JPH10224644A - Image processor - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide an image processor reduced in the sense of roughness when the number of gray levels is decreased. SOLUTION: A 1st color space value conversion means 10 obtains a 3-stimulus value from a pallet of an image forming device 3 and converts the 3-stimulus value into a 1st color space where the 3-stimulus value is arranged in a uniform color space. An adjusted 3-stimulus value generating means 20 obtains the 3-stimulus value from the input image to generate an adjusted 3-stimulus value obtained by adding errors from adjacent picture elements to the 3-stimulus value. A 2nd color space conversion means 30 converts the adjusted 3-stimulus value into a 2nd color space value where the adjusted 3-stimulus value is arranged in a uniform color space. A norm calculation means 40 obtains a difference between the 1st color space value and the 2nd color space value and calculates a norm. A minimum pallet selection means 50 selects a minimum pallet minimizing the norm. An output means 60 provides an output of a minimum pallet to the image forming device 3. An error spread means 70 obtains the 3-stimulus value between the input image and the minimum pallet uses the difference for an error, which is spread to adjacent picture elements.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an image processing apparatus, and more particularly, to reducing the number of gradations of an input image in order to output a multi-gradation input image to an image forming apparatus which forms an image with a smaller number of gradations. The present invention relates to an image processing apparatus for performing the above.

[0002]

2. Description of the Related Art Recent advances in color printers, especially ink jet printers, have made it possible to easily obtain color hard copies. At the same time, desktop publishing (DTP), which creates a publication using a small computer such as a personal computer or a workstation, has become widespread, and an environment for creating and printing visual documents has begun to be rapidly established.

[0003] When an image displayed on a color monitor such as a personal computer is printed out by a color printer, a color matching technique is used. On a color monitor, the three primary colors of R (red), G (green), and B (blue) are displayed as an additive color mixture. On a color printer, the three primary colors of C (cyan), M (magenta), and Y (yellow) are displayed. , The difference in brightness, chromaticity, and saturation range that can be expressed by the former and the latter is adjusted by this color matching.

When a color monitor is displayed using a CRT, a full-color display of 256 tones of 16.7 million colors is generally used for each color, and an ink-jet type color printer generally has eight colors of 2 tones for each color. . Therefore,
In order to adjust the difference in the number of gradations, halftoning techniques such as pattern dither and error diffusion are performed, and these techniques are used in combination with a color matching system as appropriate.

The above-mentioned error diffusion means that, for example, when the pixels arranged in line with 200 and 10 are to be displayed in two gradations of 0 and 255 as an input image, the first pixel wants to display 200, so that 0 or 255 The closest one is selected as the value of the output image.

In this case, an extra display of 255-200 = 55 is carried over to the next pixel as an error. In the next pixel, a value -45 obtained by subtracting the previous error 55 from the initial value 10 is a value to be displayed.

In this case, since 0 is closer, 0 is selected and used as the value of the output image.
(= 0 − (− 45)) is carried over to the next pixel as an error. Then, the selection and the diffusion of the error are sequentially performed for the next pixel and thereafter.

In this way, since the cumulative sum of the error between the input image and the output image is always set to 0, the average value of the pixels having a certain area (hereinafter referred to as the area average) is calculated using the input image value and the output image value. ) Are approximately equal.

Next, a processing procedure for converting a full-color image to be output to a color monitor into an image to be output to a color printer will be described. FIG. 6 is a flowchart showing a processing procedure when converting an image displayed on a color monitor into an output image of a color printer. [S20] The full-color input image to be displayed on the color monitor is subjected to color conversion by using color matching in consideration of the primary colors of the color printer (RGB → CMY). [S21] It is determined whether it is necessary to add a value corresponding to an error from a neighboring pixel. If it needs to be added, the process proceeds to step S22. If it does not need to be added, the process proceeds to step S23.
Go to Here, step S23 and subsequent steps are performed on the C image. [S22] An error value from a neighboring pixel is added to the value of the C1 pixel of the input image. [S23] A value closest to the pixel C1 is selected and output as an output image from among the values that can be output. [S24] The difference between the value to be output and the selected value is set as a nearby error. [S25] It is determined whether or not execution has been performed for all elements in the column. If all the elements have been executed, the process proceeds to step S27; otherwise, the process proceeds to step S26. [S26] The process moves to the next column element and returns to step S21. [S27] It is determined whether the process has been performed for all rows. If executed, the process ends. If not, the process proceeds to step S28. [S28] Move to the next line and return to step S21.

In the above flowchart, the error diffusion is performed on the C image, but the same error diffusion is performed on the M and Y images. In this case, when the input image and the output image of each of C, M, and Y are considered as the area average, the respective area averages can be made substantially equal.

However, when considering the color of the pixel of the output image in which the area average is synthesized, for example, when error diffusion is performed on the C data of a certain pixel, other M and Y conditions are not considered at all. However, when considered on an individual pixel basis, an optimal color has not necessarily been selected for the color of the pixel of the original input image.

For the purpose of selecting an optimum color, for example, instead of independently performing error diffusion on the values of a CMY image, the values of an input image are converted into tristimulus values, and Journal of E has proposed a method in which error diffusion is performed in a space (hereinafter referred to as XYZ space) to select a more optimal color.
Electronic Imaging 5 (1), 97-106.

FIG. 7 is a flowchart showing a procedure for converting an input image value into tristimulus values, performing error diffusion in an XYZ space, and selecting an optimum color. [S30] The colors (each palette) that can be displayed on the color printer are converted into tristimulus values (Xi, Yi, Zi). [S31] The values of the pixels of the input RGB image to be displayed on the color monitor are displayed on each channel (RGB) of the color monitor.
Is converted into tristimulus values (X1, Y1, Z1) in consideration of the input-luminance characteristics, chromaticity coordinates of each color primary color, chromaticity coordinates of a white point, and the like. [S32] It is determined whether it is necessary to add a tristimulus value as an error from a neighboring pixel. If it needs to be added, the process proceeds to step S33. If it does not need to be added, the process proceeds to step S34. [S33] A tristimulus value as an error from neighboring pixels is added to the tristimulus value of the input image. [S34] (Xi, Yi, Zi) and (X1, Y1, Z
The difference between the components of 1) is

[0014]

ＸX = Xi-X1 ＹY = Yi-Y1 ＺZ = Zi-Z1 (1) The norm of Expression (1) is defined as Expression (2).

[0015]

(Equation 2)

Then, a palette in which the norm △ Ei is minimized is selected from each palette, and this pixel is displayed. [S35] The difference between the tristimulus value of the color of the pixel to be displayed next and the tristimulus value of the color of the selected palette is determined as a neighboring error. [S36] It is determined whether the process has been performed for all elements in the column. If all the elements have been executed, the process proceeds to step S38; otherwise, the process proceeds to step S37. [S37] The process moves to the next column element and returns to step S32. [S38] It is determined whether the process has been performed for all rows. If executed, the process ends; otherwise, the process proceeds to step S39. [S39] Move to the next line and return to step S32.

In this way, by combining pixels of colors whose tristimulus values are locally close to each other, a color whose color appearance (tristimulus value) matches can be displayed globally.

[0018]

However, in the prior art as described above, the XYZ space is in this space, and when the norm is 0, it can be said that the two colors are the same, but visually the XYZ space is the respective colors. Not even in direction. For this reason, even if the pixel of the input image desired to be displayed in the XYZ space and the palette having the smallest norm are selected, the palette does not necessarily have a similar color appearance to human senses, but has the smallest color difference. Did not.

Therefore, the combination of pixels of colors that are not necessarily small in color difference makes the area-averaged color appearance uniform, so that there is a problem that a rough feeling as a noise component is felt.

SUMMARY OF THE INVENTION The present invention has been made in view of the above points, and has been made in consideration of the above-described problems, and has been made in view of an image processing apparatus which reduces roughness as a noise component when outputting a multi-gradation input image with a smaller number of gradations. The purpose is to provide.

[0021]

According to the present invention, in order to solve the above-mentioned problems, in order to output a multi-gradation input image to an image forming apparatus which forms an image with a smaller number of gradations, the input image has to be divided into In the image processing apparatus for reducing the number of gradations, for the palette in which the image forming apparatus is capable of forming an image, a tristimulus value is obtained from the palette, and then the tristimulus value is placed on a uniform color space. First color space value conversion means for converting into the arranged first color space value;
Adjusting tristimulus value generating means for obtaining a stimulus value and adding an error from a neighboring pixel to the tristimulus value, and a second tristimulus value generating means for arranging the tristimulus value on the uniform color space Second color space value converting means for converting the first color space value into a second color space value, a norm calculating means for calculating a norm after calculating a difference between the first color space value and the second color space value, A minimum pallet selecting unit for selecting a minimum pallet that minimizes the following; an output unit for outputting the minimum pallet to the image forming apparatus; and a difference between a tristimulus value of the input image and a tristimulus value of the minimum palette. Error diffusion means for determining the difference and diffusing the difference as the error to the neighboring pixels.

Here, the first color space value conversion means calculates a tristimulus value from the palette for a palette which is a color in which the image forming apparatus can form an image, and then converts the tristimulus value into a uniform color space. Convert to the first color space value located above. The adjusted tristimulus value generation means obtains tristimulus values from the input image, and
An adjusted tristimulus value is generated by adding an error from a neighboring pixel to the stimulus value. The second color space value conversion means converts the adjusted tristimulus values into second color space values arranged on a uniform color space.
The norm calculating means calculates a norm after obtaining a difference between the first color space value and the second color space value. The minimum pallet selecting means selects the minimum pallet that minimizes the norm. The output unit outputs the minimum pallet to the image forming apparatus. The error diffusion means obtains a difference between the tristimulus value of the input image and the tristimulus value of the minimum palette, and diffuses the difference to neighboring pixels as an error.

[0023]

Embodiments of the present invention will be described below with reference to the drawings. FIG. 1 is a principle diagram of the image processing apparatus of the present invention. The image processing apparatus 1 is an image forming apparatus (for example, a color printer) 3 that forms an input image of multiple gradations to be displayed on the color monitor 2 with a smaller number of gradations.
, The number of gradations of the input image is reduced.

The first color space value converting means 10 converts the color of the pallet which is capable of forming an image by the image forming apparatus 3
After obtaining the tristimulus values from the palette, the tristimulus values are converted to first color space values arranged on a uniform color space.

The adjusted tristimulus value generating means 20 obtains tristimulus values from the input image from the color monitor 2 and generates an adjusted tristimulus value by adding an error from a neighboring pixel to the tristimulus value. The second color space value conversion means 30 converts the adjusted tristimulus values into second color space values arranged on a uniform color space.

The norm calculating means 40 calculates the norm after obtaining the difference between the first color space value and the second color space value. The minimum pallet selecting means 50 selects the minimum pallet that minimizes the norm.

The output means 60 outputs the minimum pallet to the image forming apparatus 3. The error diffusion means 70 determines whether the input image
The difference between the stimulus value and the three stimulus values of the minimum palette is obtained, and this difference is diffused to neighboring pixels as an error.

Next, a detailed operation of the present invention for converting a multi-tone input image into an output image having a smaller number of tones will be described. FIG. 2 is a flowchart showing an operation procedure when converting an input image with multiple tones into an output image with a smaller number of tones using the image processing apparatus 1 of the present invention. FIG. 3 is a flowchart showing an operation procedure when the processing shown in FIG. 2 is performed for all pixels. [S1] The first color space value conversion unit 10 converts the tristimulus values of the palette, which are colors that can be displayed on the image forming apparatus 3, into a uniform color space by CIE (Commission International Commission: Commission Interna
La placed in the Lab space of nation de l'Eclairage)
It is converted to a b value (Li, ai, bi). [S2] Adjustment 3 stimulus value generation means 20
The values of the pixels of the input RGB image to be displayed on the color monitor 2 are calculated based on the tristimulus values (X1 , Y1, Z1). [S3] The adjusted tristimulus value generating means 20 determines whether it is necessary to add a tristimulus value as an error from a neighboring pixel. If it is necessary to add, go to step S4.
If there is no need to add, go to step S5. [S4] The adjusted tristimulus value generation means 20 adds tristimulus values as errors from neighboring pixels to the tristimulus values of the input image. [S5] The second color space value conversion means 30 converts the adjusted tristimulus values into Lab values (L1, a1, b1) arranged on a Lab space which is a uniform color space. [S6] The norm calculating means 40 calculates (Li, ai, bi)
And the difference between each component of (L1, a1, b1)

[0029]

## EQU3 ## When ΔL = Li−L1 Δa = ai−a1 Δb = bi−b1 (3), the norm of Expression (3) is defined as Expression (4).

[0030]

(Equation 4)

Then, the minimum pallet selecting means 50 selects a pallet having the minimum norm ΔEi (color difference) from each pallet and displays the pallet as this pixel. [S7] The error diffusion unit 70 sets the difference between the adjusted tristimulus value and the tristimulus value of the selected palette color as an error. [S8] It is determined whether the process has been performed for all elements in the column. If the processing has been executed for all the elements, the procedure goes to step S10; otherwise, the procedure goes to step S9. [S9] The process moves to the next column element and returns to step S3. [S10] It is determined whether or not execution has been performed for all rows. If executed, the process ends. Otherwise, the process proceeds to step S11. [S11] Move to the next line and return to step S3.

Next, functional blocks of the image processing apparatus 1 will be described. FIG. 4 is a diagram illustrating functional blocks of the image processing apparatus 1. The first color space value converting means 10 converts tristimulus values into first color space values arranged on a uniform color space after obtaining tristimulus values from the palette.

The tristimulus value generating means 20a obtains tristimulus values from the input image from the color monitor 2. The adjusted tristimulus value generating means 20b generates an adjusted tristimulus value by adding an error from a neighboring pixel to the tristimulus value.

The second color space value conversion means 30 converts the adjusted tristimulus values into second color space values arranged on a uniform color space. The norm calculating means 40 calculates a norm after obtaining a difference between the first color space value and the second color space value. The minimum pallet selecting means 50 selects the minimum pallet that minimizes the norm.

The output means 60 outputs the minimum palette of each of the C, M, and Y planes to the image forming apparatus 3. The error diffusion means 70 calculates three stimulus values of the input image and three stimulus values of the minimum palette.
The difference from the stimulus value is obtained for each of the X, Y, and Z planes, and the difference is fed back to the adjustment tristimulus value generation means 20b as an error and diffused to neighboring pixels.

As described above, the image processing apparatus 1 according to the present invention has a space in which colors are arranged so that they are equidistant when differences in brightness, chromaticity, and saturation are the same as human perception. The three stimulus values are arranged on the Lab space of the uniform color space.

As a result, the space is more sensuously more uniform than in the XYZ space. Therefore, if a pallet is selected in this space, a pallet having a smaller color difference (hereinafter, referred to as “appearance is close”) can be selected by the human sense. In addition, it is possible to reduce the feeling of roughness as a noise component.

Therefore, it is possible to display a color whose color appearance (tristimulus value) is globally correct while reducing roughness by using a combination of pixels having colors that are closest to human senses. become.

In the above description, the uniform color space is C
Although the IELab space is used, a CIELV space may be used. Further, the arrangement of tristimulus values is not limited to these two spaces but can be performed in various uniform color spaces.

Further, as compared with the conventional method, there is an advantage that it is not necessary to perform color conversion on a full-color image of the color monitor 2 as an input image using a color matching system. As described above, the present method is a general method for converting a multi-tone color image into a low-tone color image,
The present invention can be applied not only to a printer but also to a low-gradation image forming apparatus such as a liquid crystal display or a plasma display.

Since the gamut of a CMY three-color printer is particularly narrow in the black direction, a method for improving this is to print in four colors by adding black (K) in addition to the three colors of CMY. In this case, conventionally, since three dimensions are converted to four dimensions, one-dimensional redundancy is required, and it is difficult to perform color conversion.

However, according to the present invention, there is no need to perform color conversion.
By simply making a four-color palette, the other parts can be handled without any change. Further, the present invention can be applied to an image forming apparatus of four or more colors.

Next, a system configuration using the image processing apparatus 1 of the present invention will be described. FIG. 5 is a diagram showing a system configuration using the image processing apparatus 1 of the present invention. The input image scanned by the color scanner 4 is input to the image processing apparatus 1.
Through the color monitor 2. The input image is output to the color printer 3a after the number of gradations is reduced in the image processing device 2, and color printing is performed.

In this figure, the input image has 256 gradations for each color,
It is RGB data of a total of 16 million colors.
Internally, the data is converted into CMY data of, for example, two gradations for each color, for a total of eight colors.

Next, as a minimum palette selecting means, when the adjusted tristimulus value becomes a value in a range where image formation is not possible, a palette is selected on a linear transformation space of tristimulus values instead of a norm of a uniform color space. Will be described. According to the present invention, the color of the pixel to be displayed in the uniform color space and the palette that minimizes the color difference are selected, and the difference between the tristimulus values is error-diffused, but the original tristimulus value of the pixel to be displayed is actually Since the colors exist, they can be arranged in a uniform color space.

However, as a result of diffusing an error from a neighboring pixel, the adjusted tristimulus value of the pixel to be displayed may deviate from the color gamut that can be displayed by the image forming apparatus, or a part of the value may be negative.

However, a general uniform color space defines a color space with colors actually seen (all of the three stimulus values are positive). Therefore, extrapolation is performed outside the color gamut, or even a part of the color space is negative. If it exists, it becomes impossible to define, and the pixels to be displayed in the uniform color space cannot be arranged properly, and the palette with the minimum color difference cannot be selected. In such a case, this problem can be avoided by selecting the palette closest to the pixel to be displayed using the XYZ space or its linear transformation space.

Next, a description will be given of a case where weighting is performed on the luminance component and the chromaticity component at the time of norm calculation and calculation is performed. Since the number of gradations of the output image data to be displayed is smaller than that of the input image, the error is perceived as rough although the number is reduced by the above method.

This error is converted from the tristimulus values of the palette to the uniform color space by converting the tristimulus values of the pixels of the input image to be displayed into the uniform color space, so that the norm in this space is minimized. Since this is the result of selecting a pallet, the selection of the minimum color difference is performed in pixel units. However, the noise component is in a region having a considerably high spatial frequency.

The contrast recognition characteristic of the human visual system differs between the luminance component and the chromaticity component, and it is generally considered that the former has a band-pass characteristic and the latter has a low-pass characteristic. The uniform color space is considered to be a space in which various colors are arranged isotropically in the luminance and chromaticity directions in a region where the spatial frequency is low, such as a color chart or a color mark, but in a region with a high spatial frequency such as noise, Due to the bandpass characteristics of the components and the lowpass characteristics of the chromaticity components, the luminance component can be recognized up to higher frequencies,
It is no longer considered isotropic.

Therefore, it is necessary to adjust the sensitivity between the luminance component and the chromaticity component in order to equalize the space in a high spatial frequency region such as noise. For example, in Expression (4), the norm is calculated at the same ratio between the error of the luminance component and the error of the chromaticity component. Here, the weight in the luminance direction is newly set to KL, and the chromaticity direction (a direction) is newly calculated. ) Is defined as Ka, and the weight in the chromaticity direction (b direction) is defined as Kb, and the norm is defined as the following equation.

[0052]

(Equation 5)

By changing the weights (KL, Ka, Kb),
If the norm is calculated with an emphasis on the luminance component or the chromaticity component and error diffusion is performed, the luminance component of the noise component is K when Ka, Kb> KL, and the chromaticity component of the noise component is K when Ka, Kb <KL.
It can be reduced more than when L = Ka = Kb.

In the case of Ka / KL> 2 and Kb / KL> 2, the noise of the luminance component decreases, but the color noise increases because the luminance is given too much importance. Also, 1> Ka / KL, 1> K
In the case of b / KL, the noise of the chromaticity component is reduced, but the noise of the luminance component is increased because the chromaticity is given too much importance. Since the characteristics of the human visual system can be recognized up to the spatial frequency region where the luminance component is high as described above, it is desired to reduce noise in the chromaticity direction from the luminance direction. Therefore, 2> Ka / KL> 1, 2
> Kb / KL> 1 is appropriate.

In the above description, the uniform color space is CI
Although the ELab space is used, the noise in the luminance direction and the noise in the chromaticity direction can be similarly reduced in the CIELV space and other uniform color spaces, thereby reducing the roughness of the screen.

As described above, in order to express the data to be expressed on the multi-tone color monitor 2 by the image forming apparatus 3 with a smaller number of tones, the color of the color which is close in appearance when the tone is compressed is reduced. It was configured to combine pixels. This makes it possible to reduce the roughness of the screen while making the colors of the input image and the output image uniform.

Further, by changing the weighting factor of the error of the luminance component and the weighting factor of the error of the chromaticity component for calculating the norm when selecting the pallet, it is possible to further reduce the roughness on both sides. .

[0058]

As described above, the image processing apparatus of the present invention arranges tristimulus values of an input image and an output image to an image forming apparatus on a uniform color space, and obtains a norm from these color space values. A palette with a small difference is selected and output, and the difference between the tristimulus value of the input image and the tristimulus value of the selected palette is distributed to neighboring pixels as an error. As a result, it is possible to make the colors of the input image and the output image uniform, and it is possible to reduce the roughness that is a noise component.

[Brief description of the drawings]

FIG. 1 is a principle diagram of an image processing apparatus according to the present invention.

FIG. 2 is a flowchart showing an operation procedure when converting an input image with multiple gradations into an output image with a smaller number of gradations by using the image processing apparatus of the present invention.

FIG. 3 is a flowchart showing an operation procedure when the processing shown in FIG. 2 is performed for all pixels.

FIG. 4 is a diagram showing functional blocks of the image processing apparatus of the present invention.

FIG. 5 is a diagram showing a system configuration using the image processing apparatus of the present invention.

FIG. 6 is a flowchart illustrating a processing procedure when converting an image displayed on a color monitor into an output image of a color printer.

FIG. 7 is a flowchart illustrating a processing procedure for converting an input image value into tristimulus values, performing error diffusion on an XYZ space, and selecting an optimal color.

[Explanation of symbols]

DESCRIPTION OF SYMBOLS 1 ... Image processing apparatus, 2 ... Color monitor, 3 ... Image forming apparatus, 10 ... First color space value conversion means, 20 ...
Adjustment tristimulus value generation means, 30 second color space value conversion means, 40 norm calculation means, 50 minimum palette selection means, 60 output means, 70 error diffusion means.

Claims (3)

[Claims] 1. An image processing apparatus for reducing the number of tones of an input image in order to output a multi-tone input image to an image forming apparatus for forming an image with a smaller number of tones, A first color for obtaining a tristimulus value from the palette for a palette that is a color capable of forming an image, and then converting the tristimulus value to a first color space value arranged on a uniform color space A spatial value conversion unit, an adjusted tristimulus value generating unit that obtains tristimulus values from the input image, and generates an adjusted tristimulus value by adding an error from a neighboring pixel to the tristimulus value; And a second color space value converting means for converting the second color space value from the first color space value into a second color space value arranged on the uniform color space. A norm calculating means for calculating a norm, wherein the norm is minimized A minimum pallet selecting means for selecting a small pallet; an output means for outputting the minimum pallet to the image forming apparatus; a difference between a tristimulus value of the input image and a tristimulus value of the minimum pallet; And an error diffusion means for diffusing the error as the error to the neighboring pixels. 2. The minimum pallet selecting means, wherein the error is added to the adjusted tristimulus value, and the adjusted tristimulus value becomes a value in a range where image formation is impossible in the image forming apparatus. 2. The method according to claim 1, wherein the calculating unit calculates the norm from a difference between the tristimulus value and the adjusted tristimulus value of the pallet or a linear conversion value thereof, and selects a pallet that minimizes the norm. Image processing device. 3. The image processing apparatus according to claim 1, wherein the norm calculating means calculates the norm by weighting a luminance component and a chromaticity component.