JP2004064629A - Image processing apparatus and its method - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To reduce quantization error accompanying color transformation processing by using the multi-value dither method. <P>SOLUTION: The color transformation processing combines 2 methods, i.e., 3D-LUT (three dimensional-look up table) processing of 8 bit-10 bit, and the multi-value dither processing of 10 bit-8bit. Keeping a number of bits before and after the color transformation processing at 8 bit excludes effect to other processing sections and prevents increase of the circuit size. <P>COPYRIGHT: (C)2004,JPO

Description

[0001]
TECHNICAL FIELD OF THE INVENTION
The present invention relates to an image processing apparatus such as a scanner or a printer for expressing good gradation, and a method therefor.
[0002]
[Prior art]
In a conventional image processing apparatus, for example, as shown in FIG. 2, an RGB color 8-bit image input by an image input unit is color-converted into R′G′B ′ 8-bit images by an input color conversion unit. Then, the R'G'B 'is color-converted into 8 bits for each of CMYK by output color conversion means, and the density of CMYK is adjusted by the following output gamma correction means. Convert. Finally, the output image of C'M'Y'K 'is printed on an output medium by the image output means.
[0003]
In this case, the input color conversion means prepares a relationship of 8 bits RGB-8 bits R'G'B 'arranged in a three-dimensional lattice in advance as a 3D-LUT (208), and performs an arbitrary RGB input by a spatial interpolation operation. R′G′B ′ output value is calculated. Several methods have been proposed as interpolation methods in this case, and tetrahedral interpolation disclosed in Japanese Patent Publication No. 58-16180 is one of them.
[0004]
[Problems to be solved by the invention]
However, in the input color conversion means in this conventional technique, the quantization error accompanying the interpolation operation cannot be ignored. This will be described with reference to FIG.
[0005]
First, for the sake of simplicity, it is assumed that the input RGB value is a gray color R = G = B = Gray (8 bits). It is also assumed that the output R'G'B 'for the gray input is also gray, and R' = G '= B' = Gray '(8 bits). Further, it is assumed that the number of grid points of the 3D-LUT (208) is 16 for each RGB axis, and the grid point intervals are equal.
[0006]
There are 16 lattice points (302) on a diagonal line (301) connecting black (R = G = B = 0) and white (R = G = B = 255) in the 3D-LUT. At these grid points, an output Gray ′ value for the input Gray value is set, and it is assumed that the relationship is a downwardly convex relationship as shown in 303.
[0007]
In tetrahedral interpolation, the output value at the gray input is a linear interpolation of the output Gray 'values at two grid points on a diagonal line sandwiching the input Gray value. Therefore, the relation 304 connecting the 16 Gray 'values with a straight line is a theoretical characteristic of the output Gray' value with respect to an arbitrary input Gray value.
[0008]
However, actually, the output Gray 'value is discrete 8-bit data, so that the output has a step-like characteristic like 305 due to a quantization error, and the gradation is lost.
[0009]
As a method for solving this problem, it is conceivable to increase the number of output bits of the 3D-LUT. For example, if the output R'G'B 'at a lattice point is provided as 10-bit data, the above-described quantization error can be reduced.
[0010]
However, as a problem of this method, it is necessary to increase the total number of processing bits in the circuits subsequent to the input color conversion means 202 to 10 bits, which leads to an increase in circuit scale.
[0011]
In view of the above problems, an object of the present invention is to provide an image processing apparatus and an image processing method that perform color conversion with excellent gradation while preventing an increase in circuit scale.
[0012]
[Means for Solving the Problems]
An object of the present invention is to provide an image processing apparatus for color-converting N-bit color data 1 into N-bit color data 2, wherein the color data 1 is converted into M-bit high gradation color data where M> N. And a multi-value dither processing means for converting the high gradation color data into the color data 2 by a multi-value dither method.
[0013]
Further, the above object is to provide an image processing apparatus for color-converting N-bit color data 1 into N-bit color data 2, wherein the color data 1 is converted into M-bit high gradation color data of M> N. Means for holding a plurality of 1D-LUTs that are M-bit input-N-bit output, and means for selecting one 1D-LUT from the plurality of 1D-LUTs based on the pixel position of the color data 1 And means for converting the high gradation color data into the color data 2 using the selected 1D-LUT.
[0014]
That is, the technical contents of the present invention can solve the above problem by providing the following configuration.
[0015]
(1) In an image processing apparatus for color-converting N-bit color data 1 into N-bit color data 2, color conversion means for converting the color data 1 into M-bit high gradation color data where M> N, An image processing apparatus comprising: a multi-value dither processing unit that converts the high gradation color data into the color data 2 by a multi-value dither method.
[0016]
(2) The image processing apparatus according to (1), wherein the color conversion unit uses a three-dimensional LUT.
[0017]
(3) The image processing apparatus according to (2), wherein the three-dimensional LUT holds a value at each grid point as M-bit data.
[0018]
(4) The three-dimensional LUT holds a value at each grid point as N-bit data, and converts the N-bit data into M-bit data when performing color conversion. Image processing device.
[0019]
(5) In an image processing apparatus for color-converting N-bit color data 1 into N-bit color data 2, color conversion means for converting the color data 1 into M-bit high gradation color data, where M> N, Means for holding a plurality of 1D-LUTs that are M-bit input-N-bit output; means for selecting one 1D-LUT from the plurality of 1D-LUTs based on a pixel position of the color data 1; Means for converting the high gradation color data into the color data 2 using the 1D-LUT obtained.
[0020]
(6) In an image processing method for performing color conversion of N-bit color data 1 into N-bit color data 2, a color conversion step of converting the color data 1 into M-bit high gradation color data where M>N; A multi-value dither processing step of converting the high gradation color data into the color data 2 by a multi-value dither method.
[0021]
(7) The image processing method according to (6), wherein the color conversion step uses a three-dimensional LUT.
[0022]
(8) The image processing method according to (7), wherein the three-dimensional LUT holds a value at each grid point as M-bit data.
[0023]
(9) The three-dimensional LUT holds a value at each grid point as N-bit data, and converts the N-bit data into M-bit data when performing color conversion. Image processing method.
[0024]
(10) In an image processing method for performing color conversion of N-bit color data 1 into N-bit color data 2, a color conversion step of converting the color data 1 into M-bit high gradation color data where M>N; Holding a plurality of 1D-LUTs that are M-bit input-N-bit output; selecting one 1D-LUT from the plurality of 1D-LUTs based on a pixel position of the color data 1; Converting the high gradation color data into the color data 2 using the 1D-LUT obtained.
[0025]
BEST MODE FOR CARRYING OUT THE INVENTION
Hereinafter, embodiments according to the present invention will be described in detail with reference to the drawings.
[0026]
FIG. 1 is a block diagram showing an embodiment according to the present invention. The input color conversion means 102 is composed of two parts: a high gradation input color conversion means 106 having an input of 8 bits and an output of 10 bits, and a multi-value dither processing means 107 having an input of 10 bits and an output of 8 bits.
[0027]
First, the high gradation input color conversion means converts an 8-bit input RGB value into a 10-bit rgb value which is high gradation color data by using a preset 8-bit-10-bit 3D-LUT (108). . Here, the color conversion processing by the 3D-LUT will be briefly described with reference to FIG.
[0028]
First, eight grid points (502) surrounding the input RGB values (501) are obtained, and the respective grid point positions are set to R0G0B0 to R1G1B1. Assuming that output values at those grid points are rgb ₀₀₀ to rgb ₁₁₁ , an output value rgb for an arbitrary input RGB value is obtained by tetrahedral interpolation as follows.
_{rgb = rgb 000 + c1 * ΔR} / (R1-R0) + c2 * ΔG / (G1-G0) + c3 * ΔB / (B1-B0)
Here, ΔR = R−R0, ΔG = G−G0, and ΔB = B−B0.
[0029]
Here, c1, c2, and c3 are classified into the following six cases.
_{(1) ΔR>ΔG> c1} = rgb 100 -rgb when _{ΔB 000, c2 = rgb 110 -rgb} 100, c3 = rgb 111 -rgb 110
(2) ΔR>ΔB> when _c1 = _rgb 100 -rgb ₀₀₀ of _{ΔG, c2 = rgb 111 -rgb 101} , c3 = rgb 101 -rgb 100
(3) When ΔB>ΔR> ΔG c1 = rgb ₁₀₁ −rgb ₀₀₁ , c2 = rgb ₁₁₁ −rgb ₁₀₁ , c3 = rgb ₀₀₁ −rgb ₀₀₀
(4) when _{ΔG>ΔR> ΔB c1 = rgb} 110 -rgb 010, c2 = rgb 010 -rgb 000, c3 = rgb 111 -rgb 110
(5) When ΔG>ΔB> ΔR, c1 = rgb ₁₁₁ −rgb ₀₁₁ , c2 = rgb ₀₁₀ −rgb ₀₀₀ , c3 = rgb ₀₁₁ −rgb _{010
(6) ΔB>ΔG> c1} = rgb 111 -rgb when _{ΔR 011, c2 = rgb 011 -rgb} 001, c3 = rgb 001 -rgb 000
Then, the obtained 10-bit high gradation color data rgb is converted into 8-bit R'G'B 'data by the multi-value dithering means 107.
[0030]
Here, the multi-value dither processing will be briefly described. When 10-bit data is converted to 8-bit data by the multi-value dither method as in the present embodiment, as shown in FIG. 4, 255 dither patterns 400 having a size of 2 × 2 are prepared. In the figure, DITHER [k]; k = 1 to 255, which is filled with threshold data of 1 to 1023. Then, by comparing the input value of 10 bits with the 255 threshold values at the dither position corresponding to the pixel position (x, y), the output level of 0 to 255, that is, the output value of 8 bits is obtained. Is determined.
[0031]
Further, the 2 × 2 dither pattern is repeatedly developed in the X and Y directions according to the size of the input image, so that the above multi-value dither processing is performed on the entire image area.
[0032]
As a result, the gradation of 10-bit rgb is reduced to 8-bit R'G'B 'by the multi-value dither processing. In this case, the gradation information corresponding to 10 bits is R'G'. It is stored in B '. This can be expressed as 4 gray scales as area gray scales using a 2 × 2 dither pattern, and since they have 256 output levels in the depth direction, a total of 4 × 256 = 1024 gray scales, that is, a pseudo equivalent to 10 bits. This is because of the typical gradation expression.
[0033]
As described above, first, a desired color conversion is performed by increasing the number of bits of the high gradation color data rgb so that the quantization error can be ignored, and then, the number of bits is restored by multi-value dither processing. Thus, image processing with excellent gradation can be performed.
[0034]
In the present embodiment, since the number of output bits of the input color conversion means 102 remains at 8 bits, it is not necessary to increase the number of processing bits of the subsequent processing circuit, thereby preventing an increase in circuit scale. Also, by using multi-value dither processing, it can be realized at a very high speed.
[0035]
Also, in the above embodiment, a high gradation 3D-LUT (108) of 8 bits-10 bits is prepared in advance, but in another embodiment, a normal 8 bit-8 bit 3D-LUT is prepared. Even if the number of operation bits is increased during the execution of the interpolation processing and the 10-bit high gradation color data rgb is calculated, the quantization error can be reduced. FIG. 6 shows this state.
[0036]
First, an 8-bit-8-bit 3D-LUT (608) is prepared. Then, at the time of the arithmetic processing by the tetrahedral interpolation, the arithmetic bit number expanding means 609 multiplies the grid point data by four as follows to obtain 10-bit high gradation color data rgb.
rgb = 4 * rgb ₀₀₀ + (4 * c1 * ΔR) / (R1-R0) + (4 * c2 * ΔG) / (G1-G0) + (4 * c3 * ΔB) / (B1-B0)
As a result, it is possible to reduce a truncation error due to the division of the above equation. Then, multi-level dither processing is performed on the rgb10 bits in the same manner as in the previous embodiment, and 8-bit R′G′B ′ can be obtained.
[0037]
In another embodiment, a similar effect can be achieved by using a plurality of 1D-LUTs instead of the multi-value dither processing. This will be described with reference to FIG.
[0038]
First, similarly to the previous embodiment, rgb 10-bit data is obtained by the 3D-LUT. Next, one 4D-LUT (709) of 10 bits input and 8 bits output is set in advance by a 1D-LUT selecting means 708 based on the position information (x, y) of the processing pixel. The LUT is selected. Then, using the selected 1D-LUT, the 1D-LUT processing means 707 converts rgb10 bits into R'G'B'8 bits.
[0039]
Next, a method for creating these four 1D-LUTs (709) will be described. In the multi-level dither method shown in FIG. 4, pixel positions (x, y) = (2n, 2m), (2n + 1, 2m), (2n, 2m + 1), (2n + 1) when n and m are arbitrary integers. , 2m + 1), the value of the multi-bit dither output 8 bits for all input 10 bits can be calculated in advance for each of the four cases. If these are set as a 10-bit-8-bit 1D-LUT, the same result as the above-described multi-value dither processing can be obtained.
[0040]
As described above, according to the present embodiment, by using the multi-value dither method, it is possible to perform color conversion processing with excellent gradation.
[0041]
In the present embodiment, the input color conversion process of RGB-R'G'B 'has been described as an example, but the present invention can be applied to the output color conversion process of RGB-CMYK or a color conversion process similar to them. Needless to say.
[0042]
Further, in the above description, processing using a 3D-LUT has been described as a specific color conversion method. However, the present invention can be easily applied to color conversion processing using masking processing by matrix operation or the like.
[0043]
Further, the object of the present invention is also achieved by supplying a storage medium storing a program code of software for realizing the above-described embodiment to an apparatus, and reading and implementing the program code in the storage medium by a CPU of the apparatus. Needless to say.
[0044]
【The invention's effect】
As described above, according to the present invention, a high-quality image with excellent gradation can be obtained without causing a significant increase in circuit scale.
[Brief description of the drawings]
FIG. 1 is a block diagram of an embodiment using a multi-value dither process according to the present invention.
FIG. 2 is a block diagram including a conventional input color conversion means.
FIG. 3 is a diagram showing a quantization error generated by a conventional input color conversion means.
FIG. 4 is a diagram illustrating details of a multi-value dither process according to the embodiment;
FIG. 5 is a diagram schematically illustrating a color conversion process using a 3D-LUT according to the embodiment.
FIG. 6 is a block diagram of another embodiment according to the present invention.
FIG. 7 is a block diagram of an embodiment using a 1D-LUT capable of obtaining a result equivalent to the multi-value dither according to the present invention.
[Explanation of symbols]
102 input color conversion means 106 high gradation input color conversion means 107 multi-value dither processing means 108 high gradation 3D-LUT
400 Multi-level dither pattern

Claims (10)

In an image processing apparatus for color-converting N-bit color data 1 into N-bit color data 2,
Color conversion means for converting the color data 1 into M-bit high gradation color data in which M>N;
Multi-value dither processing means for converting the high gradation color data into the color data 2 by a multi-value dither method;
An image processing apparatus comprising: The image processing apparatus according to claim 1, wherein the color conversion unit uses a three-dimensional LUT. 3. The image processing apparatus according to claim 2, wherein the three-dimensional LUT holds a value at each grid point as M-bit data. The image processing apparatus according to claim 2, wherein the three-dimensional LUT holds a value at each grid point as N-bit data, and performs bit conversion of the N-bit data into M-bit data when performing color conversion. In an image processing apparatus for color-converting N-bit color data 1 into N-bit color data 2,
Color conversion means for converting the color data 1 into M-bit high gradation color data in which M>N;
Means for holding a plurality of 1D-LUTs that are M-bit input-N-bit output;
Means for selecting one 1D-LUT from the plurality of 1D-LUTs based on a pixel position of the color data 1;
Means for converting the high gradation color data into the color data 2 using the selected 1D-LUT;
An image processing apparatus comprising: In an image processing method for performing color conversion of N-bit color data 1 into N-bit color data 2,
A color conversion step of converting the color data 1 into M-bit high gradation color data where M>N;
A multi-value dither processing step of converting the high gradation color data into the color data 2 by a multi-value dither method;
An image processing method comprising: The image processing method according to claim 6, wherein the color conversion step uses a three-dimensional LUT. The image processing method according to claim 7, wherein the three-dimensional LUT holds a value at each grid point as M-bit data. 8. The image processing method according to claim 7, wherein the three-dimensional LUT holds a value at each grid point as N-bit data, and converts the N-bit data into M-bit data when performing color conversion. In an image processing method for performing color conversion of N-bit color data 1 into N-bit color data 2,
A color conversion step of converting the color data 1 into M-bit high gradation color data where M>N;
Holding a plurality of 1D-LUTs that are M-bit input-N-bit output;
Selecting one 1D-LUT from the plurality of 1D-LUTs based on the pixel position of the color data 1;
Converting the high gradation color data into the color data 2 using the selected 1D-LUT;
An image processing method comprising:

Images