dots arranged in the oblique direction (45 °, 135 °) were mixed in the irregularly distributed pattern, this was noticeable as a texture. Therefore, in the creation of the initial pattern, weight amplification is performed not only in the vicinity of 0 ° and 90 ° but also in the direction of 45 °. In the above description, the first quadrant with the dot position of interest as the origin is dealt with, but it is natural that the same treatment is applied to the second, third, and fourth quadrants, and the angle is adjusted according to each quadrant. Is converted (0 °
→ 180 °, 90 ° → 270 °, etc.).

After optimizing the matrix on the start point side as described above, the dot arrangement on the end point side is determined. First, an initial dot arrangement pattern similar to that on the starting point side is prepared, and the optimization processing is performed after performing processing for matching with the optimized starting point side matrix. FIG. 10 is an explanatory diagram of the matching process.
First, the initial dot arrangement pattern 21 on the end point side and the optimized dot arrangement pattern on the start point side 20 are combined. In the dot arrangement pattern 22 after composition, by deleting the excess dots one by one using the above-described optimization method with respect to the dots other than the dots on the start point side, the dots on the end point side matched with the start point side are obtained. The initial dot arrangement pattern 23 is created. Finally, the optimized dot arrangement on the end point side is determined by paying attention to the dots other than the dots on the start point side in the dot arrangement pattern 23 after the matching process and performing the optimization by the optimization method described above.

The purpose of matching with the starting point side is as follows. It is possible to provide threshold matrices with different layouts for each gradation level, but continuity between gradation levels is lost, so pseudo contours and textures may occur, and the image forming apparatus is inexpensive. Then
It may be difficult in terms of cost to mount a large number of threshold matrices. Therefore, by always matching the start point side and the end point side, it becomes possible to store all gradations that can be finally expressed in one threshold matrix.

FIG. 4 shows an example of a mask of dot arrangement on the start point side and the end point side optimized by the procedure described above.

Although not clearly shown in FIG. 1, steps S1 to S1
S3 is repeated from highlight to shadow as many times as necessary. For example, if the density level m and the density level n are picked up for the first time, the density level n is changed from the previous end point side density level to the start side density level by a predetermined level in the second time. Pick up the level as the end point concentration level. Then, the already optimized dot arrangement of the starting point side density level is used as it is, the initial dot arrangement pattern of the end point side density level is subjected to matching processing with the starting point side dot arrangement, and then the optimization is performed to make the end point side. Determines the dot arrangement of.

When the dot arrangement at the discrete gradation levels is determined as described above, the dot arrangement at each gradation level between these gradation levels is determined by the interpolation method in step S4. This will be described with reference to FIG.
The next gradation level (m +
In 1), the candidate positions of the increased dots are narrowed down to any one of the dot arrangements of the gradation level n on the end point side. From among the dot candidate positions (dotted circles), the positions of the increased dots are selected using the optimization method described above. When the dot arrangement for the gradation level (m + 1) is determined in this way, the dot arrangement for the next gradation level (m + 2) is determined by the same method. (m + 1) is regarded as the starting point side gradation level. By the same procedure, the dot arrangement of all the gradation levels sandwiched between the picked-up gradation levels is determined.

Note that it is of course possible to adopt a processing procedure in which after the dot arrangement of the gradation levels on the start point side and the end point side of a set is determined, the dot arrangement of the gradation level between them is immediately determined. Is.

When the dot arrangements of all the gradation levels are determined in this way, the appearance order of the dots from highlight to shadow is determined, so in step S5,
A layout (threshold arrangement order) of a distributed threshold matrix for expressing a halftone is created by turning on / off one type of dot, that is, by using a binary value. This layout is used as the basic layout 1 shown in FIG. As described in relation to the initial dot arrangement pattern, depending on the resolution (especially at low resolution), the dither having regularity may be more preferable in the highlight part. It is also possible to replace the entire gradation section on the highlight side with existing dither.

Next, steps S6 and S7 will be described. In this processing block, a threshold value matrix for expressing a halftone is created by turning on / off two or more types of dots, that is, with three or more values. As described above, in this type of halftone expression method, there are three methods of controlling the dot diameter, controlling the ink density, and controlling both of them.
There are two types, but the description will be given here assuming that the dot diameter is controlled.

First, the gradation reproduction range for each quantization level is calculated (step S6). A threshold matrix for each quantization level is created by assigning each calculated gradation reproduction range to a common basic layout (step S7,
S8). The threshold value matrix corresponding to each dot diameter has the same basic layout, but the individual threshold values and the threshold difference to be arranged can be formed by the dots having the corresponding diameters.
It is determined by the difference in% solid density and the matrix size. By doing so, although the density width that can be represented by the same dot area differs depending on the dot diameter, the density change per gradation level can be kept constant for all dot diameters.

FIG. 2 shows an example of a threshold matrix for expressing a halftone with four values using large, medium and small dots and a basic layout commonly applied to them. Reference numeral 1 is a basic layout, 2 is a threshold matrix corresponding to small dots, 3 is a threshold matrix corresponding to medium dots, and 4 is a threshold matrix corresponding to large dots. Basic layout 1 shown here
Is a layout of a binary optimized distributed threshold matrix, each threshold matrix 2, 3, 4 is also a distributed threshold matrix.

FIG. 11 is a simplified diagram showing the basic configuration of an image forming apparatus which expresses a halftone with four values using the threshold matrices 2, 3, and 4. A quantizer 100 quantizes the multi-valued image data IN into four values by using three quantization thresholds Th1, th2, and Th3. 102, 103, 1
Reference numeral 04 is a threshold value generator that outputs the threshold values of the threshold value matrices 2, 3 and 4 corresponding to the position of each pixel of the multi-valued image data IN in the image space as the quantization threshold values Th1, th2 and Th3, respectively. These threshold generators 102, 103, 1
04 is an R storing threshold matrices 2, 3, and 4, for example.
It can be configured by a memory such as an OM and a counter for reading information from the memories according to the pixel clock of the multi-valued image data IN. Of course, the threshold generation circuits 102, 103 and 104 and the quantizer 100 can be realized as one ASIC circuit. The quantizer 100, when the pixel value of the multi-valued image data IN is less than Th1, dot off, when the pixel value is Th1 or more and less than Th2, small dot on, and the pixel value is Th2 or more and Th3.
If it is less than this, a quantized signal indicating medium dot on is output to the image forming engine 105, and if the pixel value is Th3 or more, a large dot on is output to the image forming engine 105. The image forming engine 105 controls the on / off of large, medium, and small dots in accordance with the quantized signal of the quantizer 100 to output an image in which halftones are expressed by four values.

With respect to the image forming apparatus capable of controlling the dot diameter very finely, sufficient gradation expression can be achieved only by area modulation without introducing dispersion characteristics. However, low-priced inkjet printers, electrophotographic printers, thermal transfer printers, etc., which are currently the mainstream in the market, can only perform dot diameter modulation in two to three steps, so that smooth gradation expression is realized. In this case, the introduction of dispersion characteristics is essential. Further, since there is a difference in each dot diameter that can be modulated, if the dots are not switched as shown in FIG. 12, the maximum diameter dot is suddenly recorded depending on the layout of the threshold matrix.
There is a risk that the merit of recording above the value will be lost (same level as binarization).

In the case of a binary value, it is necessary to fill the solid with a fixed dot diameter, so the dot diameter is usually set to √2 times the resolution pitch or more. As a result, in the actual image, partially overlapping dots occur before the halftone dot area ratio on the data reaches 50%. As in the case of digital data, if the level does not change from "1" to "1" even if they overlap, the overlapping area can be simply subtracted and reflected in the threshold value. However, in the case of the inkjet method, in reality, more ink adheres to the overlapping area to increase the density, and in the electrophotographic method and thermal transfer method, not only more recording energy (charge / heat) is concentrated in the overlapping area but also the surrounding area. Also propagates to, and unnecessary dot gain is generated. Even if the value is three or more, the same problem occurs with the maximum dot, but by switching the dots as shown in FIG. 12, such a problem can be avoided except for a very limited shadow part. it can. On the contrary, in the dots other than the maximum dot, even if the halftone dot area ratio is 100%, the overlapping portion does not occur, so the dot of each size can be expressed simply by using the area ratio or Yule-Nielsen equation. Since the number of gradations can be calculated, as described above, by arranging this number of gradations along the layout of the distributed threshold matrix, a distributed threshold matrix corresponding to each dot size can be created. You can

Depending on the dot size, the number of gradations that can be expressed is not necessarily the same for each size, but since the layout of threshold arrangement is the same, the dot arrangement for each gradation is the same for each dot size. Even if the patterns are different, there is no sense of discomfort during synthesis. Furthermore, although this is an advantage of reducing the number of values, since the number of gradations can be shared for each dot size and the grain size can be improved by reducing the dot diameter even in the same layout,
It is possible to reduce the size of the threshold matrix while maintaining the image quality, and it is possible to reduce the load for developing the threshold matrix and the cost for mounting the threshold matrix in an actual image forming apparatus.

The threshold value matrix created by the method of the present invention can be applied to all image recording methods for expressing an image by dots, but particularly, in inexpensive ink jet printers, electrophotographic printers, thermal transfer printers, etc.
It is suitable for a small-value image recording method in which the dot diameter and the dot density cannot be changed to the extent that it is multivalued.

An example of the image output system using the threshold matrix created by the method of the present invention will be described with reference to FIGS. 13 and 14.

In FIG. 13, reference numeral 200 denotes a printer driver executed by a personal computer, which is a preprocessing module 20.
1 and the gradation processing module 202 in which the threshold matrix created by the method of the present invention is mounted. The pre-processing module 201 is a part that performs pre-processing such as general CMM, BG / UCR, γ correction, and scaling in the printer driver. The gradation processing module 202 quantizes the preprocessed image data into three or more halftone data using a threshold matrix created by the method of the present invention. The halftone data is output as an image by the printer 203. Such a system configuration is generally suitable for an inexpensive printer.

The system configuration shown in FIG. 14 is generally suitable for a high-speed printer, and the printer driver 210 of the personal computer has a CMM, BG. Gradation in which preprocessing such as UCR and γ correction is performed, and the controller 211 on the printer side performs scaling processing and image data is quantized by using the threshold matrix created by the method of the present invention and converted into halftone data. Let the process run. The means for this quantization is a combinational circuit or ASIC having the configuration shown in FIG.
Although it can be a circuit, it is also possible to execute the same processing by using a microprocessor. This halftone data is output as an image by the image forming engine 212 of the printer. Such a printer is also included in the present invention. Although described as a printer here, it can be replaced with an image forming apparatus such as a digital copying machine or a facsimile.

[0056]

As described in detail above, according to the threshold value matrix creating method of the present invention, for example, dots of different types having different dot diameters have consistent dispersion characteristics, and dots of different types interfere with each other. As a result, it is possible to avoid the inconvenience of generating textures having different periods / tones. Since the arrangement order of dots of each type is the same, generation of unnecessary isolated dots can be prevented. By increasing the dispersion characteristic of the largest dot that most easily affects the graininess, the graininess of small and medium dots, which are less noticeable than that, is also improved, and the graininess of the entire image can be raised. It is possible to create a threshold matrix which has advantages such as avoiding the inconvenience that the quality of dot arrangement gradually deteriorates as the gradation level changes, which is seen in a general FM screen. Further, by using the computer program or the recording medium of the present invention, a threshold matrix having the above advantages can be easily created by using a computer such as a personal computer. According to the image output system of the present invention,
It is possible to obtain an effect that a high quality image can be output by using a printer or the like that expresses a halftone by using ON / OFF of two or more types of dots.

[Brief description of drawings]

FIG. 1 is a schematic flowchart for explaining an example of a procedure for creating a threshold matrix according to the present invention.

FIG. 2 is a diagram showing an example of a threshold matrix created according to the present invention and its basic layout.

FIG. 3 is an explanatory diagram of selection of gradation levels on a start point side and an end point side.

FIG. 4 is a diagram showing an example of optimized dot arrangement at tone levels on a start point side and an end point side.

FIG. 5 is an explanatory diagram of optimization of dot positions at gradation levels between the starting point side and the ending point side gradation levels.

FIG. 6 is an explanatory diagram of evaluation value calculation by an optimization filter.

FIG. 7 is a diagram illustrating a dot pattern in which “sweep” has occurred due to error diffusion processing.

FIG. 8 is an explanatory diagram of an application range of an optimization filter.

FIG. 9 is an explanatory diagram of a relationship between dot alignment and matrix size, and an angle to be corrected.

FIG. 10 is an explanatory diagram of a matching process performed before the optimization process of the dot arrangement of the end point side gradation level.

FIG. 11 is a block diagram showing the basic configuration of an image forming apparatus that expresses halftone with four values.

FIG. 12 is an explanatory diagram of a dot switching method when gradation expression is performed using large, medium, and small dots.

FIG. 13 is an explanatory diagram of a configuration example of an image output system according to the present invention.

FIG. 14 is an explanatory diagram of another configuration example of the image output system according to the present invention.

FIG. 15 is a diagram schematically showing a state in which the degree of optimization of dot patterns sequentially decreases in the sequential optimization method.

[Explanation of symbols]

S1 to S5 Processing steps for creating a basic layout of a binary threshold matrix S6, S7 Processing steps for creating a threshold matrix for quantization of three or more values according to a common basic layout Step 1 Basic layouts 2 and 3 , 4 Threshold matrix 100 Quantizers 102 to 104 Quantization threshold generator 105 Image forming engine 200 Printer driver 201 Pre-processing module 202 Gradation processing module 203 Printer 210 Printer driver 211 Printer controller 212 Printer image forming engine

   ─────────────────────────────────────────────────── ─── Continued front page    F-term (reference) 2C262 AA24 AB07 AB13 BB01 BB06                       BB10 BB22 BB27                 5B057 AA11 BA29 BA30 CA02 CA08                       CA12 CA16 CB02 CB07 CB12                       CB16 CE13 CH07 CH18                 5C077 LL19 NN08 PP52 PP53 PQ08                       RR02 RR09 RR11 RR19 TT02

Claims (6)

[Claims] 1. A method of creating a plurality of threshold value matrices for quantizing multi-valued image data into three or more values in order to express a halftone by turning on and off a plurality of kinds of dots, wherein A first step of deciding a threshold value arrangement order of a threshold value matrix for binarizing multi-valued image data to represent halftone by turning on and off, and a plurality of steps for quantizing multi-valued image data into three or more values. And a second step of creating each of the threshold matrices in accordance with the threshold arrangement order determined in the first step. 2. In the first step, a grayscale level that is discontinuous in all grayscale intervals is selected, the dot arrangement in the selected grayscale level is determined according to optimization conditions, and the selected start point side is set. The dot arrangement at each gradation level between the end point side gradation levels is determined based on the optimized dot arrangement of the start point side and end point side gradation levels according to the optimization conditions, and each determined gradation level 2. The threshold value matrix creating method according to claim 1, wherein the threshold value arrangement order of the threshold value matrix for binarization is determined based on the dot arrangement of levels. 3. In the first step, a matching process for including an optimized dot arrangement of a starting point side gradation level in an initial dot arrangement pattern of an ending point side gradation level is performed, and after this matching processing, Based on the initial dot placement pattern,
3. The threshold value matrix creating method according to claim 2, wherein the dot arrangement on the end point side gradation level is determined according to the optimization condition while preserving the dot arrangement on the start point side. 4. A program for causing a computer to execute a threshold matrix creating process by the threshold matrix creating method according to claim 1. Description: 5. A computer-readable recording medium on which the program according to claim 4 is recorded. 6. An image comprising means for quantizing multi-valued image data into three or more values by using a threshold value matrix created by the threshold value matrix creating method according to claim 1, 2. Forming system.