JPH01176561A - Color image processing apparatus - Google Patents

Abstract

PURPOSE:To make it possible to reproduce an image faithful to an original thereof, by doing various image processings after an input coordinate system is converted to a coordinate system which meets the human sense, and then converting the coordinate system back to the input coordinate system. CONSTITUTION:For example, an image signal of R, G, B is supplied to a normal converting means 10, so that the input coordinate system R, G, B is converted to a specific colorimetric system. The normal converting means 10 is intended to convert the input coordinate system to a coordinate space fit for the human sensibility. Such colorimetric system as L*, u*, v* is suitable for the coordinate space. The image data L*, u*, v* converted to the colorimetric system is supplied to a signal processing means 2 for a predetermined signal process. The predetermined signal process indicates a gradation compensating proceeding, a color correcting proceeding, an edge proceeding or the like. After the various signal proceedings are conducted, the color displaying system of the image data is converted back to the input coordinate system by a reverse converting means 10'. Thus, the image data Ro, Go, Bo is outputted with the same coordinate system as the input coordinate system.

Description

DETAILED DESCRIPTION OF THE INVENTION [Field of Industrial Application] The present invention relates to a color image processing device suitable for use in color correction of video printers, digital color copies, and the like.

[Background of the Invention] In video printers, digital color copiers, etc., color separation image correction devices are often used for color correction such as color correction.

As is well known, this is a device that cancels the side absorption of coloring materials (toner, ink, thermal transfer ink, pigments of photographic paper, etc.) and reproduces correct colors (intermediate colors). Known color masking devices fall under this category of color separation image modification devices.

Such a color separation image correction apparatus performs various signal processing such as gradation correction, color correction, and edge processing in order to more faithfully reflect the original image.

That is, as shown in FIG. 6, the input signal (R.

Three primary color signals such as G and B, signals satisfying complementary color relationships such as Y, M and C) are supplied to the signal processing means 2, and predetermined signal processing as described above is performed.

Ro, Go, Bo, Yo, Mo, and Co indicate outputs after signal processing.

Similar signal processing is performed when converting from R, G, and B to other color systems such as Y, M, and C.

[Problems to be Solved by the Invention] By the way, in a color image processing device similar to -, image processing on image data is simply performed using R, G,
, B, Y, M, and C, which are signals unique to the device.

However, if the correction is made by focusing only on the signal level, the correction will not match human sensibilities.

For example, even if the input level is increased by n times, it will not be perceived as n times higher by the human eye.

There are also methods that simply calculate a correction coefficient and add, subtract, multiply, and divide this correction coefficient, but this:? It is a correction that does not fully take into account human sensibilities.

This is because the correction process is not performed in a color system (coordinate system) suitable for the human eye.

Therefore, this invention has been developed to solve these conventional problems, and it is a color image that can obtain more appropriate color images by performing signal processing that takes human sensibilities into account. This paper proposes an image processing device.

[Technical Means for Solving the Problems] In order to solve the above-mentioned problems, the color image processing device according to the present invention includes: forward conversion means for converting color image processing data into values in a color isochromatic space; A signal processing means that performs predetermined signal processing on the forward-converted image data, and an inverse conversion means that converts the image data that has been subjected to the predetermined signal processing into the original color system or another color system. It is special effects to have. .

[Work] A color system suitable for human sensitivity is (L.

u'', vゝ) color system is suitable.Other color systems that cannot be used include (L'', a'', b''
) and Munsell color system.

Therefore, as the forward conversion means 10, (L'', u).

v ”) The color system is used up, and the input color system is converted to a color system that is compatible with the human eye.

(L", Lλ", v"') After the color system conversion, various image processing is performed in the signal processing means 2. After that, (L
", u", v') is inversely transformed from the color system to the input color system. 10' shows this inverse conversion circuit.

After being converted into a form suitable for the human eye, various signal processing is performed, and even after inverse conversion processing, the contents of the oil stopper suitable for the human eye are maintained as they are.

[Example] Next, an example of a color image processing device according to the present invention will be described.
The case where the present invention is applied to the color separation image correction apparatus described above will be described in detail with reference to FIG. 1 and subsequent figures.

In FIG. 1, 1 indicates the entire color image processing device.

Input color image information, for example R, G, B image quality (digital image data), is supplied to the forward conversion unit 0,
The input coordinate system (R, G, B) is converted to a specific color system.

This forward conversion means 10 is a means for converting an input coordinate system into a coordinate space suitable for human sensibilities, and the coordinate space is (L'z, u'2, v'
> ) Since the color system is appropriate, the following example will explain the case where this color system is used.

(L", u"", v") The image data L 1%, ul), y fB converted into the color system is the signal processing means 2
and undergoes predetermined signal processing.

The predetermined signal processing includes gradation correction processing, color correction processing, edge processing, and the like.

The gradation correction process is performed on image data, particularly data indicating the brightness of the image. (L = k.

u ” + V ”) In the color system, L' representing brightness
can be considered to be image data representing the brightness of the image.

Therefore, only the image data L' is supplied to the gradation correction processing means 3, and predetermined gradation correction is executed.

In contrast to 11, color correction processing such as hue correction is performed on other image data u', excluding image data L', contrary to the above.
Only “v” is applicable.

The gradation-corrected image data and color-corrected image data L, u'+v. are supplied to an edge processing means 5, where edges at image contours are emphasized or smoothed.

The edge processing may be performed only on the image data L''.

Although gradation correction processing, color correction processing, etc. can be calculated each time, the correction or correction data calculated in advance may be compiled into a table and the data may be referred to. Therefore, in that case, means 3 and 4
An LUT is used in both cases.

After the various signal processings have been performed, the image data Ro is inversely transformed into the coordinate system of the 10 inputs in the inverse transformer 10', and has the same coordinate system as the input.

Go and Bo are output. Alternatively, it can be converted to another coordinate system (Ro', Go-, Bo-) suitable for use after output.

Therefore, gradation correction, etc., can be performed using a coordinate system that is compatible with human sensibilities, so that unbalanced image processing does not occur.

By the way, in order to convert the input coordinate system to the (Ll, u·+v'') color system, the following means can be adopted.

(L", u", v") In order to create new image data based on the color system, is it possible to use the conventionally known linear masking method or nonlinear masking method? Since all of these methods use polynomial approximation equations, they have the disadvantage that positive errors other than key colors become noticeable.

In the example described below, such a method is not adopted.

In other words, in the color conversion by coordinate conversion described below, for data other than known color conversion data, the color conversion represented by the coordinates of each vertex of a cubic or rectangular spatial area containing input image data is performed. It is calculated by interpolation processing based on information (known image data for color conversion).

The interpolation process eliminates color conversion errors and skips during color conversion, and improves color conversion accuracy compared to conventional color conversion.

Next, an example of the forward conversion means 10 that performs such interpolation processing will be explained with reference to FIG. 2 and subsequent figures.

2 and 3 are diagrams for explaining the case where an example of a new color masking method is applied to the color masking method described above, and FIG. 4 is a diagram showing an example for realizing this color masking method. This is a concrete means to show that.

The color masking method according to this embodiment does not have an LUT (lookup table) for all color combinations, but has accurate color conversion data for color combinations with determined values that are irregular. .

If the color is not that color, interpolation is attempted from the image data of the surrounding points (already calculated color conversion data) using a weighted average.

For convenience, if we explain using the curve of one person's output system,
As shown in FIG. 2, the curve L1 is a color conversion curve for input image data, and the ◯ marks on this line are the discrete color conversion data stored in the cheetah table.

If we assume that X is the color conversion data for which we are trying to find, then this color conversion data
2 allows interpolation. As a result of such interpolation processing, the color conversion data X to be obtained is the previously calculated color conversion data M1. It is always located on the straight line connecting M2. As a result, color conversion errors can be suppressed to a minimum, and skipping of corrected colors can be eliminated.

These color conversion data are prepared according to the type of document. That is, only the color conversion data multiplied by a color correction coefficient suitable for the original or the type of original is prepared. As described above, when there are three types of originals, there are three types of color conversion data.

A specific method of calculating corrected color data will be described next.

In this example, it corresponds to eight color conversion data (L-u'', v') including a rectangular parallelepiped space W (with X at its diagonal apex) determined by three input image data R, G, and B. A rectangular parallelepiped spatial region V formed by known calculated color conversion data (P1 to P8) is determined.The spatial regions W and 2 each have Pl as a reference point.

And for each color, 0.32, 64, 96, 128, 160°192.22
It is assumed that there is a color conversion value for each color combination at each point of 4,255 points. That is, in this case, 8X8X8
=512 spatial regions.

At this point, each input image data R, G, B (100
, 130, 150), it can be interpolated using the color conversion data of the vertices (lattice points) of the spatial area surrounded by the 8 points shown below.

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 conversion data L'' at that time.
i, u"i, v"i are shown.

Pl: (96,128,128) = (L″″1.uJ, v″′1) P2: (128,128,128) = (L″2, u″2, v″2) P3: (96 ,160,128) = (L"3, u"3 + v'3) P4: (128,160,128) = (L"4. u"4. v"4) P5: (96,128,160 ) = (L″′5.u”5.v”5)P6: (128,128,160) = (L′6.u′:6.V′z6)Pl
: (96,160,160) = (L″'?, u”?, v”7) P8 : (128,160,160) = (L”8. u”8. v”8) Each of these vertices Pi The relationship between the spatial region V having , and the spatial region W formed by the input image data is shown in FIG.

In this invention, weighting coefficients for each vertex Pi of these spatial regions (2) are calculated.

There are several ways to calculate the weighting coefficient, but the simplest method is to calculate the volume of the rectangular parallelepiped spatial region created by the opposite vertex of the point with a positive value (I*) and the point This is the weighting coefficient at the point.

Therefore, the weighting coefficient of point P8 is determined by using the coordinates of Pl and the coordinates of The volume of the rectangular parallelepiped spatial region created by is 4x2x22=176, which becomes the weighting coefficient of point P8.

Similarly, the weighting coefficients for the remaining points P1 to P7 can be calculated.

P1=8400 P2=1200P3=560
P4= 80P52 18480 P6=
2640P7=1232P8=17 The sum of these weighting coefficients is the same as the volume of the cubic spatial region (2), and in this example, it is 32768 (assumed to be a).

Therefore, the correction value v″′x, u″x, L′x at point X
is v"x=1/a (P1v1+P2v"2+P3v"3+P4v'4+P
5v”5+P5v”6+P7v”7+P8v”8)u”
x=1/a (P1uJ+ P2u"2+ P3u'3+ P4u"
40P5u”5+P6u”6+P7u”7+P8u”8
) L”x=1/a (PIL”1+ P2L”20 P3L”3+ P4L”
40 P5L"5+ P6L'+3+ P7L"? ＋P
8L"8). Then, the correction values of a point X to be found and the eight points surrounding it are v"'i, u"'i,
If there are L months and each weighting coefficient is Ai, then v"x=(1/nuAi) A i v "iu"x=
(1/AAi)AAiu”iLゞx= (1/疏Ai
) can be expressed as A i L ”i.

Note that the above-mentioned color conversion data is just an example, and in reality, the number of color conversion data may vary depending on the ROM capacity, etc.
It is set to the power of 2. Therefore, 256 bits of R
When using OM, each color can have 32 points of color conversion data (323=32768 points for all three colors).

In this case, the number of divided space regions is (32-1)3=297
It becomes 91.

There are countless ways to obtain color conversion data for grid points.

A simple method is to apply a conventional nonlinear masking method. That is, approximation is performed using a high-order polynomial that minimizes the error, and color conversion data at each grid point is calculated using the polynomial. Since the polynomial at this time is unrelated to hardware and is calculated in advance, it does not matter if it contains any complicated terms (for example, reciprocal, n-th power, logarithm, etc.).

According to the nonlinear masking method, the polynomial is expressed as a general formula: ■゛ゝ=fv (B, G, R) u”=fu (B, G, R) L”=fL (B, G, R) However, , B, G, and R are grid points.

FIG. 4 shows an example of the forward conversion means 10, in which the present invention is applied to a simultaneous color separation correction device that attempts to obtain three color conversion data L", u", and v" at the same time. This is the case.

As is clear from the above equation, this forward conversion means 10
is a color conversion information storage means (color conversion data storage means) 20 that stores a plurality of color conversion data, and an information storage means (color conversion data storage means) for weighting.
The processing means includes a weighting coefficient storage means) 24, a multiplication and accumulation means 30 for multiplying the referenced color conversion data and the weighting coefficient, and accumulating the values, and a division means. Of these, the division means can be omitted depending on the configuration.

The color conversion data storage means 2Q divides the color space formed by the three-color separation image information that can be inputted for color conversion into a plurality of spatial regions, and stores color conversion information for the combination of three-color separation image information located at the apex thereof. (L ” + u ” +
data corresponding to "v") are stored.

The weighting coefficient storage means 24 can output weighting information for a plurality of color conversion data selected from the color conversion information storage means based on the input three-color separation image information.

The processing means processes a plurality of pieces of color conversion information selected from the color conversion data storage means 20 based on the input color separation image information;
Based on the weighting coefficients, the cost of the corrected color separated image data to be finally obtained is calculated and output.

The color conversion data storage means 20 described above includes LtJT21~
In this example, each input image data B, G,
Color conversion data for R (color conversion data that takes into account three types of color correction coefficients corresponding to the original) is stored in each LUT21~
It is stored in 23.

Reference numeral 24 denotes a weighting coefficient storage means, which is also configured as a UT.

Address signals for reading are supplied to the color conversion data storage means 20 and the weighting coefficient storage means 24, respectively. Therefore, the input image data B, G, and R are once supplied to the address signal forming means 40, and an address signal corresponding to the input level is output.

The address signal forming means 40 is also composed of LUTs 41 to 43, respectively. A bipolar ROM is suitable as the LUT.

For these LUTs 41 to 43, the controller 50
A signal is supplied for 1 bit from the 1 bit, the details of which will be described later.

The color conversion data and data indicating weighting coefficients (hereinafter simply referred to as weighting coefficients) referenced by the address signal corresponding to the input level of the input image data are sequentially supplied to the sliding accumulator 30 side a total of eight times.

As described above, the multiplication/accumulation means 30 is for sequentially executing AiKi (Ki is a general term for L'', u', v'') and calculating the sum thereof, and in this example, the multiplier 34 -36 and accumulators 37-39.

Therefore, each multiplier 34-36 has R of 512
The corresponding color conversion data (8
bit) and weight @Ai are supplied. -(,A1K1
The multiplication processing of the upper 8 bits is supplied to subsequent stage accumulators (ALU) 37 to 39, and the multiplication processing is sequentially performed.

Accumulators 37 to 39 are operated with 16-bit precision;
Only the upper 8 bits are used as the cumulative output (product-sum output). By this, the cumulative output is changed to the weighting coefficient Ai
You will get the same output as dividing by . I'm here,
By doing this, the divider can be omitted.

The accumulated outputs of the upper 8 bits are latched by latch circuits 45-47, respectively. Latch pulse is controller 5
Generated with 0.

The configuration of each part will be explained in detail.

The color conversion data storage means 20 stores each color B, as shown in the figure.
LUT 21 ~ in which accurate color conversion data corresponding to G and R (including correction values corresponding to the original; the same applies hereinafter) is stored.
23 is used.

256 bit capacity ROM as LUT21~23
When using , only 32 points are extracted from the minimum level to the maximum level of the input image data. This gives 32 points per color (so for 3 colors, 32" = 3
2768 points) of color conversion data can be stored.

Therefore, when the input level is 256 gradations, 32
For example, the distribution of points is "8" starting from O, as shown below.
Divide into 0.8, 16. ...240,248, total 3
Divide it evenly so that there are 2 pieces, and the 33rd point is 249.
Points above 255 points will not be used. Or 249~
The point 255 is treated as 248.

The color conversion data at each distribution point is accurately calculated, and the calculated color conversion data is applied to each LUT.
21 to 23.

Note that setting the distribution points to 32 in this manner allows the use of a general-purpose ROM with an 8-bit output, which has the advantage of making the storage means 20 inexpensive.

The weighting coefficient A1 at each distribution point is stored in the LUT 24 for weighting coefficient storage means. Now, when allocating 8 bits at a time as described above, 8 weighting coefficients Ai
The total of
If you try to use C, the weighting coefficient (
512), the number of elements increases, so in this example, an approximate value compressed to approximately 1/2 of the theoretical value is used as the actual value of the weighting coefficient.

In the example shown below, the sum of the eight weighting coefficients is always set to 256, and the maximum weighting coefficient of each is 256.
5.

In such a case, for example in FIG. 3, if X is at the same position as Pl, each of the weighting coefficients P1 to P8 will be PI, P2, . P3.
P4. P5. P6. P7. P8255, O
,0,0,0,O,0,1(512,○,○,o
, o, o, o, ○), and the total sum of the weighting coefficients is 256.

Also, if X is located between Pl and 1)3, and is away from Pl by 3 (therefore, 5 from P3), then the weighting coefficients of P1 to P8 are as follows. .

PI, P2. P3. P4. P5. P6. P
7. P81G0. 0. 96. O, O, O
,0,1(320,0,192,0,0,O,○,
O), and the weighting coefficients are appropriately selected so that the total sum of the weighting coefficients in this case also becomes 256.

Similarly, X is separated by 3 from the plane of P1 to P4, and Pl
, P3. P5. Move away by 1 from the surface of P7, move to PI,
P2. P5. If you are 5 degrees away from the PC surface,
The weighting coefficients P1 to P8 are as follows.

PI, P2. P3. P4. P5. P6. P
7. P2S5, 7. 88. 12. 32. 4
．． 53. 7 (105, 15, 175, 25, 63,
9, 105, 15), and each weighting coefficient can be appropriately selected so that the sum of the weighting coefficients in this case also becomes 256.

The above-mentioned 1-bit distribution signal is a control signal for specifying the color conversion data MX, M2 before and after the point X, if explained with reference to FIG.

In other words, for convenience of explanation, -32 distribution points (lattice points)
The relationship between the address signal and the corresponding address signal is set as shown in FIG.

Now, when the level of the input image data is 100, the address signal (12, 13) is outputted from the color conversion data storage means 20 to output the previous and subsequent color conversion data (96 and 104) including this input level. need to be formed.

Therefore, the distribution ζ′) signal or the address signal (
12) is output, and the address signal (13) is controlled so that the distribution is a signal or 1 and the larger color conversion data (104) is referred to.

However, when the maximum value to be used (248 in this case) is used, when the signal is O, the color conversion data of its own value is selected, and when the signal is 1, the smaller color conversion data is selected. Select color conversion data (240 in this case).

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

Also, from the L”, u − v′ coordinates, the original (Ro,
Go.

Bo) A device for converting into coordinates can also be realized with the same configuration as described above. That is, in this case, it is sufficient to simply replace the data in the color conversion data storage means 2Q.

[Effects of the Invention] As explained above, according to the present invention, when performing image processing, the input coordinate system is first transformed into a coordinate system that suits human sensibilities, and then various image processing is performed. This is an inverse transformation to the input coordinate system.

By performing such processing, gradation conversion, color correction, edge processing, etc. can be performed in a state that is compatible with human sensibilities.

Therefore, it is possible to realize more natural image processing, and it is possible to reproduce an image that is faithful to the original image.

For this reason, the digital image composition apparatus according to the present invention is extremely suitable for application to image processing apparatuses such as color blueprints, video printers, and digital color copiers.

[Brief explanation of the drawing]

FIG. 1 is a system diagram showing an example of a color image processing device according to the present invention, FIG. 2 is a characteristic curve diagram for explaining a forward conversion means to which this invention is applied, and FIG. 3 is an explanation of a color masking method. 4 is a system diagram of essential parts showing an example of the forward conversion means, FIG. 5 is a diagram showing the distribution relationship of grid points, and FIG. 6 is a system diagram of the color image processing apparatus. 1... Color image processing device 2... Signal processing means 3... Gradation correction means 4... Color correction means 5... Edge processing means 10... Forward conversion means 10 -... Reverse Conversion means 20...color conversion data storage means 30...multiplying and accumulating means 40...address signal forming means 50...controllers V, W...spatial area χ...color conversion data

Claims (2)

[Claims] (1) Forward conversion means for converting color image processing data into values in the isochromatic space; signal processing means for performing predetermined signal processing on the forward-converted image data; and image data subjected to predetermined signal processing. 1. A color image processing device comprising: an inverse conversion means for converting a color into an original color system or another color system. (2) The color image processing apparatus according to claim 1, wherein the predetermined signal processing includes edge enhancement or smoothing signal processing. (3) The color image processing device according to claim 1 or 2, wherein the conversion means uses an interpolation calculator using a look-up table.