JPH0716257B2 - Color image correction method - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION [Industrial field of application] The present invention is intended to reproduce colors on a color CRT on photographic paper in an apparently larger color than an area originally defined as a color reproduction range. The present invention relates to a color image correction method capable of creating a reproduction range.

BACKGROUND OF THE INVENTION When a TV image signal is hard-copied using a video printer, a digital color copying apparatus, etc., the color systems are different from each other. In many cases, a color separation image correction device is used for color correction such as.

For example, as is well known, a color masking device, which is one of color separation image correction devices, cancels a sub-absorption of a color material (toner, ink, thermal transfer ink, photographic paper, or other coloring matter) to obtain a correct color ( This is a device for reproducing intermediate colors).

That is, a television image is a color image formed by the color-addition method, and its color system uses the R, G, B coordinate system of the phosphor. On the other hand, a photographic paper or the like forms a color image by the subtractive color method, and its color system uses a color system such as L ^* , u ^* , v ^* . This is because in such a case, it is necessary to convert the signal data (color correction) between these color systems.

For example, in the color masking device 10 shown in FIG. 29, new image data (image data after color correction, in this example, by the numerical operation of the input three primary color image data of R, G, B, Cyan C, magenta M, and yellow Y) are formed, and a color image is recorded based on the new image data C, M, Y.

In the figure, 11 is a television receiver, 12 is a color printer, and 13 is a recording medium such as photographic paper.

If you can accurately grasp the color characteristics of color printers,
Basic colors that reproduce a specified color (3 colors or 4 colors
Since the color) combination can be accurately obtained, the color conversion error becomes small and the color reproducibility is remarkably improved.

Basic colors that reproduce a specified color (3 colors or 4 colors
The following two methods are conventionally known as methods for calculating a combination of colors.

If you want to make a hard copy using photographic paper,
As shown in the figure, the spectral absorption density of each single color (Y, M, C) is measured, and the total absorption characteristic is calculated using the density additivity. After that, it is converted into a color system such as X, Y, Z, L ^* , u ^* , v ^* .

The density additivity refers to a method in which the density of each color at each spectral absorption density is added and calculated.

In printing and the like, the combination of basic colors is estimated by the Neugebauer equation.

[Problems to be Solved by the Invention] By the way, among the above calculation methods, in the case of photographic paper, the density additivity does not hold in an actual system. Therefore, the accuracy when estimating the color reproducibility is poor.

Even when the Neugebauer equation is used, since this is an approximate expression, there is a large difference between the approximate expression and the actual value, and the accuracy of color reproducibility is also insufficient.

Further, for example, the color reproduction range of the output system such as photographic paper and the color reproduction range of the input system such as color CRT are the same.
When expressed in a color system such as L ^* , u ^* , and v ^* , the color reproduction range for this color system is generally different. In the case of the input / output system as described above, the color reproduction range of the output color system is narrower than the color reproduction range of the input system.

FIG. 31 shows the relationship between the printed matter and the photographic paper. The curve L1 is the color reproduction range of the printed matter, and the curve L2 is the color reproduction range of the photographic paper.

In this way, if there is no such part in the color gamut of the output system, such as a part with high lightness and saturation in the color gamut of the input system (* 1 area in Figure 31), Cannot express the color of.

Therefore, for example, the saturation is compressed. But,
This makes it impossible to reproduce a clear image.

Moreover, since the color correction data relating to the color image information that exceeds the color reproduction range of the output color system was appropriately compressed within the color reproduction range, the hue, saturation, or lightness are different from the human visual characteristics. It was changing in a different direction and it was very unnatural.

Therefore, the present invention was developed in order to solve such a conventional problem, has a small color correction error, significantly improves color reproducibility, and improves the color reproduction range of the output color system. It proposes a color image correction method that is apparently expanded to improve color reproducibility.

[Technical Means for Solving Problems] In order to solve the above problems, in the color image correction method according to the present invention, the expression range of the saturation and lightness of the input system is the saturation of the output system. When the brightness is larger than the reproduction range of the brightness, the output system level corresponding to the white background level of the input system is set to be darker than the white background level of the output system.

[Action] It is nothing but the judgment that the human eye feels fresh because it is judged to be relatively fresh compared with the surrounding bright parts.

For example, with the yellow paper placed on a white background as shown in FIG. 32A and the same yellow paper placed in black as shown in FIG. 32B, the latter looks much clearer. This is because it is made up of a relative comparison in which the freshness of yellow paper is felt compared to the surrounding area.

Therefore, when the expression range of the saturation and brightness of the input system is larger than the reproduction range of the saturation and brightness of the output system, the level of the output system corresponding to the white background level of the input system is set darker than the white background level of the output system. .

In this way, from the above-described relative color contrast sensation, it is possible to make the image feel fresher if it is set slightly darker than the original brightness of the reference white surface of the image expression medium.

This method is applied to the color image correction process for correcting the difference in color system. The color image is corrected by the following process.

Firstly, there is a step (first step) of preliminarily outputting a plurality of samples having a color tone close to the desired intermediate color.

Secondly, there is a step (second step) of checking the mixing amount of the basic color of each sample and the value of the color system for it.

Thirdly, there is a step (third step) of determining whether or not the target value T'exists outside the color reproduction range of the output color system.

By these three steps, the mixing amount of the basic color that reproduces the intermediate color is calculated.

The sample in the first step is obtained as follows.

It is actually printed on photographic paper by a signal of discrete n points (the total is n.n.n points) about a basic color composed of a specific color system, for example, Y, M, C coordinate system. and color prints, n ^m pieces of color patch image is formed.

Actually measure the colors of multiple color patches formed by color printing, and plot the measured data on the color system of photographic paper (for example, L ^* , u ^* , v ^* color system, and so on). As a result, the colors in the Y, M, C coordinate systems are mapped as the values in the L ^* , u ^* , v ^* color system. These mapped values become sample values.

In order to convert the measurement data into the value of a specific color system, a specific conversion formula for the color system is used. The value obtained by changing the brightness of the reference white surface described above is used as the value of the brightness of the reference white surface in this conversion formula.

In the third step, the sample value closest to the intermediate color is calculated by converging the sample values while sequentially interpolating them. The converged sample value is used as the basic color mixture amount (Y, M, C
Data of each color).

A plurality of these mixing amounts are prepared as color correction data, and these are referred to by the input color information.

In the color separation image correction device, these color correction data are tabulated, and the corresponding color correction data is referred to by the input color separation image information.

As a result, a color image is recorded based on the corrected color separation image information.

By applying the above-mentioned means for apparently changing the color reproduction range, for example, a color having a high lightness and saturation of a color CRT can be expressed on a photographic paper which is supposed to be impossible to express.

In addition, yellow with high brightness and saturation of printed matter can be expressed on photographic paper or color CRT.

[Embodiment] Next, an example of a color image correction method according to the present invention will be described in detail with reference to FIG.

The principle of the present invention will be described.

As described above, the fact that the color looks fresh means that the color saturation C is expressed by the following relationship in colorimetry.

L ^* ＝ 116 (Y / Y _o ) ^1/3 −16 where Y; brightness of input system u, v; coordinates of chromaticity diagram of input system when lightness is ignored Y _o ; reference in output system Brightness of white surface u _o , v _o ; Chromaticity diagram coordinate Y _o of the reference white surface in the output system is slightly darker than the original brightness of the reference white surface.
As a numerical value, it is set to a value smaller than normal (= 100).

What should be noted in the above equation is that the term of lightness L ^* is multiplied in calculating the saturation C. This shows that when the value Y _o is set small by lowering the level of the reference white surface, the lightness L ^* increases, and as a result, the saturation C also increases.

Here, the lightness L ^* in the color system of L ^* , u ^* , and v ^* is represented by the above-mentioned formula, and the relationship between the other hue u ^* and the saturation v ^* is described below.

u ^* = L ^* (u−u _o ) v ^* = L ^* (v−v _o ) By applying the above principle, it is possible to apparently reproduce lightness and saturation that cannot be normally represented.

A specific example is shown.

[Specific Example 1] The printing color is represented by photographic paper (see FIG. 1).

FIG. 1 is a characteristic diagram showing the relationship between saturation and brightness, which is a curve
L1 represents a color reproduction range of a printed matter projected on a surface formed by lightness and saturation.

On the other hand, the curve L2 is the color reproduction range when the white background of the photographic paper is the reference white surface.

As is clear by comparing the two, the color reproduction range of the printed matter is wider in the part of the color (* 1) with high lightness and saturation,
This color cannot be displayed on photographic paper as it is.

Therefore, in the present invention, the reference white surface of the printing paper is set darker than the original reference white surface of the white background. In other words, reference white surface = reference white surface + D, where D is set as When the color reproduction range when the density D is selected to be about 0.15, it is as shown by the curve L3 in FIG.

According to the curve L3, the brightness exceeds 100, but the shaded area * 2 is not used in principle.

As a result, the color reproduction range of photographic paper apparently expands,
Most of the color reproduction range of printed matter can be covered.
After making level adjustments like this, the colors created by printing
By mapping to Y, M, C of photographic paper showing L ^* , u ^* , v ^* , the final correction value of the color image is obtained.

Here, the mapping to Y, M and C results in the movement of the color as shown by the arrow in FIG. As a result, the white color of the printed matter on the photographic paper is a little white (that is,
It will be light gray).

The processing of the first specific example is the same when a color CRT is used as the input system and photographic paper is used as the output system.

In this way, when a system such as a printed matter or a color CRT that can produce colors with higher brightness and saturation than a white background is shown on photographic paper, for example, when reproducing a color image in a color proof or a color hard copy system, This means is particularly effective.

If there is a white background edge other than the screen area of the hard copy output using this method, cut it off or mask it with an appropriate saturation such as black or ash to show the original white background. It is even more effective if devised so that it does not exist.

If the density D for changing the brightness of the reference white surface is too high, it becomes unnatural, so D ≦ 0.3 is preferable. However, this limitation is not imposed on a transparent original such as a slide (transparent slide).

[Specific Example 2] As a next specific example, a case where a printed matter is reproduced on a color CRT will be described.

A curve L4 in FIG. 3 shows a color reproduction range of a printed matter projected on a surface formed by lightness and saturation.

On the other hand, the curve L5 is the color reproduction range when the brightest white of the color CRT is the reference white surface.

Looking at these two ranges, the color reproduction range of the printed matter is wider in the color part (* 3) having high brightness and saturation and the part (* 4) having high saturation and low brightness.

Therefore, as described above, the reference white surface is set to 1/2 the brightness of the brightest white of the color CRT, and the color C at that time is set.
The calculated color gamut of RT is curve L6. In the figure,
The shaded area * 5 has L ^* > 100, but this area is not used in principle.

In the case of the curve L6, the color reproduction range of the color CRT has a high lightness and high saturation, and the color reproduction range of the printed matter can be almost covered.

After performing such level adjustment, the color created by printing
A corrected color image is calculated by mapping onto B, G, R of a color CRT indicating L ^* , u ^* , v ^* .

The mapping to the color CRT involves the movement of colors as shown by the arrow in FIG.

It should be noted that it is better to map the area * 4 by changing the lightness or the saturation or both without changing the hue to a system in which the reference white surface is changed. An example in which both the lightness and the saturation are changed is shown in FIGS. 18 to 20.

The means shown in the specific example 2 is suitable for application to a system that monitors the color of a printed matter by a color CRT.

In this case, unlike the first specific example, the color CRT is a light-emitting body, and thus the reference white surface may be in any dark state if the environment in which the color CRT is observed is dark. It suffices to set the reference white surface to the brightness at which the input system can be reproduced. However, in reality, the limit is a brightness level that is about 1/3 lower than the normal brightness level.

Next, the basic principle of the color-separated image correction method required for mapping onto photographic paper or a color CRT will be described below.

For convenience of explanation, a case where the basic colors are three colors of Y, M, and C will be described.

An intermediate color on a recording medium, for example, a color photographic light-sensitive material such as photographic paper can be expressed innumerably by combining the densities of Y, M and C, but the expression range is three-dimensionally shown. When expressed in the Y, M, C coordinate system, the expression range is a cube as shown in FIG. The coordinate system of Y, M, C is changed to another color system, for example, X, Y, Z
When converted to the color system, a solid as shown in FIG. 6 is obtained.

In the figure, each vertex A to H corresponds to A'to H '.

As is clear from FIG. 6, most of the solids that determine the expression range are distorted, and each side is not always a straight line, and each side is a complicated curved surface.

In this solid, by a proper combination of Y, M, C,
A given intermediate color can be reproduced. Therefore, it is necessary to form the color correction data so as to be included in this solid.

For simplicity, the basic colors will be described as two colors (for example, Y and M), and then the algorithm using the original basic colors will be described.

The color level of each intersection (5 × 5 = 25 grid points in the embodiment) of FIG. 7 is supplied to a color printer, and the color level is recorded on a recording medium (hereinafter referred to as photographic paper) to print the color. Form a patch.

While measuring the actual color from the obtained color patch,
Convert the measured values into color system values (sample values) using the color system conversion formula (calculation formula of L ^* , u ^* , v ^* ) described above.
This is plotted in FIG. 8 for each grid point.

Although the number of color patches is large, it takes a long time to actually measure the color. Therefore, in the embodiment, about 5 × 5 = 25 color patches are used.

More color patches may be used. In that case, the number of color patches may actually be increased, but the number of color patches can also be increased by interpolation processing. For example, when the middle of a color patch of 5 × 5 = 25 is interpolated, this interpolation processing expands and propagates to a color patch of 9 × 9 = 81.

In the example shown below, the combination of basic colors is estimated by 5 × 5 = 25 color patches.

Although the above-described color patch setting does not consider human perception characteristics (discrimination ability), the number of color patches may be selected in consideration of human perception characteristics as described later.

Here, as shown in FIG. 8, when a certain intermediate color is indicated by x (target value T ′), the combination of the Y and M coordinate systems indicating this color is represented by the grid points a to d in FIG. Presumed to be within the enclosed area.

The calculation process of which grid point is the closest is obtained by converging the color system of FIG. 4 in correspondence with the coordinate system of FIG.

As described above, the reason why the convergence calculation is performed using only the color system of FIG. 8 and the convergence result is not estimated in association with the coordinate system of FIG. 7 is that the coordinate system of FIG. This is because, although the conversion for the color system is known, the inverse conversion operation is very complicated, and its preferable conversion formula is not known yet.

For this reason, the target value T shown in the coordinate system of FIG. 7 is to be estimated by the following processing. The estimation processing operation will be described in detail with reference to FIGS. 9 and 10.

First, using the target value T'and a total of 25 basic grid points (see FIG. 10), the grid point closest to the target value T'is calculated.

In practice, the grid point that minimizes the difference between the two is calculated.
If this grid point is b ', it can be estimated that the target value T is close to the grid point b corresponding to the grid point b'in FIG. 9 as well.

Next, at the level interval at which the grid point interval is 1/2, the grid point b
A total of 9 grid points surrounding the grid point are set, and these grid points are calculated by the weighted average of the surrounding grid points. For example, it is calculated by weighting the surrounding 2 or 4 lattice points.

The values corresponding to the newly calculated grid points e to 1 are plotted again in the color system of FIG.

Then, the plotted grid points e'to 1 '(total 9
The grid point closest to the target value T ′ is obtained by the same method as described above, and the grid point (in this example,
The grid point h in FIG. 9 corresponding to h ′) and the eight grid points m to t including the grid point h are calculated by further narrowing the grid spacing to 1/2. By repeating such division of the lattice, the lattice becomes gradually narrower and finally converges. The target value T of FIG. 9 corresponding to the value of this converged grid point is
It indicates the combination of basic colors (mixing amount of Y, M, C) for reproducing the intermediate color.

The above estimation operation is executed for each given target value.
The estimated target values may be made into a table, and the target values may be referred to by the values of the input color separation image.

When actually applied to a color separation image correction device, etc.,
You will be using a lookup table. An example will be described later.

Next, the algorithm when three colors are used as the basic colors will be described.

It is assumed that each of Y, M, and C has a level of 256 steps from 0 to 255. 5 of these levels in this example
One level is extracted. For example, five levels of 0, 64, 128, 192, and 255 are extracted for each of Y, M, and C.
Color patches of all these combinations of colors (5 × 5 × 5 = 125) are created.

An example of a color patch is shown in FIG. As the color system of each color patch, the CIE L ^* , u ^* , v ^* , L ^* , a ^* , b ^* color system is suitable.

The number of color patches is not always the same for each color. That is, when a color patch is constructed in consideration of the discrimination ability of human eyes, generally, each color does not have the same number of levels. Because M (magenta) has the highest discrimination ability of human eyes and Y (yellow) has the lowest discrimination ability,
This is because it is conceivable to make Y smaller and M larger in color patches.

FIG. 12 shows an example thereof, and illustrates the case where Y · M · C = 3 × 5 × 4.

As described above, when the color patch is created in consideration of the discriminating ability of the eyes, almost the same result can be obtained as when the color patch is formed by dividing Y, M, and C at equal intervals.

As a result, the number of color patches is reduced, and the actual measurement time for the corresponding colors is shortened.

If these relationships are generalized, the patch numbers PY, PM, and PC of Y, M, and C may be set so as to satisfy the following relationship.

PY <PC ≤ PM To increase the number of color patches by interpolation processing, do the following.

As a basic lattice, when 5 × 5 × 5 = 125, L ^* , u ^* , v ^*
The color system is interpolated by curve interpolation shown in the following calculation formula.

In this case, as shown in FIG. 13, when the black circles ● are grid points (sample points), if the △ and × marks are points to be interpolated, two grid points exist before and after like the △ marks. When doing, and when there are 1 point and 3 points before and after like the × mark,
Different interpolation formulas are used.

When the color system of the points to be interpolated is L _m ^* , u _m ^* , v _m ^*, and the color system of each sample point is Li ^* , ui ^* , vi ^* (i = 1 to 4) In the former case, interpolation is performed by the following interpolation formula.

L _m ^* =-(1/16) L1 ^* + (9/16) L2 ^* + (9/16) L3 ^* -(1 /
16) L4 ^* u _m ^* =-(1/16) u1 ^* + (9/16) u2 ^* + (9/16) u3 ^* -(1 /
16) u4 ^* v _m ^* =-(1/16) v1 ^* + (9/16) v2 ^* + (9/16) v3 ^* -(1 /
16) v4 ^{* In the} latter case, the following interpolation formula is used.

L _m ^* = (5/16) L1 ^* + (15/16) L2 ^* + (15/16) L3 ^* -(1 /
16) L4 ^* u _m ^* = (5/16) u1 ^* + (15/16) u2 ^* + (5/16) u3 ^* − (1/1
6) u4 ^* v _m ^* = (5/16) v1 ^* + (15/16) v2 ^* + (5/16) v3 ^* − (1/1
6) Fig. 14 shows an example of the v4 ^* interpolation processing sequence. The numbers I, II, and III are interpolated in this order.

By such an interpolation process, the number of color patches can be expanded to 729 by electrical processing, even though only 125 color patches are actually measured, and the Y, M, C coordinate system at that time is used. The color patches shown are as shown in Fig. 15.

Figure 16 maps this to the L ^* , u ^* , and v ^* color system.

FIG. A is the color system seen from the top of FIG.
Is a mapping on the L ^* , v ^* plane side, and FIG. 6C is a mapping on the L ^* , u ^* plane side.

The target value T estimation process described above is executed using a total of 729 color patches created by such an interpolation process.

Here, in order to calculate which lattice point the target value T ′ is close to, the following evaluation function ΔE may be used.

_{^{ΔE = | L T * -Li *}} | + | u T * -ui * | + | v T * -vi * | ΔE = [(L T * -Li *) 2 + (u T * -ui *) 2 ＋ (v _T ^* −vi ^* )
² ] It does not matter which one of the _1/2 evaluation functions is used.

When all the final target values T are calculated by the convergence calculation processing, the processing time is too long.
In practice, an algorithm is adopted that ends in several convergence processes.

For example, when the interval of the basic lattice is divided by 64 quantization steps as in the above example, the lattice interval (division interval) becomes 32 by the above-described interpolation processing. , Then the grid spacing is 16,
The algorithm is such that the processing is completed by sequentially repeating the processing of a total of 5 times of 8,4,2,1.

As a result, the target value can be estimated with sufficient accuracy.

When the color patch as shown in FIG. 15 is obtained by the interpolation processing, the calculation of the interpolation point (each vertex of the solid) is performed as described above in the first to fifth convergence processing. Although it can be calculated by a curved approximation, the following examples are all cases of linear approximation.

Interpolation processing by linear approximation is as follows.

Assuming an interpolation point s as shown in FIG. 17, assuming that L ^* , u ^* , v ^* color system of the interpolation point s is Ls ^* , us ^* , vs ^* , An example of the formula is shown below.

The interpolated color system Ls ^* , us ^* , vs ^* is associated with the values of the Y, M, C coordinate system.

Mi is the volume of a rectangular parallelepiped including the diagonal vertices and the interpolation point s. In the case of FIG. 15, Becomes A specific example of the interpolation will be described later in the color-divided image estimation device.

By the way, in the above, the estimation process when the target value T is inside the solid shown in FIG. 8 has been described. However, even if the reference white surface is changed and the color reproduction range is apparently increased, it is still 18th. As shown in the figure, the input system color may exist outside the solid body.

In such a case, it is estimated by the following processing. The Y, M, and C coordinate systems are not used to simplify the explanation.

The target value T'exists outside the solid because the color reproduction range of the output system is narrower than the color reproduction range of the input system.

In this case, the hue of the color is not changed, the color is moved in the achromatic direction, and the color at the point where the straight line 1 in the achromatic direction intersects the boundary of the color reproduction range is used as the target value T ^*. It is something to do.

Also in this case, the target value T ^* is considered to be on the line connecting the grid points q1 and q2 in FIG. 7, and Y,
It is estimated by dividing and converging q1 'and q2' (Fig. 18) while associating them with the M and C coordinate systems.

As the algorithm of this estimation operation, the following algorithm is added to the above algorithm.

First, when any of Y, M, and C is O or maximum, it is determined that the target value T'is outside the solid, that is, outside the color reproduction range.

In that case, as shown in FIG. 19, a straight line passing through the target value T ′ and the achromatic axis (this is one point of the L ^* axis) is assumed, and the straight lines (hereinafter referred to as convergence lines) l and u ^*. , Θ with respect to the v ^* plane is expressed as follows.

l = ar + b θ = arc tan (u _T ^* / v _T ^* ) where a and b are arbitrary real numbers, which are different from a and b in FIG. 7.

In the case of setting not to change the lightness in addition to the hue, l = LT ^* .

Next, the cylindrical coordinates (θ, r, l) = (hue, saturation, lightness) of the sample points on the outer surface are calculated and stored in memory.

Then, among the sample points (lattice points indicated by black circles ● in Fig. 20) on the outer surface memorized in this way, the minimum quadrilateral consisting of four sample points is assumed, and their cylindrical coordinates are assumed. Is represented by (θi, ri, li).

It is always checked whether any of the four points satisfies the following conditional expression.

θ ≦ θi ≦ θ + 180 ° (i = 1~4) θ-90 ° = θi ≦ θ + 90 ° (i = 1~4) θ-180 ° ≦ θi ≦ θ (i = 1~4) ar i + b-l i ≧ O (i = 1 to 4) ar _i + b−l _i ≦ O (i = 1 to 4) When these conditions are satisfied, the convergence line 1 passes through the set minimum quadrangle. Probability is high.

It should be noted that such conditional expressions can be considered innumerably, but the conditional expressions described above are examples in which they can be performed by a simple operation.

Next, we take this quadrilateral by the weighted average from its vertices,
The midpoint indicated by a circle in Fig. 20 is obtained, and the outer surface is divided into four.

The above conditional expressions are referred to again for these four surfaces, and thereafter, the same operation is repeated seven times. Then, the average value of the values of the Y, M, and C coordinate systems corresponding to the seventh vertex is set to the target value T.
It is used as an alternative value T ^* of.

Next, an example of a color separation image correction device (color masking device) embodying the above-described color separation image correction method according to the present invention will be described in detail with reference to FIG.

In this example, the target value calculated as described above,
That is, the color correction data is stored in advance in the LUT (lookup table). For example, if the input system is a color CRT, the color correction data of each grid point associated with the basic color coordinate system (coordinate system similar to that in Fig. 15) determined by B, G, and R is stored, and the grid points Color correction data other than is calculated by interpolation.

When there are few input gradations or output gradations, it is possible to store all the color correction data instead of the intermittent color correction data.

The interpolation processing of the corrected color data will be described with reference to FIG.

In this example, a rectangular parallelepiped space W determined by three input image data R, G, B (interpolation point s is located at the diagonal vertex)
A rectangular parallelepiped space region V formed by eight color correction data (known calculation color correction data P1 to P8 corresponding to C, M, and Y) including the above is defined. Each of the spatial regions W and V uses P1 as a reference point.

Then, it is assumed that each color has a color correction value for the color of the combination at each point of 0, 32, 64, 96, 128, 160, 192, 224, 255.

At this time, if the input image data R, G, B each have a value of (100, 130, 150), interpolation is performed using the color correction data of the vertices (lattice points) of the spatial area surrounded by 8 points shown below. Be done.

Here, Pi (i = 1 to 8) on the left side indicates the coordinate value of each vertex of the spatial region V, and the right side indicates the color correction data Ci, Mi,
Indicates Yi.

P1: (96,128,128) = (C1, M1, Y1) P2: (128,128,128) = (C2, M2, Y2) P3: (96,160,128) = (C3, M3, Y3) P4: (128,160,128) = (C4, M4, Y4) P5: (96,128,160) = (C5, M5, Y5) P6: (128,128,160) = (C6, M6, Y6) P7: (96,160,160) = (C7, M7, Y7) P8: (128,160,160) = (C8, Therefore, the relationship between the spatial area V formed by these eight vertices P1 to P8 and the spatial area W formed by the input image data is as shown in FIG.

The weighting coefficient for each vertex Pi of the spatial region V is calculated as follows.

As the method of calculating the weighting coefficient, the same calculation formula as in the above-described L ^* , u ^* , v ^* color system can be used.

In this, the volume of the rectangular parallelepiped spatial region W formed by the interpolation points s and the vertex opposite to the correction value to be obtained is used as the weighting coefficient at the correction value to be obtained.

Therefore, the weighting factor of the point P8 is (100,130,150)-(96,128,128) = (4,2,22) using the coordinates of P1 and the coordinates of s. The volume of is 4 × 2 × 22 = 176, which is the weighting factor for point P8.

Similarly, the weighting factors of the remaining points P1 to P7 are calculated.

P1 = 8400 P2 = 1200 P3 = 560 P4 = 80 P5 = 18480 P6 = 2640 P7 = 1232 P8 = 176 The sum of these weighting factors is the same as the volume of the cubic space region V, and in this example, 32768 (a And).
Therefore, the correction values Cs, Ms, Ys at the s point are Cs = 1 / a (P1C1 + P2C2 + P3C3 + P4C4 + P5C5 + P6C6 + P7C7
+ P8C8) Ms = 1 / a (P1M1 + P2M2 + P3M3 + P4M4 + P5M5 + P6M6 + P7M7
+ P8M8) Ys = 1 / a (P1Y1 + P2Y2 + P3Y3 + P4Y4 + P5Y5 + P6Y6 + P7Y7
+ P8Y8). That is, the correction value of a desired point s and eight points surrounding it is Ci, Mi, Yi (this is the interpolated value Ls ^* , u of the color system).
s ^* , vs ^* are values in the Y, M, C coordinate system corresponding to) and each weighting coefficient is Ai, Can be expressed as

The points of the color correction data described above are examples.

In practice, the number of color correction data is set to a power of 2 in consideration of the ROM capacity and the like. Therefore, 256 kbit RO
When using M, it is possible to have 32 points of color correction data for each color (32 ³ = 32768 points for all three colors).

FIG. 22 shows an example of the color masking device 10.

As is clear from the above-mentioned arithmetic expression, this color masking device 10 includes a color correction information storage means (color correction data storage means) 20 for storing a plurality of color correction data, and a weighting information storage means (weighting coefficient storage means). 24, a multiplication accumulation means 30 for multiplying the referred color correction data and the weighting coefficient, and accumulating the value, and a processing means including a division means. Of these, the dividing means can be omitted depending on the configuration.

The color correction data storage means 20 divides the color space formed by the three-color separated image information that can be input for color correction into a plurality of spatial regions, and corrects the color of the combination of the three-color separated image information located at the vertex. Information is stored.

The weighting coefficient storage means 24 outputs weighting information for each of the plurality of color correction information selected from the color correction information storage means based on the input three-color separated image information.

In the processing means, a plurality of color correction information selected from the color correction data storage means 20 based on the input color separation image information and the corrected color separation image data to be finally obtained are calculated based on the weighting coefficient. Is output.

FIG. 22 shows a case where the present invention is applied to a simultaneous type color masking device which simultaneously obtains three color correction data C, M, Y, and FIG. 28 shows three color correction data C, M, Y. Is a case where the present invention is applied to, for example, a so-called sequential color masking device which is configured to sequentially output in these orders.

Then, the simultaneous color masking device 10 in FIG.
The configuration of each part of will be described.

Reference numeral 20 is a color correction data storage means, and in this example, color correction data for each color C, M, Y is stored in each LUT 21-23. Reference numeral 24 is a weighting coefficient storage means, which is also configured as an LUT.

In the color correction data storage means 20 and the weighting coefficient storage means 24,
Address signals for reading are respectively supplied. Therefore, the input image data B, G, R are temporarily transferred to the address signal forming means 4
It is supplied to 0 and the address signal corresponding to the input level is output. The address signal output means is also composed of LUTs 41 to 43, respectively. A bipolar ROM is suitable as the LUT. A 1-bit distribution signal is further supplied to the LUTs 41 to 43 from the controller 50, the details of which will be described later.

The color correction data referred to by the address signal corresponding to the input level of the input image data and the data indicating the weighting coefficient (hereinafter simply referred to as weighting coefficient) are sequentially supplied to the multiplication accumulating means 30 side eight times in total.

The multiplying and accumulating means 30 is for sequentially executing AiKi (Ki is a general term for C, M, and Y) as described above, and is for obtaining the sum thereof. In this example, the multipliers 34 to 36 are used. Accumulator 37 ~
It consists of 39 and.

Therefore, for each of the multipliers 34 to 36, a 512 Kbit ROM is used, the corresponding color correction data (8 bits) and the weighting coefficient Ai are supplied, and the multiplication process of AiKi is executed.
The higher 8 bits of the multiplication output is the accumulator (AL
U) are supplied to 37 to 39 and the multiplication outputs are sequentially added.

The accumulators 37 to 39 operate with a precision of 16 bits, and the higher 8 bits of them are used as the cumulative calculation power (sum of products output). As a result, the same output as that obtained by dividing the cumulative calculation force by the weighting coefficient Ai is obtained. That is, by doing this, the divider can be omitted.

The cumulative calculation power of the upper 8 bits is latched by the latch circuits 45 to 47, respectively. The latch pulse is generated by the controller 50.

The configuration of each unit will be described in more detail.

The color correction data storage means 20 is for each color C, M, Y as shown in the figure.
LUTs 21 to 23 that store accurate color correction data corresponding to
Is used.

When a ROM with a capacity of 256 Kbits is used as the LUTs 21 to 23, only 32 points are extracted from the minimum level to the maximum level of the input image data. With this, 32 per color
It is possible to store color correction data for points (thus, for 3 colors, 32 ³ = 32768 points).

Therefore, when the input level is 256 gradations, the distribution of 32 points is 0,8,16, ... 240,248 in total, divided into "8" in order from 0 as shown below. And divide it into equal parts to get the 33rd point
Do not use from 249 points to 255 points. Or 249-255
Points are treated as 248.

The color correction data at each distribution point is accurately calculated, and the calculated plurality of color correction data is stored in each LUT2.
It is stored in 1 to 23.

Note that setting the distribution points to 32 points in this manner has the advantage that the storage means 20 can be constructed at a low cost because a general-purpose ROM with 8-bit output can be used.

The weighting coefficient Ai at each distribution point is stored in the LUT 24 for the weighting coefficient storage means. Now, when 8 bits are allocated as described above, the total of the weighting coefficients Ai of 8 times is 8 × 8 × 8 = 512, but as described above, a commercially available general-purpose IC whose output is 8 bits.
If you try to use, the number of elements will increase if you have the weighting factor (up to 512) according to the theoretical value. Therefore, in this example, the approximate value of compressing the theoretical value to about 1/2 is used as the actual value of the weighting factor. It

In the example shown below, the sum of eight weighting factors is set to always be 256, and the maximum weighting factor of each is 255.

In such a case, for example, in FIG. 21, when s is at the same position as P1, the weighting factors of P1 to P8 are P1, P2, P3, P4, P5 as shown by their theoretical values in (). , P6, P7, P8 255,0,0,0,0,0,0,1 (512,0,0,0,0,0,0,0), and the total weighting coefficient is 256.

Further, when s is in the middle of P1 and P3 and is apart from P1 by 3 (thus, 5 from P3), the weighting factors of P1 to P8 are as follows.

P1, P2, P3, P4, P5, P6, P7, P8 160,0,96,0,0,0,0,1 (320,0,192,0,0,0,0,0) Each weighting factor is appropriately selected so that the sum of the weighting factors is also 256.

Similarly, s is separated from the plane of P1 to P4 by 3 and P1, P3, P
If the distance from the surface of 5, P7 is 1 and the distance from the surface of P1, P2, P5, P6 is 5 only, the following weighting factors P1 to P8
Becomes

P1, P2, P3, P4, P5, P6, P7, P8 53,7,88,12,32,4,53,7 (105,15,175,25,63,9,105,15) and the weighting factor in this case Each weighting coefficient is appropriately selected so that the total sum of the above is also 256.

The above-mentioned 1-bit distribution signal is a control signal for designating the color correction data before and after including the point s.

That is, for convenience of explanation, the relationship between 32 allocation points (lattice points) and the address signals corresponding thereto is set as shown in FIG.

Now, when the level of the input image data is 100,
It is necessary to form an address signal (12, 13) from which the color correction data (96 and 104) including this input level are outputted from the color correction data storage means 20.

Therefore, when the distribution signal is 0, the address signal (12) is output so that the smaller color correction data (96) is referred to, and when the distribution signal is 1, the larger color correction data (104) is output. Address signal (1
3) is controlled to be output.

However, when the distribution signal is 0 at the maximum value to be used (248 in this case), the color correction data having its own value is selected, and when the distribution signal is 1, the smaller color correction data ( In this case, select 240).

The distribution signal is supplied to the weighting coefficient storage means 24.

By the way, when a recording medium such as photographic paper is used,
Since each lot has a sensitivity difference, if such a sensitivity difference is taken into consideration, it is necessary to have color correction data capable of correcting a plurality of sensitivity differences for each lot. However, the color correction data storage means according to the sensitivity difference is
Preparing 20 is practically impossible and not realistic.

If the color correction data storage means 20 is commonly used, the color correction data storage means 20 can be realized without any difficulty.

FIG. 24 shows an example of a color masking device 10 suitable for use in such a configuration, and the input image data B, G,
R is once supplied to the color masking device 10 via the LUTs 55 to 57 for input value correction. The color correction data storage means 20 stores color correction data corresponding to one type of sensitivity.

The corrected image data is calculated from the color correction data from the color correction data storage means 20 and the weighting coefficient at that time. The corrected image data is supplied to the sensitivity correction LUTs 61 to 63, and is corrected according to the sensitivity of the printing paper used.

Here, a plurality of types of sensitivity correction values corresponding to different sensitivities are stored in the sensitivity correction LUTs 61 to 63, and the correction values are selected according to the sensitivity of the printing paper to be used.

The visual characteristics of humans are taken into consideration as the input / output characteristics of the sensitivity correction LUT. Then, by using the sensitivity correction curve of the input / output characteristic as shown in FIG. 25, it is possible to minimize the occurrence of the pseudo contour due to the quantization error.

By the way, although the above example is not configured to fully use 256 gradations, a color masking device that fully uses 256 gradations can be realized, for example, by following the idea below. (See Figure 28).

For that purpose, first, 8-bit intervals and 9-bit intervals are distributed in a mixed form as grid points. By adopting the mixed type, an identification signal of 8-bit interval and 9-bit interval is prepared. Therefore, the output of the address signal forming means 40,
The relationship between the grid points and the identification signal is set as shown in FIG.

As a result, for example, when the input is 216, the relationship between the output from the address signal forming means 40 and the output from the controller 50 is controlled as follows.

Distribution signal 0 1 Address signal to 24 6 Address signal to 3 20 26 27 Identification signal 1 1 Here, the value of the address signal to the weighting coefficient storage means 24 is
When the distribution signal is 0, the difference (= 6) from the smallest grid point 210 closest to the input 216 is selected, and when the distribution signal is 1, the difference between the input 216 and the next grid point 219 (= 3). Is selected.

The identification signal 1 represents a grid point having a 9-bit interval, and 0 represents a grid point having an 8-bit interval. The identification signal is required for the following reason.

That is, if the intervals of the grid points are different, the spatial region created by the grid points of three colors is not a cube but a rectangular parallelepiped, and its volume is 512 (= 8 × 8 × 8), 576 (= 8 × 8 × 9). ) 648 (= 8 × 9 × 9) and 729 (= 9 × 9 × 9). Therefore, an identification signal indicating whether one side is 8 bits or 9 bits is required.

Further, in the weighting coefficient storage means 24, the respective weighting coefficients are set in accordance with this identification signal so that the sum thereof is still 256.

For example, when the image data value of each color is (64,143,216), it becomes as shown in FIG.

Therefore, from the weighting factors and the color correction data as shown,
The final color correction data is obtained according to the above calculation formula.

In this way, if the bit interval of the grid points is appropriately selected, 256
Full gradation can be used. However, in this case, it goes without saying that the controller 50 is configured to generate the above-described identification signal.

FIG. 28 shows the case where the present invention is applied to the color masking device 10 configured in a sequential manner, in particular, the configuration in which 256 gradations are fully used, and the portions corresponding to those in FIG. , The description is omitted.

In this example, since the maximum distance between grid points is 9 bits,
As the address signal for referring to the weighting coefficient corresponding to this distance, 4-bit data is generated by the address signal forming means (pre-LU
T) 40 is supplied to the weight coefficient storage means 24 side. The address signal forming means 40 further outputs an identification signal (1 bit configuration) at 8-bit intervals and 9-bit intervals, and this is supplied to the weighting coefficient storage means 24.

LUTs 21 to 23 for color correction data have control terminals ▲
Control signals for selecting chips in ▼ ▲ ▼, ▲
▼, ▲ ▼ are supplied, and the color correction data are sequentially read from the LUTs 21 to 23, for example, and then supplied to the multiplication and accumulating means 30.

Also in the multiplication accumulating means 30, the correction value calculation for each color is sequentially processed.

As the multiplication / accumulation means 30, a multiplication / accumulation device composed of a single chip is used as shown in the figure, and the upper 8 bits of the product-sum output (cumulative calculation power) are sequentially output for each color. .

The controller 50 includes a 9-ary counter 51 and a latch circuit 52 for adjusting the output timing. counter
The reference clock to 51 is the clock input terminal of the multiplier / accumulator 30.
The color correction data Ki and the weighting coefficient Ai, which are commonly supplied to Xck and Yck and are input to the X and Y terminals, are arithmetically processed at the clock timing. Then, the clock that counts down the reference clock to 1/9 is supplied to the Zck terminal so that the final color correction data is output from the output terminal ZOUT at the next timing when the product-sum output for eight times is obtained. It

When the level of the arithmetic processing control pulse supplied to the accumulation terminal ACC is 1, the product-sum processing of X.Y + Q (Q is the immediately preceding product-sum output) is executed. The 0-level control pulse is generated at each timing when the ninth reference clock is obtained, whereby the sum-of-products output is reset and prepared for the next color correction calculation process.

Therefore, at the time of this reset, the weighting coefficient of all 0 is input to the terminal Yin so that the storage means 24 has
A reset signal is supplied to the terminal. As a result, the Yin data becomes 0 due to the pull-down resistor Rp, and XY (=
The reset process of 0) is executed.

The embodiment described above can be modified as follows.

First, in the above description, the final color correction data is calculated from the color correction data of eight lattice points, but it may be interpolated from the color correction data of two points of diagonal vertices. Such an interpolation method is suitable especially when the correction data of more points than the above is used as the color correction data.

Second, although the color correction data is stored in the ROM-configured LUT in the above, RAM is used as the color correction data storage means, and another memory (ROM, disk memory, etc.) is prepared for storing the color correction data. However, it is also possible to read the color correction data from this separate memory when necessary and write it in RAM for use.

According to this configuration, since the S-RAM can be used as the RAM, the calculation processing time can be speeded up.

In such a configuration in which another memory is used and downloaded when necessary, color inversion data, data for selecting a certain output color, data with a color tone changed according to the type of illumination light, and color Data for special effects such as emphasis data can be prepared. Special effects can be created relatively easily by downloading and using these when and as needed.

Thirdly, when the color masking device is applied for printing, it is only necessary to separately prepare an LUT storing black (smear) data in the color correction data storage means 20. In this case, the sequential color masking device is advantageous because it can simplify the configuration.

Fourthly, the weighting factor may be calculated as the reciprocal of the distance from the point Pi (or its nth power) instead of using the volume of the rectangular parallelepiped as the weighting factor.

Fifthly, if a latch circuit is connected between the respective stages of the color correction data storage means 20 and the multiplication and accumulation means 30, the processing between the respective stages can be separated from each other, so that high-speed arithmetic processing can be performed.

Sixth, the conversion of color space coordinates is (B, G, R), (L ^* , a ^* , b
^* ), (X, Y, Z), etc. can be easily understood.

[Effects of the Invention] As described above, according to the present invention, when the saturation and lightness expression range of the input system image is larger than the saturation and lightness reproduction range of the output system, the white background level of the input system is set. Since the level of the corresponding output system is set darker than the white background level of the output system, the color reproduction range of the output system can be apparently expanded.

As a result, it's so bright that you can't express it.
Since it can be expressed with highly saturated colors, it has the characteristic that a clear image can be reproduced.

When expanding the color reproduction range in this way,
Since unnaturalness is not the color shift as in the past, a more natural image can be easily reproduced.

Therefore, the present invention mixes a plurality of basic colors forming a color separation image to reproduce an intermediate color of a certain color system,
The value of the color system close to the target intermediate color set in the color system is sequentially converged while being associated with the value of the coordinate system composed of the basic colors, and the value of the color system closest to the target intermediate color is calculated. It is extremely suitable to be applied to a color separation image correction apparatus that corrects a target value of a coordinate system corresponding to a value as a combination of basic colors that reproduces an intermediate color to be obtained.

[Brief description of drawings]

1 to 4 are diagrams showing the color reproduction range of the input / output system used for the explanation of the present invention, FIG. 5 is an explanatory diagram of the Y, M, C coordinate system, and FIG. 6 is L ^* , u ^* , v ^* Explanatory diagram of color system, Fig. 7 is an explanatory diagram of Y, M coordinate system which is a simplified version of the coordinate system of Fig. 5, and Fig. 8 is a diagram of color system showing lightness and saturation at that time. Explanatory diagrams, FIGS. 9 and 10 are explanatory diagrams of interpolation processing, FIGS. 11 and 12 are diagrams showing an example of color patches, FIG. 13 is an explanatory diagram of curve approximation, and FIG. 15 and 16 are explanatory diagrams of the coordinate system and color system obtained by sample point extension, FIG. 17 is an explanatory diagram of interpolation processing, and FIG. Explanatory drawing when there is a target value in
Figure is a cylindrical coordinate diagram showing the color reproduction range in the color system, Figure 20 is an illustration of the convergence operation, Figure 21 is an illustration of interpolation processing similar to Figure 17, and Figure 22 is a color masking device. Fig. 23 is a systematic diagram of the main part showing an example, Fig. 23 is a diagram showing the distribution relation of grid points, Fig. 24
Fig. Is a schematic system diagram showing another example of the present invention, Fig. 25 is a diagram showing a sensitivity correction curve used at that time, and Figs. 26 and 27 are distribution signals, color correction data, identification signals, etc. Fig. 28 shows the relationship between Fig. 28 and Fig. 28 showing the other example of the present invention.
A system diagram similar to Fig. 22, Fig. 29 is a block diagram of a conventional color separation image correction device, Fig. 30 is a spectral absorption density curve diagram, and Fig. 31 is a printed matter as an input / output system having a different color reproduction range, and a printed image. FIG. 32 is a curve diagram showing the relationship of paper, and FIG. 32 is a diagram showing eye adaptability. 10 …… Color masking device 20 …… Color correction data storage means 30 …… Multiply accumulation means 40 …… Address signal forming means 50 …… Controller

─────────────────────────────────────────────────── ─── Continuation of the front page (51) Int.Cl. ⁶ Identification code Internal reference number FI Technical display location H04N 5/91 7734-5C H04N 5/91 H B41J 3/00 B

Claims (4)

[Claims] 1. When an expression range of saturation and brightness of an image of an input system is larger than a reproduction range of saturation and brightness of an output system,
A color image correction method characterized in that the level of the output system corresponding to the white background level of the input system is set darker than the white background level of the output system. 2. The color image correcting method according to claim 1, wherein a printed matter is used as the input system and a color photographic paper is used as the output system. 3. A color image correction method according to claim 1, wherein a color CRT is used as an output system. 4. A color image correction method according to claim 1, wherein a transmission slide is used as an output system.