JP2001028685A - Half-tone matrix generating method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To generate a half tone matrix of an FM screen at a high speed even when the size of the matrix is large. SOLUTION: In the method for generating a half tone matrix used for binary processing or multi-values processing that converts multi-value image data into binary data or multi-value data, an entire area of the half tone matrix to be generated is divided into a plurality of division areas, the half tone matrix for each division area is generated individually, and the half tone matrices generated individually for each division area are combined to generate the entire half tone matrix.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to binarization processing or multi-level processing of an image, and more particularly to a halftone matrix creation method.

[0002]

2. Description of the Related Art Digital image data handled by a computer often has one pixel of 8 bits or more per color in order to express continuous tone. With 8-bit data, gradation expression of 2 to the eighth power = 256 gradations is possible. On the other hand, when outputting such digital image data, there are various types of output devices. For example, in a printer called a thermal sublimation type printer that sublimates ink by heat and transfers it to paper, continuous tone can be reproduced at each reproduction dot.
Some can reproduce the gradation. That is, the dot density is variable for each dot. If it is 8-bit image data, it can be printed as it is. On the other hand, there is a printer called an ink jet printer that performs printing by spraying ink droplets on paper. Many printers of this type reproduce dots of each color in binary. That is, in the case of monochrome for each dot, it can be expressed only in either white or black.
For such a printer, for example, 8 bits (256
To print using (gradation) data, it is necessary to convert 8-bit data into 2-bit data. here,
If the data is simply binarized, there arises a problem that the tone information originally contained in the image is lost. For this reason, it is necessary to perform the binarization so that the gradation is visually expressed in the reproduced image while performing the printing in the binary. That is, pseudo halftone processing is required.

Many methods are known for pseudo halftone processing. Among them, the relatively simple one is the systematic dither method. In the systematic dither method, data called a dither matrix is used. FIG. 7 is an explanatory diagram showing the concept of binarization by the conventional systematic dither method. 7A shows the original image data, FIG. 7B shows the dither matrix, and FIG. 7C shows the binarization result. Here, 8-bit data is assumed, and the data value is 0 to 255. First, a threshold value for binarization is arranged in the dither matrix. The input image as shown in FIG. 7A is sequentially associated with the dither matrix one pixel at a time as shown in FIG. 7B, and is binarized by a threshold value at the corresponding position. That is, if the value of the image data is smaller than the threshold, the binarization result is turned off, and if the value of the image data is larger than the threshold, the binarization result is turned on. As a result, a binarization result as shown in FIG. 7C is obtained.

Here, the dither matrix will be described. Here, a 4 × 4 matrix generally used is described as an example. FIG. 8 is an explanatory diagram showing threshold levels. In a 4 × 4 matrix, there are 17 different cases for the number of ON / OFF dots. Therefore, gradation expression of 17 levels becomes possible. When the input image data is 8 bits and can represent from 0 to 255, if the thresholds are equally set, the dynamic range of the image data is equally divided by the number of levels that can be represented by the dither matrix as shown in FIG. Take an intermediate value. Next, it is necessary to determine how to arrange these thresholds in a matrix. The dot arrangement is roughly divided into a dot concentration type and a dot dispersion type. 9 and 10 are explanatory diagrams showing the arrangement of dots. FIG. 9 shows how dots are sequentially turned on in the dot concentration type. In the dot concentration type, as the gradation level increases, the dots gradually increase in thickness around a certain point.
On the other hand, FIG. 10 shows a state in which dots are sequentially turned on in the dot dispersion type. In the dot dispersion type, dots are dispersed as a whole without concentrating as much as possible.
The so-called Bayer type is a typical example. By such processing, pseudo halftone processing can be performed.

[0005] However, such a pseudo halftone using a dither matrix has a drawback that a regular pattern is generated. That is, the arrangement of the dots for a specific density is fixed and determined, and is repeated successively, so that a density-specific regular pattern is generated. Therefore, a technique called an error diffusion method is known. The error diffusion method uses a fixed threshold value when binarizing the data.
This method uses an error with the binary data after the value conversion and diffuses the error to adjacent peripheral pixels. At this time, it is general to weight the error to be diffused according to the relative position from the target pixel. According to such a method, the error which has been conventionally discarded at the time of binarization can be used in the subsequent binarization processing, and the error can be minimized as a whole. In addition, as a result of such processing, continuous tone reproduction can be reproduced in a form in which dots are dispersed, instead of the limited number of tones as in the above-described systematic dither method, in reproducing continuous tone.

[0006] However, the above error diffusion method also has a disadvantage. First, the error is calculated and diffused while weighting the surrounding pixels, so that it is necessary to perform a product-sum operation for each pixel, and there is a problem that high-speed processing is difficult to perform. In addition, there is a problem that a regular stripe pattern is generated when a specific gradation, particularly a value obtained by dividing the whole gradation level by a simple integer or a value close thereto is generated. further,
In a portion where the level of the image changes rapidly, there is a problem that the generation of dots is delayed due to accumulation of errors, a pseudo contour is generated, and an unnatural impression is given to the appearance.

Therefore, a technique called an FM screen is receiving attention. This is the same as the systematic dither method in principle, but is characterized by using a relatively large dither matrix and creating the matrix so that the dots are arranged randomly and dispersed. The size of the dither matrix is 16 × 16 or 32 × 32
If the size is relatively small such as, for example, a repeated pattern is generated, so that the size is 128 × 128 or 256 × 25.
6 is often used.
By doing so, continuous tone can be reproduced continuously and smoothly, and the number of calculations can be reduced. If the matrix is made large to some extent and properly prepared, no unexpected regular stripe pattern will occur and no delay in dot generation will occur.

As described above, it is important to create a halftone matrix in an FM screen.
Several methods are known as a method for creating a halftone matrix of an FM screen. For example, there is a method using a Fourier transform. First, a white noise image having a target matrix size is created. That is, elements of the halftone matrix are regarded as pixels of an image, and the halftone matrix is handled as an image. Here, white noise is random noise whose frequency components are uniform. Next, Fourier transform is performed on this white noise image to convert it into a frequency component. Next, of these frequency components, low frequency components are cut. The noise from which the low frequency component has been cut is called blue noise. This data is subjected to inverse Fourier transform to return to image data. The returned image becomes an FM screen, and the image is used as a dither matrix to perform a binarization process.

Another method is to arrange the dots by calculation when the elements of the halftone matrix are similarly regarded as pixel dots of an image. Creating a halftone matrix corresponds to FIGS. 9 and 10
As can be seen from the above, this is equivalent to performing ranking for determining the order in which the corresponding elements are turned on for each dot in the halftone matrix. Therefore, taking into account the arrangement of the dots that have been ranked up to that point, the remaining dots are sequentially ranked by a specific calculation procedure, and repeated until all the dots in the halftone matrix are ranked. There is a method of creating a matrix. Here, as a specific procedure for determining the ranking, for example, from an unranked dot which is not ranked in the halftone matrix,
There is a method of calculating the distance to the already ranked dots, finding the minimum value of the distance, and performing the following ranking for the unranked dots having the maximum minimum value. This corresponds to setting the center of the largest circle in the space of the unordered dots as the dot to be ranked next. In this case, it is necessary to first rank some dots in advance, which is performed, for example, by ranking arbitrary points in advance.

[0010]

However, there is a problem that it takes an enormous amount of time if an attempt is made to create an FM screen by the above-described specific calculation procedure. That is, in the method of ranking dots by calculation as described above, when determining the next dot to rank, the evaluation parameters are calculated for all the unranked dots, and the evaluation parameter for each dot is calculated. , The most suitable dots are ranked. Therefore, when the dots in the middle stage are ranked, it is necessary to perform the calculation again for the remaining unranked dots. For example, if the size of the halftone matrix is n × n (= N), and the number of ranked dots is k, the following calculation is performed for each of the Nk unranked dots. It is necessary to calculate the distance to k dots, and the calculation amount is (N−k)
× k. Then, in order to obtain an estimate of the total amount of calculation, the total amount of calculation when k is changed from 1 to N-1 is obtained. Becomes Actually, the initial dot cannot be calculated from one dot, and must be started from three or more dots.
It can be seen that the power increases, that is, n raised to the sixth power. Therefore, there is a problem that the larger the size of the halftone matrix is, the more the amount of calculation is drastically increased and the calculation time is drastically increased.

[0011]

SUMMARY OF THE INVENTION In view of the above problems, the present invention relates to a halftoning matrix creation method used for binarization processing for converting image multivalued data into binary data or multivalued data or multivalued data. , The whole area of the halftone matrix to be created is divided into a plurality of divided areas,
A halftone matrix for each divided region is separately created, and the halftone matrix for each separately created divided region is combined to create an entire halftone matrix. The halftone matrix of the divided area sequentially arranges dots according to a specific procedure from the position of a small number of initial dots,
This is created by repeating until all the dots in the divided area are ordered.

[0012]

Embodiments of the present invention will be described below with reference to the drawings. FIG. 1 is an explanatory diagram of area division of a halftone matrix according to the present invention. Here, 256
A case where a halftone matrix of × 256 is used will be described. First, a 256 × 256 halftone matrix is equally divided into 16 64 × 64 regions. Next, dots are arranged in each region by calculation.

FIG. 2 is a flowchart showing an algorithm for calculating the dot arrangement. Here, as a procedure for determining the ranking, the distance from the unranked unranked dots of the halftone matrix to the already ranked dots is calculated, the minimum value of the distance is calculated, and the minimum value is calculated. For the largest unranked dot,
A case where the following ranking is performed will be described using an example. This corresponds to setting the center of the largest circle in the space of the unordered dots as the dot to be ranked next. FIG. 3 is an explanatory diagram showing an algorithm for ranking the dots. For each of the ranked dots indicated by the black dots, the unranked white dots shown in the figure are the points at which the minimum value of the distance to the ranked dots is the maximum, and this point is ranked next. Also, this point corresponds to the center of the largest circle that enters the space of the unranked dots.

In FIG. 2, first, in step 21, an initial dot pattern is generated, and the dots at these positions are ranked. Next, in step 22, the distance between each of the unranked dots and the ranked dot is calculated and stored in the memory. In step 23, the dot having the minimum distance value is determined. In step 24, the unranked dots obtained in step 23 are ranked. In step 25, it is determined whether the pixel assigned the threshold value in step 24 is the last one, that is, whether there is any unprocessed pixel.
If there is an unprocessed dot, the process returns to step 22 and repeats the process. If there are no unprocessed dots,
In step 26, the creation of the halftone matrix ends.

[0015] In the above halftone matrix creating method, the first step is to generate an initial dot pattern,
The ranking of the dots at these positions is performed. FIG. 4 is an explanatory diagram showing an example of the initial dot pattern. FIG. 4A shows an example in which the dot arrangement is regular, and FIG. 4B shows an example in which the dot arrangement is irregular. In the case where the initial dots are as shown in FIG. 4A, the arrangement of the threshold values sequentially determined thereafter is also regular, and tends to be close to a Bayer pattern. In addition, when the initial dots are as shown in FIG. 4B, the arrangement of the threshold values sequentially determined thereafter is likely to be random.

Thus, the halftone matrix is
The result depends on the initial dot pattern. FIG.
FIG. 8 is an explanatory diagram showing an example in which the initial dots are changed for each divided area of the halftone matrix according to the present invention. As shown in FIG. 5, when the entire halftone matrix is divided, the halftone matrix of each divided region can be finely changed by changing the initial dot pattern for each divided region.
In addition, repetition of a pattern can be avoided in a small area, and randomness can be maintained in the entire halftone matrix. Further, since the calculation is performed for each divided region, the amount of calculation is small, and the entire halftone matrix can be obtained at high speed.

On the other hand, there is a problem that it takes time and effort to decide to change the initial dot pattern for each divided area obtained by dividing the halftone matrix. . Therefore,
First, for the entire halftone matrix, ranking is performed halfway without performing region division, and after sufficient ranking dots are generated for each divided region, a method of calculating the ranking for each divided region is considered. Can be FIG. 6 is a flowchart showing an algorithm for calculating the dot arrangement according to the present invention. In FIG. 6, first, in step 61, the division of the area of the halftone matrix is determined. In step 62, an initial dot pattern is generated for the entire halftone matrix, and the dots at these positions are ranked. Next, in step 63, a ranking process is performed on each of the unranked dots. In step 64, it is determined whether or not initial dots are sufficiently generated for each of the divided areas divided in step 61. If not, the ordering process of step 63 is repeated. If sufficient, the halftone matrix ranking process is in the state shown in FIG. 5 at this time. Therefore, the calculation is performed by a calculation routine for determining the halftone matrix in step 65 in the same manner as described with reference to FIG. In step 66, the creation of the halftone matrix ends.

Although the above description has focused on the case of binarization processing, the present invention is also applicable to the case of multi-level processing of three or more values. Here, multi-value conversion means that m-value image data is converted into n-valued (m>
n). For example, if the dots of the output device are reproduced in three values, ternary multivalue processing is required. In the above description, a case of a 256 × 256 halftone matrix has been described, but other sizes may be used. Also, a case has been described in which the halftone matrix is divided equally, but the present invention is applicable even if the halftone matrix is not divided evenly. Also, a specific procedure for creating the halftone matrix is to calculate the distance from the unranked unordered dots of the halftone matrix to the already ranked dots, find the minimum value of the distance, and calculate the minimum value. Although the method of performing the following ranking for the largest unranked dot has been described as an example, the present invention can be implemented using other procedures.

[0019]

As described above, according to the present invention,
Even if the halftone matrix of the FM screen has a large size, it can be created at high speed, and a high-quality print can be obtained using the halftone matrix.

[Brief description of the drawings]

FIG. 1 is an explanatory diagram of area division of a halftone matrix according to the present invention.

FIG. 2 is a flowchart illustrating an algorithm for calculating the arrangement of dots.

FIG. 3 is an explanatory diagram showing an algorithm for ranking dots.

FIG. 4 is an explanatory diagram showing an example of an initial dot pattern.

FIG. 5 is an explanatory diagram showing an example in which an initial dot is changed for each divided region of a halftone matrix according to the present invention.

FIG. 6 is a flowchart illustrating an algorithm for calculating a dot arrangement according to the present invention.

FIG. 7 is an explanatory diagram showing the concept of binarization by a conventional systematic dither method.

FIG. 8 is an explanatory diagram showing threshold levels;

FIG. 9 is an explanatory diagram showing an arrangement of dots of a dot concentration type.

FIG. 10 is an explanatory diagram showing an arrangement of dots of a dot dispersion type.

[Explanation of symbols]

21: Generation of initial dot pattern 22: Calculation of distance 23: Judgment of dot with minimum distance value 24: Ranking of unranked dots 25: Final pixel Judgment 26 ... End of halftone matrix creation 61 ... Segmentation of halftone matrix area 62 ... Generation of initial dot pattern 63 ... Ordering process 64 ... Determine whether enough initial dots are generated 65: Routine for calculating halftone matrix 66: End of halftone matrix creation

 ──────────────────────────────────────────────────続 き Continued on the front page F term (reference) 5B057 CA08 CA12 CA16 CB07 CB12 CB16 CC02 CE13 5C077 LL18 MP01 NN09 NN19 NP01 PP21 PP43 PP68 PQ08 RR02 RR03 RR05 RR09 RR16

Claims (4)

[Claims] 1. A halftone matrix creation method used for binarization processing or multivalued processing for converting image multivalued data into binary data or multivalued data, wherein a whole area of a halftone matrix to be created is divided into a plurality of segments. The method is characterized in that a halftone matrix for each of the divided regions is separately created, and the halftone matrix for each of the separately created divided regions is combined to create an entire halftone matrix. Halftone matrix creation method. 2. A halftone matrix for each area is created by sequentially arranging dots in accordance with a specific procedure based on a small number of initial dots, and repeating until all the dots in the area are ordered. 2. The method of producing a halftone matrix according to claim 1, wherein: 3. The halftone matrix creating method according to claim 2, wherein the initial dots are created in different positions in each area. 4. A sequence of dots is sequentially assigned to all regions according to a specific procedure from the position of an initial dot, and after a sufficient number of dots are assigned to each of a plurality of divided regions to be divided, 3. The halftone matrix creating method according to claim 2, wherein in the divided areas, the dots that have been ordered so far are set as initial dots of each divided area, and the remaining dots are ordered for each divided area.