JP2002027249A - Half tone processor, and medium having stored its procedure - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a half tone processor which can get a high-quality half tone image high in reproducibility, and can finish the processing in a short time, and a medium having recorded its procedure. SOLUTION: These are a half tone processor consisting of a block dividing means, which divides a input image into blocks consisting of specified size of picture elements, an adaptation computing means which makes the picture element of each block a vector and compares this with a code word to serve as a candidate of conversion and computes the adaptation, an image generating means which generates an image in the block, based on the code word where the adaptation is maximum, an error diffusing means which diffuses the average stratum error of the block to the neighboring block of a block being processed thereafter, and a storage medium having recorded its procedure.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

The present invention belongs to the technical field of processing image data by a computer. In particular, the present invention relates to a halftone process for generating a halftone image (pseudo grayscale image) from a grayscale image.

[0002]

2. Description of the Related Art Halftone processing is a processing which, in a reproduction system having a limit on the number of gray scales that can be expressed in principle, allows expression closer to the original grayscale image (full-tone image) beyond the limit. . For example, in offset printing, in principle, two parts, ie, a halftone dot portion where the ink is deposited on the printing paper and a portion of the paper itself which is not deposited on the printing paper.
Reproduce the image with only one part. In the case of color printing, a combination of four color dots of CMYK (cyan, magenta, yellow, black) and the overprinting of the halftone dots is used.
Reproduce an image using only colors.

As is well known, in offset printing, halftone dot% is changed to exceed such a limit. Halftone dot% is the ratio of the area occupied by halftone dots on the printed surface. Since these halftone dots are fine, when viewed from a distance (approximately 20 cm or more) at which the fine structure cannot be seen, the halftone dots are recognized as intermediate colors due to the visual characteristics, and a full-color full-tone image is reproduced in a pseudo manner.

[0004]

Conventionally, the process of generating the halftone dots in offset printing includes a contact / printing process.
A screen is used. That is, a halftone dot film is obtained by superposing the contact screen and the film of the original image on the photosensitive film, and exposing the photosensitive film to exposure. Since the contact screen is a periodic image,
The dots in this dot image are also arranged periodically (FIG. 7).
(A)).

[0005] Overprinting these periodic halftone dots in color printing causes moiré. In order to avoid this, the screen angle is actually changed for each color. However, even if moiré can be avoided,
A glazing pattern (rosetta pattern) which is a visual obstacle and lowers the print quality remains and cannot be avoided at the same time (see FIG. 6).

[0006] It is recognized that the generation of the turtle pattern is caused by the fact that the halftone dots are periodic and the fine dots constituting the halftone dots are concentrated in the periodic part. Therefore, it has been proposed to make the halftone dots dispersed or aperiodic. This is because arbitrary dots can be generated by computer processing in recent years.

The basis of the processing is a comparison operation between an original image and a threshold matrix (dither matrix) of a predetermined size.
As this method, a fixed mask method, a systematic dither method, a random dither method, and the like are known, but there is a problem that the periodicity cannot be excluded because the threshold matrix is regularly arranged in the entire image ( FIG. 7 (B)).

In order to reduce the recognition of the periodicity,
Larger block size, application of different threshold matrices,
Although application of a dither value, a random number, addition processing of blue noise, and the like to the original image has been proposed, periodicity is also recognized (see FIG. 8D).

As a method not based on a comparison operation with a threshold matrix, there are known an average error minimum method, an error diffusion (ED) method, and the like. In addition to the advantage that there is no periodicity, the preservation of the average gradation and the reproducibility of details are excellent because the main purpose is to reduce the quantization noise. However, there is a problem that the graininess remains, the gradation characteristic changes, and a unique worm-like texture is generated (see FIG. 8C). In addition, there is a problem that the load of data processing is large due to an increase in the amount of calculation and a corresponding processing time is required.

The present invention has been made to solve the above problems. An object of the present invention is to provide a halftone processing apparatus which can obtain a high-quality halftone image with high reproducibility and can perform the processing in a short time, and a medium in which the processing procedure is recorded. .

[0011]

The above object is achieved by the present invention described below. That is, the halftone processing apparatus according to claim 1 of the present invention converts an N-bit pixel value with M> N from an image having an M-bit pixel value, where L, M, and N are positive integers. A halftone processing device for deriving an image having the image, comprising: a block dividing unit; a fitness calculating unit; an image generating unit; and an error diffusion unit.
, And the degree-of-fit calculation means converts the pixels of the divided block into an L-dimensional M-bit vector, compares the vector with an L-dimensional N-bit code word that is a conversion candidate, and The image generating means generates an L-dimensional N-bit image in the block based on the codeword having the highest fitness, and the error diffusion means processes the average gradation error of the block thereafter. The error diffusion is performed to a block adjacent to the block.

According to the present invention, the input image is divided into blocks of pixels of a predetermined size L by the block dividing means, and the pixels of the blocks are converted into L-dimensional M-bit vectors by the fitness calculating means. Candidate L
The degree of conformity is calculated by comparing with the code word of dimension N bits, an image of L dimension N bits is generated in the block based on the code word having the maximum degree of conformity by the image generation means, and the average of the block is averaged by the error diffusion means. The tone error is diffused to a block adjacent to the block to be processed thereafter. That is, since the image generated in the block is based on the codeword having the highest degree of matching, an optimized image can be obtained from the entire halftone image. Further, since the error diffusion is performed, the image has excellent gradation reproducibility. Therefore, there is provided a halftone processing device capable of obtaining a high-quality halftone image with high reproducibility.

According to a second aspect of the present invention, in the halftone processing apparatus according to the first aspect, the error diffusion performed by the error diffusion means calculates an average gradation error of the block. The process is executed by canceling the error in the process. According to the present invention, since the error is canceled in the process of calculating the average gradation error, the processing load in error diffusion is extremely small.

The halftone processing apparatus according to a third aspect of the present invention is the halftone processing apparatus according to the first or second aspect, wherein J and K are positive integers, and the block dividing means converts the input image into a predetermined image. It is divided into rectangular blocks composed of pixels of size J × K = L. According to the present invention, processing is performed by rectangular blocks, and the basic and simplest processing is performed, and the reproducibility of horizontal lines and vertical lines is excellent.

According to a fourth aspect of the present invention, in the halftone processing apparatus according to any one of the first to third aspects, the block dividing means converts the input image into pixels of a predetermined size L. , Hexagons or triangles, or tilings that combine them (tilin
g) is divided into blocks. According to the present invention, the recognition of the periodicity is eased, and the reproducibility in the directions is equalized.

According to a fifth aspect of the present invention, there is provided a halftone processing apparatus according to any one of the first to fourth aspects, wherein an optimization parameter for setting an optimization parameter for calculating a degree of conformity is set. This has a setting means. According to the present invention, since the calculation of the degree of conformity is optimized by setting the optimization parameter, a higher-quality halftone image with higher reproducibility can be obtained.

According to a sixth aspect of the present invention, there is provided the halftone processing apparatus according to any one of the first to fifth aspects, wherein the codebook in which an L-dimensional N-bit code word generated in advance is described. Is stored in the storage unit.
According to the present invention, a codeword from a previously generated codebook is applied. Therefore, in the halftone processing, the processing time is reduced by the time of the codeword operation. Also, since the number of codewords can be appropriately reduced in advance, the processing time is further reduced.

A halftone processing apparatus according to a seventh aspect of the present invention is the halftone processing apparatus according to any one of the first to fifth aspects, wherein the codeword sequential generation means for sequentially generating an L-dimensional N-bit codeword. Is provided. According to the present invention, a codeword suitable for the state of the halftone processing can be generated, and processing relating to unnecessary codewords is omitted.

According to an eighth aspect of the present invention, in the halftone processing apparatus according to any one of the first to seventh aspects, the total number of the L-dimensional N-bit codewords is 2 ＾ L. (2 to the power of L). According to the present invention, the speed of the halftone process is increased.

According to a ninth aspect of the present invention, there is provided the halftone processing apparatus according to any one of the first to eighth aspects, wherein the halftone processing apparatus includes a plurality of processing units, The block divided by the dividing means is assigned to each of the plurality of processing units, and the plurality of processing units perform parallel distributed processing. According to the present invention, the speed of the halftone process is increased.

According to a tenth aspect of the present invention, in the halftone processing apparatus according to the ninth aspect, the calculation of the fitness performed by the fitness calculating means is performed in parallel by the plurality of processing units. This is to be processed. According to the present invention, the speed of the halftone process is increased.

In the halftone processing apparatus according to claim 11 of the present invention, in the halftone processing apparatus according to any one of claims 1 to 10, the calculation of the fitness by the fitness calculating means includes the step of: Instead of performing a full search, a codeword with the highest fitness is searched by an optimization technique. According to the present invention, the halftone processing is remarkably speeded up.

The halftone processing apparatus according to claim 12 of the present invention is the halftone processing apparatus according to any one of claims 1 to 11, wherein P is an integer of 2 or more, and the image is a P color image. , The pixels of the block are (L ×
P) It is processed as a vector of dimension M bits. According to the present invention, halftone processing is performed on a multicolor image.

According to a thirteenth aspect of the present invention, in the halftone processing apparatus according to any one of the first to eleventh aspects, the image is a multicolor image, and each color of the multicolor image is The above processing is performed for each image. According to the present invention, the speed of halftone processing for a multicolor image is increased.

According to a fourteenth aspect of the present invention, in the halftone processing apparatus according to the thirteenth aspect, the processing for each color image is performed by the plurality of processing units in parallel distributed processing for each color image. It is something to do. According to the present invention, halftone processing for a multicolor image is remarkably speeded up.

The medium on which the halftone processing procedure according to claim 15 of the present invention has been recorded can be obtained from an image having pixel values of M bits where L, M, and N are positive integers.
A medium in which a halftone processing procedure for deriving an image having a pixel value of N bits represented by N is recorded, wherein the halftone processing procedure includes a block division step, a fitness calculation step, an image generation step, In the block dividing step, the input image is divided into blocks each having a pixel of a predetermined size L, and in the fitness calculating step, the pixels of the divided block are defined as L-dimensional M-bit vectors, An L-dimensional N-bit image is generated by comparing with an L-dimensional N-bit code word that is a conversion candidate, and in the image reproducing process, an L-dimensional N-bit image is generated in the block based on the code word having the highest degree of conformity. In the error diffusion process, the average gradation error of the block is error-diffused to a block adjacent to the block to be processed thereafter.

According to the present invention, the input image is divided into blocks each having a predetermined size L in the block division process, and the pixels of the block are converted into L-dimensional M-bit vectors in the fitness calculation step. Comparability is calculated by comparing with a candidate L-dimensional N-bit codeword, and an L-dimensional N-bit image is generated in the block based on the codeword having the highest conformity in the image generation means. That is, since the image generated in the block is based on the codeword having the highest degree of matching, an optimized image can be obtained from the entire halftone image. Further, since the error diffusion is performed, the image has excellent gradation reproducibility. Therefore, high reproducibility,
A medium in which a processing procedure of a halftone process for obtaining a high-quality halftone image is provided.

[0028]

Next, an embodiment of the present invention will be described. The present invention is based on simple error diffusion. Further, the present invention extends the basic techniques such as vector quantization and optimization by adding error diffusion. Therefore, it is fundamentally different from the conventional method based on threshold processing or the like.

First, the error diffusion technique will be briefly described. For example, assume that a halftone image is obtained by binarization with a block size of 4 pixels × 4 pixels. In this case, simply, the average gradation is expressed in 1 + 16 = 17 steps. As a method of converting a full-tone original image of a standard 8 to 12-bit gradation into a halftone image, the gradation expression ability is insufficient. With the error diffusion technique, this lack of gradation expression ability can be complemented.

FIG. 3 is a diagram for explaining the error diffusion process. If the original image has 8-bit gradation, the pixel value is 0 to
It has a value of 255. If the halftone image has a 1-bit gradation, the pixel value has a value of 0 or 255 (or 0 or 1). In FIG. 3, the pixel value input to the entrance of the quantizer is converted to 0 or 255 by the quantizer and output to the exit of the quantizer. For example, conversion is performed in which 127 or less is set to 0 and 128 or more is set to 255.
The converted pixel value is output from the outlet of the quantizer to the output side. This converted pixel value is the pixel value of the halftone image.

By the quantization in this quantizer, 0 to 0
Since the value of 255 is rounded to 0 or 255, an error (quantization error) occurs between the original value and the value after quantization. As shown in FIG. 3, the pixel value input to the entrance of the quantizer is subtracted by the pixel value (converted pixel value) output to the exit of the quantizer. That is, a quantization error is calculated. This quantization error is input to the entrance of the filter.

The filter plays a role in determining how to spread the error. The simplest spreading method is a method of performing a simple delay and adding the result to the next pixel. Otherwise, since scanning of the original image is mostly raster scanning performed from the upper left pixel to the lower right pixel, in this case, a plurality of pixels in the right direction and the lower direction in the pixel on which the quantization error has been calculated are used. There is also a way to distribute.

In FIG. 3, an original image is scanned from the input side and pixel values are sequentially input. A quantization error is added to the pixel value. The quantizer inputs the pixel value to which the quantization error has been added and quantizes it. The quantized pixel value is output to the exit of the quantizer. This converted pixel value is the pixel value of the halftone image. Through this series of processes,
Error diffusion is performed. As a result, it is possible to obtain a halftone image that is excellent in reproducibility of relatively fine details and reproducibility of average gradation.

In the above, error diffusion in scalar quantization in which quantization is performed for each pixel represented by one numerical value has been described. The error diffusion in the vector quantization is basically the same. The description of applying error diffusion in the process of image vector quantization will be described later, and here, the vector quantization will be briefly described. In general quantization, that is, scalar quantization, a continuous or discontinuous scalar value (one-dimensional value) is rounded to a discrete scalar value (quantized representative value). For example, the operation of rounding off the decimal part to obtain an integer value is scalar quantization. The discrete scalar values need not be equally spaced and need not be integer values.

On the other hand, vector quantization is developed into vector values (multidimensional values). Vector quantization is
The space spanned by the vectors is divided into Voronoi polygons (multi-dimensional polygons) with the quantized representative value as the generating point, and all vector values included in the polygon are quantized as the generating points of the polygon. Rounded to a representative value.

FIG. 4 shows an example of rounding in a scalar value and an example of quantization in a simplest two-dimensional vector value. FIG. 4A shows a case of a scalar value.
(B) shows the case of a vector value. In FIG. 4, black points (black circles) represent generating points of the Voronoi polygon, that is, quantization representative values. The quantization is a process of rounding a scalar value or a vector value in an area surrounded by each boundary line to a quantization representative value represented by a black point.

In FIG. 4, each boundary line is a vertical bisector of a line connecting adjacent black points and is a set of equidistant points.
A boundary line equivalent to that in FIG. 4 exists in any dimension of the Euclidean space. When the quantized representative value is a vector, the quantized representative value is also referred to as a quantized representative value vector.
Among the many quantized representative value vectors, the quantized representative value vector in which a specific vector value is rounded is the nearest quantized representative value vector of the vector value, as is clear from FIG.

That is, a specific vector value is rounded to a quantized representative value vector that minimizes the Euclidean distance. Numerically, the quantized representative value vector is selected such that the sum of squares of the difference between the corresponding elements in the two vectors is minimized. Here, the quantized representative value vector is referred to as a codeword, and a set thereof is referred to as a codebook.

In the present invention, it is assumed that the original image is an image having 256-level (8-bit) gradation and the halftone image to be obtained is a binary (1-bit) image.
At this time, if the vector is two-dimensional (in units of two pixels),
Quantized representative value vectors are (0,0), (0,255),
(255, 0) and (255, 255). If the vector is four-dimensional (in units of four pixels), there are 16 types. FIG. 5 shows a quantized representative value vector when the vector is four-dimensional. In FIG. 5, the white rectangle is “0”,
Assuming that the black rectangle is “255”, the average gradation is 0, 64, 64, 128, 64, 128, 128,
191,64,128,128,191,128,19
1,191,255 (rounded off), and each gradation is 1, 4, 6, 4, 1 in the order of gradation (to be a binomial coefficient).

Next, the halftone process in the present invention will be described with reference to an example. M = 8 bits (256) Image having pixel values of gradation expression (original image)
, A halftone process for deriving an image (halftone image) having pixel values of N = 1 bit (2) gradation expression will be described. As a matter of course, specific numerical values in the processing to be described below are merely examples, and in specific embodiments of the present invention, numerical values different from the description are selected to suit individual cases. can do. The image may be a multi-color image, or the image size and block size, the number of tones of the original image, and the number of tones of the halftone image may be different.

FIG. 1 shows an example of the processing steps in the halftone processing apparatus of the present invention. First, step S1 in FIG.
In, the original image is input to the storage unit of the halftone processing device. The original image is 1024 × 1024 = 1048576
It is assumed that the image is a grayscale image of pixels. Next, step S2
In, the size of a predetermined block is designated. As an example, a block size of 4 pixels × 4 pixels is specified.

Next, in step S3, the original image is
It is divided into blocks of pixels × 4 pixels. That is, J = K =
4, L = 4 × 4 = 16. Original image is 1024 × 10
Since it is a grayscale image of 24 pixels, 256 × 256 = 65
It is divided into 536 blocks. This block can be represented as a 16-dimensional vector. Its 6553
Processing will be performed on six 16-dimensional vectors. FIG. 2 is an explanatory diagram of the division of the original image into blocks and the relationship between 16-dimensional vectors. FIG. 2A is an original image, and FIG. 2B is a block.

Note that, for simplicity, the arrangement order of the elements (pixel values) of the 16-dimensional vector is raster-scanned from the upper left to the lower right of the square block, and the processing order of the vector (block) is also from the upper left to the right of the image. Raster scan below.

Next, in step S4, the arithmetic average of the element is determined for each vector. That is, the average gradation of each vector is obtained. The arithmetic average is generally a non-integer, but can be rounded to an integer value for simplicity. In that case, the number of gradations in the average gradation of each vector matches the expression of M = 8 bits (256) gradation of the original image. Of course, the processing time is long but the processing can be performed without rounding.

Next, in step S5, an evaluation function for determining the degree of conformity is set. In the above-described vector quantization, a codeword (quantized representative value vector) that minimizes the Euclidean distance is determined to be most suitable, and that codeword is selected. Here, since the purpose is to derive a halftone image, the factor of the average gradation obtained in step S4 is taken into consideration.

For example, an evaluation function is set as in the following equation 1.

F (Gi, Ho) = α × | ΔD | + β × ΔQ / L where F (Gi, Ho); evaluation function Gi; vector of original image (block of original image) Ho; codeword α, β; coefficient (α + β = 1, 0 ≦ α, β ≦ 1) | ΔD |; average gradation error (| x | represents an absolute value of x) ΔQ; vector quantization error L; number of dimensions (16) )

Next, in step S6, the degree of conformity is calculated. The fitness is given by the evaluation function shown in Equation 1. In the evaluation function shown in Expression 1, the smaller the value of the evaluation function is, the higher the fitness is. Therefore, calculating the degree of matching means searching for Hj that minimizes the evaluation function for each Gi. Codewords are sequentially generated (2 16 powers), and a search for minimizing the evaluation function F shown in Expression 1 is fully performed.

Here, the subscript o is 0 ≦ o <2 ＾ 16 (2
16). That is, the total number of codewords. The subscript i is a number satisfying 0 ≦ i <65536. That is, it is the ordinal number of a 16-dimensional vector (block) in the original image. The subscript j is 0 ≦ j
<16 = L. That is, it is a number representing the order of each element of the vector.

Next, in step S7, an L-dimensional N-bit image is generated in the corresponding block based on the codeword having the highest degree of matching. By processing a specific 16-dimensional vector (block) in the original image in this manner, an L-dimensional N-bit image,
That is, that block portion of the halftone image is generated.

Next, in step S8, the difference between the average gray level of the code word having the highest fitness and the average gray level of the 16-dimensional vector (block) in the original image, that is, the average gray level error, is calculated. Error diffusion is performed on blocks to be processed thereafter. Here, error diffusion is error diffusion in vector quantization.

Here, application of error diffusion in the process of image vector quantization will be described. In the vector quantization, as described above, there is a difference between the average tone in a specific block portion in the original image and the average tone in the block portion in the halftone image, that is, an error p. This error diffusion can be performed, for example, by adding an error p to all pixel values constituting a block to be processed thereafter. Alternatively, the calculation can be performed by correcting the average tone error with the error p.

More specifically, the correction for calculating the average tone error is performed by correcting the value of | ΔD |, which is the average tone error in the evaluation function shown in Expression 1, by the error p. it can. The evaluation function when error diffusion is applied in the process of vector quantization of this image is:
For example, Equation 2 below is obtained.

F (Gi, Ho) = α × | ΔD−p | + β × ΔQ / L where F (Gi, Ho); evaluation function Gi; vector of original image (block of original image) Ho; codeword α, β; coefficient (α + β = 1, 0 ≦ α, β ≦ 1) | ΔD |; average grayscale error (| x | represents the absolute value of x) p; average grayscale error in the previous block ΔQ: error of vector quantization L: number of dimensions (16)

Next, in step S8, it is determined whether or not the processing in steps S6 and S7 has been performed on all blocks. If the processing has not been completed for all the blocks, the process returns to step S6 and the subsequent steps are repeated. If the processing has been completed for all blocks, the process proceeds to step S9. Next, in step S9, the generated halftone image is stored.

An example of the halftone process according to the present invention has been described above with reference to FIG. Next, a modified example thereof will be described. In the evaluation functions shown in Expressions 1 and 2, the previous terms | ΔD | and | ΔD-p | represent errors in average gradation, and thus represent rough reproducibility of shading. The later term ΔQ / L is RMSE (root mean square err)
or) indicates the reproducibility of the microstructure. Therefore, when α is increased, the reproduction of the fine structure is sacrificed at the expense of good reproduction of the density. Conversely, when β is increased, the fine structure is not reproduced well, but the reproduction of light and shade is sacrificed. Therefore, to optimize the calculation of fitness,
It is preferable that α and β as parameters (auxiliary variables) for giving weight by instruction input can be set to arbitrary values. In order to obtain a good overall reproduction, for example, α = β = 0.
5 is assumed.

Note that the evaluation function in the present invention is not limited to the evaluation function shown in Expression 1. In the evaluation function shown in Expression 1, the average gradation is considered, but dot gain and dot continuity can be taken into consideration.

In the block division in step S3, the original image is divided into blocks of 8 pixels × 8 pixels. That is, the shape of the block is a square. The block shape in the present invention is not limited to a square. It may be a J × K rectangle. In addition, a (pseudo) hexagon or (pseudo) triangle may be used as long as the plane can be essentially filled. Further, tiling which combines them alternately may be used.

Further, in the above description of step S6, it has been described that a full search is performed as an example. However, the calculation of the fitness in the present invention is not limited to the full search. For example, genetic algorithm (GA)
And other optimization techniques can be applied.

In order to reduce the processing time of the full search, a comparison with a plurality of codewords is performed by a number of processing units (C
(PU; central processor unit, etc.) to increase the processing speed. Further, the speed can be increased by setting a partial image in the original image, allocating a block to each of the plurality of processing units for each of the partial images, and performing parallel distributed processing using the plurality of processing units. In addition, in the respective terms of the evaluation function, that is, in the above-described equation 1, the α × | ΔD | and β × ΔQ / L terms are speeded up by performing parallel distributed processing using different processing units. be able to.

Further, the processing can be speeded up by terminating the operation of the evaluation function conditionally on the way. For example, when α × │ΔD│ is found to be larger than the minimum value obtained so far, it cannot be the minimum value, and the next evaluation may be performed.

This operation can be simplified if the average gradation values of the code words are calculated in advance and arranged in the order. Further, in the above description of step S6, a description has been given of the method of sequentially generating codewords. However, a codeword is generated in advance and stored as a codebook together with its average gradation value. It is preferable to perform the processing by referring to the processing because the processing speed is increased.

In the example shown in FIG. 1, description has been made as if the original image was a monochromatic image. When a multicolor image is handled in the same manner, the vectors may be connected in the pixel arrangement order or the multicolor plane (layer) order. For example, when P is an integer of 2 or more and the original image is a P-color image, the pixels of the block are processed as (16 × P) -dimensional vectors.

Of course, the above-described processing can be performed on each of the color images in the multi-color image instead of connecting the vectors. In such a case, the processing can be speeded up by performing the parallel processing on each of the color images by a plurality of processing units.

The present invention has been described with reference to the embodiment. Here, the processing in the halftone processing apparatus has been mainly described. However, the present invention is not limited to a halftone processing device. Of course, a medium on which the processing procedure is recorded is also included.

[0064]

As described above, the halftone processing apparatus according to the first aspect of the present invention provides a halftone processing apparatus capable of obtaining a high-quality halftone image with high reproducibility. Is done. Claim 2 of the present invention
According to the halftone processing device according to (1), the error is canceled in the process of calculating the average gradation error, and the processing load in error diffusion can be extremely reduced. Further, according to the halftone processing apparatus according to the third aspect of the present invention, the processing is basic and simplest, and the reproducibility of horizontal lines and vertical lines is excellent. Further, according to the halftone processing device of the fourth aspect of the present invention, recognition of periodicity is eased, and reproducibility in directions is equalized. Further, according to the halftone processing device according to claim 5 of the present invention, the reproducibility is higher,
A higher quality halftone image can be obtained. Further, according to the halftone processing device according to claim 6 of the present invention, the processing time involved in the codeword operation can be reduced,
The processing time can be further reduced by appropriately reducing the number of codewords. Further, according to the halftone processing device of the seventh aspect of the present invention, it is possible to omit processing relating to unnecessary codewords depending on the situation. Further, according to the halftone processing device of the present invention, the speed of the halftone processing can be increased. Claim 9 of the present invention
According to the halftone processing device according to the above, the speed of the halftone process can be increased. Claim 10 of the present invention
According to the halftone processing device according to the first aspect, the halftone processing can be remarkably speeded up. Further, according to the halftone processing device of the present invention, the speed of the halftone processing can be increased. Further, according to the halftone processing apparatus according to the twelfth aspect of the present invention, it is possible to perform halftone processing for a multicolor image.
Further, according to the halftone processing device according to the thirteenth aspect of the present invention, it is possible to speed up halftone processing for a multicolor image. Further, according to the halftone processing apparatus according to the fourteenth aspect of the present invention, it is possible to remarkably speed up halftone processing for a multicolor image. Further, according to the medium in which the halftone processing procedure according to claim 15 of the present invention is recorded, a medium in which the processing procedure of the halftone processing capable of obtaining a high-quality halftone image with high reproducibility is provided. Is done.

[Brief description of the drawings]

FIG. 1 is a diagram showing an example of a processing process in a halftone processing device of the present invention.

FIG. 2 is an explanatory diagram of the division of an original image into blocks and the relationship between 16-dimensional vectors.

FIG. 3 is an explanatory diagram relating to an error diffusion process.

FIG. 4 is a diagram illustrating an example of rounding in a scalar value and an example of quantization in a simplest two-dimensional vector value.

FIG. 5 is a diagram showing a quantized representative value vector when the vector is four-dimensional.

FIG. 6 is a diagram showing a turtle pattern (rosetta pattern) generated when a periodic halftone dot is overprinted.

FIG. 7 is a diagram showing an example of a halftone image (A is a contact screen, B is an image based on a periodically dispersed halftone dot).

FIG. 8 shows an example of a halftone image (C is an error diffusion method,
FIG. 3 is a diagram showing an image by a blue noise mask method.

Claims (15)

[Claims] 1. A halftone processing device which derives an image having pixel values of N bits where M> N from an image having pixel values of M bits, where L, M and N are positive integers. A block dividing unit, a fitness calculating unit,
An image generating unit; and an error diffusion unit. The block dividing unit divides the input image into blocks each having a predetermined size L. The fitness calculating unit divides the pixels of the divided blocks into L. A vector of M dimensions is obtained, and this is compared with an L-dimensional N-bit code word which is a conversion candidate to calculate the degree of conformity. The image generating means assigns L-dimensional N to the block based on the code word having the maximum degree of conformity. A halftone processing apparatus, wherein a bit image is generated, and the error diffusion unit performs error diffusion on an average gradation error of the block to a block adjacent to the block to be processed thereafter. 2. A halftone processing apparatus according to claim 1, wherein said error diffusion performed by said error diffusion means is executed by canceling the average gradation error in the process of calculating the error. Halftone processing device. 3. The halftone processing apparatus according to claim 1, wherein J and K are positive integers, and said block dividing means divides the input image into rectangles each having a predetermined size of J × K = L pixels. A halftone processing device, wherein the halftone processing device is divided into blocks. 4. The halftone processing device according to claim 1, wherein said block dividing means converts the input image into hexagons or triangles consisting of pixels of a predetermined size L, or a combination of these. Ring (tilin
g) A halftone processing device, which is divided into shaped blocks. 5. The halftone processing device according to claim 4, further comprising an optimization parameter setting means for setting an optimization parameter for calculating a fitness. Processing equipment. 6. A halftone processing apparatus according to claim 1, further comprising a codebook storing means for storing a codebook in which a previously generated L-dimensional N-bit codeword is described. And a halftone processing device. 7. The halftone processing apparatus according to claim 1, further comprising a codeword sequential generation means for sequentially generating an L-dimensional N-bit codeword. 8. The halftone processing apparatus according to claim 1, wherein the total number of said L-dimensional N-bit code words is less than 2 ＾ L (2 to the power of L). And a halftone processing device. 9. The halftone processing device according to claim 1, wherein said halftone processing device includes a plurality of processing units, and the block divided by said block dividing means is divided into said plurality of processing units. Wherein the plurality of processing units perform parallel distributed processing. 10. The halftone processing apparatus according to claim 9, wherein the calculation of the fitness performed by the fitness calculating means is performed in parallel by the plurality of processing units. 11. The halftone processing apparatus according to claim 1, wherein the calculation of the fitness by the fitness calculating means does not perform a full search of the codeword, but calculates the maximum fitness. A halftone processing device for searching for a codeword by an optimization technique. 12. The halftone processing apparatus according to claim 1, wherein P is an integer of 2 or more, the image is a P-color image, and pixels of the block are (L
× P) A halftone processing device for processing as a vector of dimension M bits. 13. The halftone processing apparatus according to claim 1, wherein said image is a multicolor image, and said processing is performed for each color image of said multicolor image. Halftone processing device. 14. The halftone processing apparatus according to claim 13, wherein the processing for each color image is performed by the plurality of processing units in parallel and distributed processing for each color image. 15. A halftone processing procedure for deriving an image having a pixel value of an N-bit expression where M> N from an image having a pixel value of an M-bit value, where L, M, and N are positive integers, is recorded. The halftone processing procedure includes a block division process, a fitness calculation process, an image generation process, and an error diffusion process process. In the block division process, the input image is converted to a predetermined size L. In the fitness calculation step, the pixels of the divided block are defined as an L-dimensional M-bit vector, which is compared with an L-dimensional N-bit code word which is a conversion candidate. In the image reproduction process, an L-dimensional N-bit image is generated in the block based on the codeword having the highest fitness, and in the error diffusion process, A medium in which a halftone processing procedure is recorded, wherein error diffusion of an average gradation error of the block to a block adjacent to the block to be processed thereafter is performed.