JPH03157060A - Intermediate tone estimate system for dither picture - Google Patents

Abstract

PURPOSE:To prevent blur of an estimated picture by preparing plural kinds of scanning openings and selecting an optimum scanning opening so as to apply intermediate tone estimation. CONSTITUTION:In the case of estimating an intermediate tone, a picture element input means 111 fetches an input picture element corresponding to the size of a dither matrix and the location of the fetched input picture element is converted into a normalizing means 121. After the conversion of an input picture element, an intermediate tone estimate means 131 selects a scanning opening corresponding to a binary picture, and the intermediate tone is estimated in response to the ratio of the black picture element in a binary picture of the selected scanning opening. Moreover, an edge estimate decision means 141 applies intermediate tone estimation depending whether the binary picture element of the fetched input picture element is coincident with the edge picture. Any of the estimation result by the intermediate tone estimate means 131 or the edge estimate decision means 141 is selected and outputted by an intermediate tone selection means 151.

Description

[Detailed Description of the Invention] [Table of Contents] Overview Industrial Field of Application Conventional Technology Problems to be Solved by the Invention Means for Solving the Problems Action Examples Effects of the Invention [Summary] Pseudo-halftone image created Regarding the halftone estimation method for dithered images that estimates the original halftone image from the dithered image, the purpose of this method is to prevent blurring of the image and apply it to mixed images of text, etc. and dithered images. Regarding the dithered original image, a pixel input means takes in pixel data of an input pixel corresponding to the size of the dither matrix, and a pixel position is set for the taken pixel data according to the array of threshold values in the dither matrix. A normalization means that performs normalization and a plurality of scanning apertures are set, and a plurality of binary images obtained by dither matrix are prepared for each scanning aperture assuming that the density is constant, and each of these binary images is input. a halftone estimating means for estimating a halftone according to the ratio of black pixels in the matched binary image of the scanning aperture when the binary image of the pixel is compared; and a predetermined edge image; A plurality of edge images are prepared, and each of these edge images is compared with the binary image of the imported input pixels. and halftone selection means for selecting the halftone estimation result of either the estimation means or the edge estimation means.

[Industrial application field]

The present invention relates to a method for estimating halftones of a dithered image, in which an original halftone image is estimated from a dithered image displayed in pseudo-halftones.

Currently, image input/output devices such as facsimile machines and scanners often handle not only characters and line drawings, but also pseudo-halftone images such as dithered images. By extracting information such as resolution and gradation from general binary images, including images, and faithfully reproducing halftone images, dithered images can be easily handled using conventional image processing for multivalued images. For example, high frequency enhancement, edge detection, enlargement, reduction, etc. can be easily realized.

[Conventional technology]

As a conventional halftone estimation method for pseudo-halftone images such as dithering, there is a known method for reconstructing a gray level image from a dithered image (method of ).This method estimates halftones from the number of black pixels in a dither matrix.

In addition, as another conventional example of the halftone estimation method, there is
No. 1-288567 ``Method for estimating halftone image of dithered image'' and Japanese Patent Application Laid-Open No. 61-288568 ``Method of estimating halftone image of binary image'' (method ■). In this method, multiple scanning apertures are prepared, and one of the scanning apertures is selected based on whether it matches or does not match the dot pattern within the scanning aperture, or one of the scanning apertures is selected depending on the arrangement of pixels within the scanning aperture. This method estimates halftones according to the number of black pixels within the selected apertures.

[Problem to be solved by the invention]

By the way, in method (2) described above, multiple pixels in the dither matrix correspond to one pixel in the halftone image, and the entire image is smoothed, so the density of characters, line drawings, etc. There is a problem in that the estimated image becomes blurred when there is an edge portion where the value changes sharply.

In addition, in method (2), the dot pattern of the prepared scanning aperture is obtained by dithering an image with uniform density using a dither matrix. Since the scanning aperture that coincides with becomes smaller, the estimation error increases,
It was not possible to accurately reproduce edge parts such as characters and line drawings where the density changes sharply. Therefore, when estimating halftones from a mixed image of characters, line drawings, etc. and dithered images, there are many estimation errors and it is difficult to apply this method.

The present invention was created in view of these points, and provides halftone estimation for dithered images that can prevent blurring of estimated images and can be applied to mixed images of characters, line drawings, etc. and dithered images. The purpose is to provide a method.

[Means to solve the problem]

FIG. 1 is a principle block diagram of a halftone estimation method for dithered images according to the present invention.

In the figure, pixel input means 111 takes in pixel data of input pixels corresponding to the size of the dither matrix for an original image that has been dithered using a dither matrix.

The normalizing unit 121 normalizes the pixel position of the pixel data input by the pixel input unit 111 by converting the pixel position according to the arrangement of the threshold values in the corresponding dither matrix.

The halftone estimating means 131 sets a plurality of types of scanning apertures,
A plurality of binary images obtained by dither matrices corresponding to each scanning aperture are prepared for each scanning aperture for an image having a constant density, and each of these binary images and the normalizing means 12
When the binary image of the input pixels normalized by 1 is compared and they match, the halftone is estimated according to the ratio of black pixels in the matched binary image of the scanning aperture.

The edge estimating means 141 prepares a plurality of predetermined edge images, compares each of these edge images with the binary image of the input pixel taken in by the pixel input means 111, and when they match, the edge portion is Perform a corresponding halftone estimation.

The halftone selection means 151 selects and outputs the halftone estimation result of either the halftone estimation means 131 or the edge estimation means 141.

Therefore, the overall configuration is such that halftone estimation is performed by comparing with a plurality of types of scanning apertures, and halftone estimation is performed by extracting edge portions by comparison with edge images.

[For production]

A plurality of types of scanning apertures are set, and a plurality of binary images obtained by using a dither matrix corresponding to each scanning aperture are prepared for each scanning aperture for an image having a uniform density. Therefore,
If the density of the human image for which halftone estimation is performed is uniform over a wide range, the binary image of the input pixel matches the binary image prepared for a large scanning aperture. As the density change of the input image becomes more severe, the scanning aperture at which the binary images match becomes smaller. Compare the binary images of each scanning aperture prepared in this way, select the only scanning aperture whose input pixels match the binary image, and calculate the ratio of black pixels included in this binary image. Halftone estimation is performed accordingly. Also,
When the binary image of the input pixel matches a prepared edge image, the value corresponding to the edge portion is obtained as the halftone estimation result.

When performing such halftone estimation, first, the pixel input means 111 takes in human pixels corresponding to the size of the dither matrix, and the positions of the taken human pixels are converted by the normalization means 121. This conversion of pixel positions is a process to make the array of threshold values in the dither matrix common to the input pixels that have been imported, and is performed using one type of array of the dither matrix (usually one of the rows and columns of the dither matrix or Therefore, the number of binary images of scanning apertures to be prepared can be reduced.

After converting the input pixel position, the halftone estimation means 131 selects a scanning aperture that matches the binary image of the input pixel, and performs halftone estimation according to the ratio of black pixels in the binary image of the selected scanning aperture. . Further, the edge estimating means 141 performs halftone estimation based on whether the binary image of input pixels input by the pixel input means 111 matches the edge image. If they match, it is determined that the pixel of interest is an edge portion, and halftone estimation appropriate for the edge portion is performed. Either the estimation result by the halftone estimation means 131 or the edge estimation means 141 is selected by the halftone selection means 151 and output.

According to the present invention, by preparing a plurality of types of scanning apertures, selecting the optimum scanning aperture, and performing halftone estimation, it is possible to prevent blurring of the estimated image. Furthermore, since the edge portions in the input image are determined, it becomes possible to perform halftone estimation for mixed images in which edge portions such as characters and line drawings are included in the dithered image.

〔Example〕

Hereinafter, embodiments of the present invention will be described in detail based on the drawings.

FIG. 2 shows the configuration of an embodiment to which the dithered image halftone estimation method of the present invention is applied. Further, FIG. 3 shows an outline of the operation procedure of halftone estimation processing in the embodiment.

In FIG. 2, 211 is a pixel input circuit, 221 is a pixel position conversion circuit, 231 is a halftone estimation circuit, and 235 is a pixel input circuit.
241 indicates an edge estimation circuit, and 241 indicates a halftone selection circuit.

The pixel input circuit 211 takes in a systematic dither image read by a facsimile machine, scanner, etc. (FIG. 3, step 31). For example, the input image that has been dithered using a 4×4 Bayer type dither matrix shown in FIG. 4 is targeted.

311, 313.31 in the pixel input circuit 211
5 and 317 each indicate a shift register.

Each shift register shifts the binary data of each pixel that is input sequentially and outputs it as 4-bit (for 4 pixels) parallel data. Therefore, the four shift registers 311 to 317 output binary data of 166 pixels corresponding to the 4×4 dither matrix described above.

The pixel position conversion circuit 221 converts the input pixel position of the pixel data input by the pixel input circuit 211 to normalize the pixel position (FIG. 3, step S2). Here, normalization refers to an operation in which columns and rows of the input pixel group are replaced as necessary to correspond to the dither matrix. In the pixel position conversion circuit 221, 321, 323, 32
5, 327, 331, 333, 335, 337 are rotate registers, 329 is a column counter, and 339 is a row counter, respectively.

In Fig. 5 (a), (b), and (C), the input pixel group (1
The relationship between 66 pixels and a halftone image determined by this input pixel group is shown. As shown in the figure, one pixel of a halftone image is determined based on an input pixel group consisting of 166 pixels.

In addition, in Figure 6, which shows the procedure for normalizing pixel positions by the pixel position conversion circuit 221, "column 09 row O" corresponds to the initial position of the input pixel group to be imported as the original image, and in a general outline A black dot (•) within the enclosed input pixel group indicates the center of movement of the input pixel group. Therefore, when the input pixel group shifts to the right by one column, the center of movement of the input pixel group shifts to (column 12
It becomes line O). Further, "pixel group after position conversion" in the same figure indicates the position of the input pixel group after normalization, and indicates to which position the movement center of the input pixel group has shifted.

Specifically, considering the input pixel group in step 1 as a reference,
Rotation is performed by the number of times that is the remainder of dividing the column and row positions of the center of movement by the number corresponding to the size of the scanning aperture. As the scanning aperture, we are considering a 4 x 4 size corresponding to the dither matrix, so divide the row or column by 4. For example, step 2
Then, since the column remainder is 1, the last column of the input pixel group is rotated to the first column. In step 3, the rear two rows are rotated, and in step 4, the rear three rows are rotated. Also, in step 5, the remainder of the row is 1, so the last row is rotated to the beginning, and step 6
Now rotate both rows and columns.

By interchanging the positions of the columns or rows of input pixels in this way, the positions of the threshold values arranged in the corresponding dither matrix can be fixed, and the processing by the next-stage halftone estimation circuit 231 can be standardized. The processing procedure can be simplified.

When the above-mentioned slant normalization process is performed by the pixel position conversion circuit 221, the column counter 329 calculates the column remainder, and the row counter 329 calculates the row remainder. In addition, each of the four rotation registers 321 to 327 takes in data for four pixels output from each of the shift registers 311 to 317, and performs a rotation operation according to the number of remaining columns calculated by the column counter 329. This is what we do.

Specifically, the rotation register 321 takes in the 4-bit data output from the shift register 311, performs a predetermined rotation process, and then transfers each bit of the rotated 4-bit data to the next stage's four rotation registers. It is input to each of registers 331-337. Similarly, the rotate registers 323 to 327 take in 4-bit data from the corresponding shift registers 313 to 317, perform predetermined rotation processing, and then transfer each bit to each of the four rotate registers 331 to 337. Enter.

Each of the four rotate registers 331 to 337 receives four input signals from the rotate registers 321 to 327.
It takes in bit data and performs a rotation operation according to the number of remaining rows calculated by the row counter 339. The 16-bit data (4×4 bits) after rotation processing is output from the pixel position conversion circuit 221.

The halftone estimation circuit 231 uses a normalized input pixel pattern corresponding to the 16-bit data output from the pixel position conversion circuit 221 and dot patterns of multiple types of scanning apertures (hereinafter referred to as gradation patterns) prepared in advance. ), and the halftone is estimated according to the matched tone pattern (FIG. 3, step S3).

FIG. 7 shows an example of a scanning aperture prepared as a comparison target. For example, it is assumed that nine types of opening O to closed opening are prepared. The position of a black dot (.) included in each aperture shown in the figure indicates a position common to all the apertures, and when comparing patterns, pattern comparison is performed centering on this black dot.

Incidentally, there are 2 to the 16th power of gradation patterns (binary images) in the aperture O, but the number is limited to 17 if there is no density change in the original image. When an original image with a -like density is dithered using the dither matrix shown in Figure 4, the density of the original image is thickened by one gradation (and the resulting dithered image changes from completely white to one pixel). The pixel gradually changes to black, and finally becomes completely black.Therefore, for aperture 0, 17 gradation patterns from all white to all black are prepared.

Similarly, when there is no density change in aperture 1, the number of patterns is limited to nine, and these nine gradation patterns are prepared. Thereafter, in the same manner, a gradation pattern obtained assuming that there is no change in density is prepared for each opening.

In this way, if you prepare a dot pattern for the case where there is no density change as the gradation pattern for each aperture to be compared, then when there is no density change in the input pixel group (when there is no density change in the image before dither processing), , the dot pattern of the input pixel and any gradation pattern of aperture 0 should match. Furthermore, as the density change becomes more severe, the aperture where the dot pattern of the input/output pixel matches the gradation pattern becomes smaller.

The halftone estimation circuit 231 compares the dot pattern and the tone pattern of the input pixel group, selects the tone pattern of the aperture with which the dot patterns match and has the largest area, and selects the tone pattern of the aperture that has the largest area within the selected tone pattern. Halftone estimation is performed based on the number of black pixels.

Note that since the size of each aperture is different, the number of black pixels is multiplied by a constant value according to the size of the aperture to obtain an intermediate tone. The relationship between the size of this aperture and the constant value is shown in Figure 8. .

The "multiply value" in the figure indicates the inverse ratio between the size of the corresponding aperture and the size of aperture 0, and the difference in aperture size is adjusted by this multiplier value.

FIG. 9 shows a specific example of the aperture deriving operation. The "input pixel" in the figure indicates a dot pattern of 166 pixels after normalization output from the pixel position conversion circuit 221 or a part thereof.

First, in step 1, the dot pattern of the input pixel and the 17 gradation patterns of aperture 0 are compared. In this case, since all the patterns do not match, the process proceeds to step 2.

In step 2, the dot pattern of the input pixel corresponding to the aperture 1 is compared with nine types of gradation patterns of the aperture 1.

In this case, too, since there is no match, the process proceeds to step 3.

In step 3, the dot pattern of the input pixel corresponding to the aperture 2 and the nine types of gradation patterns of the aperture 2 are compared.

In this case, the fourth gradation pattern matches, and the number of black pixels in this pattern is 3, so a value of 6, which is the sum of the multiplication values of 2 and 6, is output as the halftone estimate.

Actually, the halftone estimation circuit 231 is implemented as a halftone estimation ROM.
233, the 16-bit data corresponding to the input pixel is input as an address, and an estimated halftone value is obtained and outputted by comparing it with the gradation pattern of each aperture and multiplying it with a multiplication value.

The edge estimation circuit 235 uses an input pixel pattern before normalization corresponding to the 16-bit data outputted from the pixel input circuit 211 and dot patterns of multiple types of scanning apertures prepared in advance for edge extraction (hereinafter referred to as edge patterns). 3) and perform edge estimation according to the matching edge pattern to estimate the halftone (Fig. 3, Step 3).
4).

FIG. 10 shows an example of a scanning amount Ron for edge extraction prepared as a comparison target. For example, it is assumed that five types of openings O to 4 are prepared. The positions of black dots (・) included in each aperture shown in the same figure indicate the positions common to all apertures.
When comparing patterns, pattern comparison is performed centering on this black point.

Further, FIG. 11 shows an edge pattern assuming that each opening shown in FIG. 10 includes an edge portion. Also, rB at the bottom of each edge pattern in FIG.
indicates that the pixel of interest is a black pixel in the edge part
indicates that the pixel of interest is a white pixel in the edge portion.

Therefore, when the input pixel pattern matches each edge pattern corresponding to "B", the edge estimation circuit 235
The corresponding pixel is determined to be a black pixel in the edge portion, and the halftone estimated value 16, which is the same as when all pixels in the edge pattern are black pixels, is output. Further, when the input pixel pattern matches each edge pattern corresponding to rWu, the corresponding pixel is determined to be a white pixel in the edge portion,
The halftone estimated value O, which is the same as when all pixels in the edge pattern are white, is output. Note that in the comparison with the edge pattern, the comparison with opening 0 is performed first, and cases where edges exist over a wide range are preferentially extracted.

In reality, the edge estimation circuit 235 is implemented as an edge estimation ROM.
237, the output of the pixel input circuit 211 is used as the address input, and the halftone estimated value O when the edge portion is determined.
Or find 16 at once.

The halftone selection circuit 241 selects and outputs the halftone estimated value output from either the halftone estimation circuit 231 or the edge estimation circuit 233 (FIG. 3, step S5).

This selection condition is shown in FIG. In the same figure, “O”
``x'' indicates a comparison with the edge pattern shown in FIG. 11. Edge estimation circuit 2 that obtains halftone estimated values by
This corresponds to the case where 35 is selected.

In the halftone selection circuit 241, 243 indicates a multiplexer, and 245 indicates a selection decision ROM245.
33 and the edge estimation ROM 237 (information indicating which gradation pattern or edge pattern matched), selection instruction data is created and sent to the multiplexer 243. R for selection decision
The OM 245 outputs selection instruction data according to the selection conditions shown in FIG. Select the key estimation circuit.

In this way, each pixel position of the input pixel group taken in by the pixel input circuit 211 is normalized by the pixel position conversion circuit 221,
The binary image of this normalized input pixel group is compared with the gradation patterns of multiple apertures prepared, and in this comparison, the halftone is Make an estimate. In addition, it is determined whether a predetermined edge pattern appears in the binary image of the input pixel group captured by the pixel input circuit 211, and if it appears, regardless of the ratio of surrounding white pixels and black pixels. (Regardless of the tone pattern), 0 or 16 is obtained as the estimated halftone value corresponding to the edge portion.

Therefore, by determining the edge pattern and obtaining the estimated halftone value corresponding to the edge portion, it is possible to perform accurate halftone estimation even for portions of the dithered image where the density changes are steep, such as characters and line drawings. This makes it possible to apply this method to mixed images of characters, line drawings, etc. and dithered images.

Further, for areas other than edge portions, it is possible to perform optimal halftone estimation according to the state of density change, so that blurring of the estimated image can be minimized.

In the embodiment of the present invention, the dot patterns were compared using the nine types of scanning apertures shown in FIG. More than nine types of scanning apertures or less than nine types of scanning apertures may be used as long as the number of scanning apertures is such that the number of scanning apertures is equal to or less than nine.

Further, the same applies to the edge extraction scanning aperture shown in FIG. 11, and each gradation pattern and edge pattern are not limited to those of the embodiment.

Furthermore, in the embodiment, the halftone estimation circuit 231 performs halftone estimation by comparison with the gradation pattern first;
The operations of the halftone estimation circuit 231 or the edge estimation circuit 235 may be performed either first or in parallel.

[Effects of the Invention] As described above, according to the present invention, it is possible to prevent blurring of an estimated image by preparing multiple types of scanning apertures, selecting the optimal scanning aperture, and performing halftone estimation. It becomes possible. Furthermore, if there is an edge part in the input image, halftone estimation is performed by determining this edge part.
It can also be applied to mixed images of characters, line drawings, etc. and dithered images.

[Brief explanation of the drawing]

Fig. 1 is a principle block diagram of the halftone estimation method for dithered images of the present invention, Fig. 2 is a block diagram of an embodiment to which the halftone estimation method for dithered images of the present invention is applied, and Fig. 3 is an embodiment. Figure 4 is an illustration of the 4x4 Bayer dither matrix, Figure 5 is an illustration of the correspondence between an input image and a halftone image, and Figure 6 is an illustration of input pixel position conversion. , Fig. 7 is an explanatory diagram of the scanning aperture, Fig. 8 is a diagram of the correspondence between the aperture and the multiplication value, Fig. 9 is an explanatory diagram of a specific example of the aperture selection operation, and Fig. 10 is an explanation of the scanning aperture for edge extraction. 11 is an explanatory diagram of an edge pattern, and FIG. 12 is an explanatory diagram of halftone selection conditions. In the figure, 111 is a pixel input means, 121 is a normalization means, 131 is a halftone estimation means, 141 is an edge estimation means, 151 is a halftone selection means, 211 is a pixel input circuit, 221 is a pixel position conversion circuit, and 231 is a pixel position conversion circuit. Halftone estimation circuit; 233 is a ROM for halftone estimation; 235 is an edge estimation circuit, 237 is an edge estimation ROM, 241 is a halftone selection circuit, 243 is a multiplexer, and 245 is a selection determination ROM. 311.313, 315, 317 are shift registers, 321.323, 325, 327, 331, 333.3
35 and 337 are rotation registers, 329 is a column counter, and 339 is a row counter. 1 L double 3] [Nio 14] [Technical asl [Takumi 6] Input pixel's f hill 1 block is removed βRI2 Edge is used for the exit run j'-mouth 44B month figure 10 middle g1 ftOF-IjGyu・Mokomyo-zu Figure 12 Sekiguchi 0 Opening 1 Opening 2 Opening 3 Frontage 4 -1-T

Claims (1)

[Claims] (1) Pixel input means (111) that captures pixel data of input pixels corresponding to the size of the dither matrix for the original image that has been dithered with the dither matrix.
and normalizing means (121) for normalizing pixel positions by converting the pixel positions of the pixel data taken in by the pixel input means (111) according to the arrangement of threshold values in the corresponding dither matrix. Then, a plurality of scanning apertures are set, and a plurality of binary images obtained by the dither matrix corresponding to each scanning aperture are prepared for each scanning aperture for an image with a constant density, and each of these binary images is The image is compared with the binary image of the input pixels normalized by the normalization means (121), and if they match, the image is compared with the binary image of the input pixels normalized by the normalization means (121), and when they match, the image is A halftone estimation means (131) for estimating halftones and a plurality of predetermined edge images are prepared, and each of these edge images and the pixel input means (11
an edge estimating means (141) that performs halftone estimation corresponding to an edge portion when comparing the binary image of the input pixels captured in step 1) and estimating a halftone corresponding to an edge portion; A halftone estimation method for a dithered image, comprising: a halftone selection means (151) that selects and outputs the halftone estimation result of one of the edge estimation means (141).