JPH02226867A - Color decomposition picture correction device - Google Patents

Abstract

PURPOSE:To shorten the interpolation processing time by utilizing split spaces being the number of minimum apexes and applying an interpolation processing. CONSTITUTION:A latch output from latch circuits 12R, 12G, 12B is separated into high-order 5 bits and low-order 3 bits to specify a triangular pyramid used for the interpolation processing respectively. A ROM 17 used for a distributing means discriminates to which of 6 triangular pyramids in total formed by the apexes of a basic grid an interpolation point belongs based on the low- order 3-bit signal and the ROM 17 outputs the distributing signal to specify the apexes forming the triangular pyramid sequentially. Then the interpolation processing is implemented by utilizing the split space with the minimum number of apexes such as triangular pyramid in the case of 3-dimension by 3 inputs and super-triangular pyramid (triangular pyramid in 4-dimension) in the case of 4-dimension with 4 inputs. Thus, the interpolation processing time is shortened.

Description

DETAILED DESCRIPTION OF THE INVENTION [Field of Industrial Application] The present invention is directed to a color separation ViA image correction device suitable for application to a correction device for video printers, digital color copies, and the like.

[Background of the Invention] When making a hard copy of a television image using a color printer, digital color copying device, etc., each color system is different, so a color separation image correction device is used to match the reproduced colors. be done.

FIG. 1O is an example of this, and the input red R0 green G,
The three primary color print data for blue B is yellow Y, magenta M, and cyan C at color masking stage 10.

This Y, M,
C.

K il image data is supplied to the color printer 100.

In this case, Y, M corresponding to one R, G, B
.

There are countless silk combinations for C and K, and this Y and M.

There are various methods for determining the combination of C and K.

This method generally uses tJ CR (Under Co1or
It is called the Re-ov-1) method, but one of the methods is 1.
There is a 00% UCR.

In this 100% OCR method, for example, R, G, B are Y
, M, and C, and this Y, M
.

The minimum concentration of C is the concentration of K! For example, 1 degree of Y, M, C before substitution is the 1st Im
As shown in A, when YO9Mo+Co and Yo is at the minimum concentration, Y and M are II depleted.

The concentrations of C and K are the same! ! As shown in IB, Yn,
Mn, Cn, Kn (Yn = 0).

Furthermore, in Japanese Patent Application No. 83-291813, the present applicant has proposed the following method (for convenience, this is referred to as "improved type 100%
proposed the UCR method).

That is, Y, M, corresponding to R, G, B
Among the combinations of C and K, the maximum possible value of K is determined as K, and other combinations of Y, M, and C are determined from this.

Now, in the color masking device 10, R2O,B
iii image data by the 100% UCR method or the improved 100% NCR method as described above.

M, C, K's i! In order to convert to i-image data, R
Y, M, C, corresponding to the image data of 2O,B
K image data is stored in a lookup table (LUT) in advance.
It is conceivable to store the R, G, and B image data into Y, M, C, and K image data by referring to this LUT.

However, if Y, M, C, and K image data corresponding to all R, G, and B image data are to be stored in the LUT, there is a drawback that the capacity of the LUT becomes enormous.

Therefore, in order to reduce the memory capacity, the applicant has
The color space formed by R, G, and B image data is divided into multiple basic grids, and the LUT contains Y, M, C, and K for the combination of R, G, and B image data located at the vertices. Store image data of R, G
, Y and M corresponding to the image data of B.

When C, K image data does not exist, this R, G
, B image data (interpolation points) are included in the basic lattice vertices.

For example, as shown in FIG. 12, if there is an interpolation point P in the basic grid consisting of vertices A-H, then the interpolation point The volume of the rectangular parallelepiped is 5. The vertices A to H are Y, M, and C.
, K is used as a weighting coefficient for the image data.

That is, the vertices A~ of the basic lattice that include this interpolation point P
Image data of Y, M, C, and K of H. M Meng, Ci, Ki (Mu=l~8), Y of vertices A~H, M
, C, K, the weighting coefficients for the image data are Ai (
i=1 to 8), then Y of interpolation point P.

M, C, K image data Yp, M+), C
p and Kp are calculated by the following formula.

Mp=(x/l)±AiMi ill ill Cps(1/ ゑ) AζA1C1 K9= cx /i:+ @ A Natsume Ki chant rumor 1. II [Problems to be Solved by the Invention] With such interpolation calculation processing, Y and M of the interpolation point P.

When calculating C and K image data Yp, Mpt Cp, and Kp, eight multiplication and accumulation processes are required for each, which has the drawback of increasing processing time.

SUMMARY OF THE INVENTION Therefore, it is an object of the present invention to provide a color separation image correction apparatus that can shorten processing time without the same inconvenience.

[Means for Solving the Problem] The present invention divides a color space formed by n color separation image data (n is an integer of 2 or more) that can be input for color correction into a plurality of basic grids, and data storage means each having image data for yellow Y, magenta M, cyan C, and black corresponding to the n-color separation image data located at the n-color separation image data; weighting coefficient generating means for generating weighting coefficients for each of the (n+1)II image data of Y, M, C, K;
Y, M, and C are selected from the data storage means based on the color separation image data.

processing means for outputting K (n+1) image data and Y, M, C, K image data neatly processed by multiplication and accumulation processing of the weighting coefficients generated by the weighting coefficient generation means; The (n+1) image data of , M, C, and K are divided into (n+1) of the divided space formed by dividing the above-mentioned basic grid so as to include dividing lines parallel to the achromatic color axis.
Y for the above n color separated image data located at the vertices of
, M, C, K image data.

[Function] In the above configuration, in the case of three dimensions with three inputs, a pentagonal pyramid,
In the case of four dimensions with four inputs, the interpolation process is performed using a divided space with the minimum number of vertices, such as a construction pyramid (four-dimensional pentagonal pyramid), so the interpolation process time can be shortened.

Furthermore, when a divided space with a minimum number of vertices is divided to include, for example, a dividing line perpendicular to the achromatic axis, there is an inconvenience that interpolation errors occur, but in the above configuration, it is possible to minimize Since the divided space having the number of vertices is divided to include the dividing line parallel to the achromatic axis, interpolation errors can be suppressed.

[Example] Hereinafter, an example of the present invention will be described with reference to the drawings.

In this example, the basic lattice is further divided and interpolation processing is performed using a divided space with the minimum number of vertices, that is, a pentagonal pyramid.

As shown in FIG. 1, a total of 6g of pentagonal pyramids are formed by dashed-dotted lines on a basic lattice composed of vertices A-H. In this case, a pentagonal pyramid is formed by dividing so as to include a dividing line parallel to the achromatic axis (see the dashed line connecting A-G).

Here, since the interpolation point P exists within this basic grid, it is determined in which pentagonal pyramid it is included.

For example, if the coordinates of vertices A to H are as shown in the figure, [n point P
(7) ml [KaP (5, 1, 2) Fart 1 Ni is
This interpolation point P is the vertex A as shown in FIG. 2A.

It can be seen that it is included in the pentagonal pyramid T formed by B, C, and G. Figure 8 shows the color system related to this pentagonal pyramid T. If vertices A' to G' correspond to vertices A to G, respectively, then the interpolation point P' is within the triangular pyramid T'. exist.

Once the pentagonal pyramid T is resolved, the interpolation point P is connected to the vertices A, B, C, and G, forming a total of four new pentagonal pyramids, as shown in A in the same figure. Volume of V B
CGP, V ACGP, V ABGP, V
ABCP is required. These volumes V BCGP,
V ACGP, V ABGP, VA8CP
The interpolation point P' is calculated from the following equation from the vertices A', B', and C'G' of the color system shown in FIG. In addition, V A
BCG is the volume of the pentagonal pyramid T.

P' = 1/VABCG (VBCGP-A'+VAC
GP-B'+VABGP-C'+VA8CP-G'
) If the coordinates of the interpolation point P differ, the pentagonal pyramid T used will also differ. The coordinates of point P between 31 and 31 are P(3°1.6
), then 1. Since this interpolation point P is included in a pentagonal pyramid T formed by vertices A, C, D, and G as shown in FIG. 3, this pentagonal pyramid T is used.

In this way, by performing the interpolation process using the pentagonal pyramid, the interpolation point P' can be calculated by performing the multiplication and accumulation process four times.

FIG. 4 shows an example of a color masking device 10 in which such interpolation processing is performed.

R supplied to terminal 11R, IIG, IIB,
G.

The 8-bit image data of B is sent to the latch circuit 12.
It is supplied to R, 12G, and 12B and latched (Fig. 5A, D).

In this case, the reference clock CLK supplied to the terminal 14
(B in the figure) is supplied to the four-way counter 16. The 2-bit count output of the quaternary counter 15 (C in the same figure) is
The latch circuit 128.12G is separated into an upper bit and a lower bit, and the signal of the upper bit is used as a first latch pulse (E in the figure).

12B.

The latch outputs from the latch circuits 12R, 12G, and 12B are separated into upper 6 bits and lower 3 bits in order to specify the pentagonal pyramid used in the interpolation process. In this case, the upper 5 bits represent the reference point (reference lattice point) of the vertex of the basic lattice that includes the interpolation point P determined by the image data of RlG, B (vertex A in Figure 1), and the lower 3 bits represent the interpolation point. It represents the position within the basic grid of P. In other words, in this example, the basic grid is divided into eight equal parts in each direction, and the number of interpolation points is seven.

The lower 3 bits of the signal are distributed by the RO which functions as a means.
It is supplied to M17. This ROM 17 contains the 2-bit counter output from the quaternary counter 15, the 51st! I.E., F.
) are also supplied. In this ROM17, it is determined which of the total 6gM pentagonal pyramids formed by the vertices of the basic lattice the interpolation point P is included based on No. 18 of the lower three bits. Signals are sequentially output for identifying the vertices forming the pentagonal pyramid.

For example, when the coordinates of the interpolation point P are P (5, 1, 2), the coordinates of the vertices A to H of the basic lattice including this interpolation point P are the 11th! It is determined as shown in l.

Next, it is determined which of the six pentagonal pyramids formed by the vertices A to H of this basic lattice contains the interpolation point P, and the vertices A and B are selected as shown in FIG. 2A.

Triangle formed by C and G! IT is identified.

A (0, O, O) B (8,0,0) C (8,0,8) G (8,8,8) In this case, the basic grid spacing (pitch) is ) Mj signal, which corresponds to the octagonalization level), vertex B is specified, and
Using vertex A as a reference, if you move by the basic grid interval in each of the X and 2 directions, vertex C can be identified.Furthermore, if you move from vertex A by the basic grid interval in each of the X, Y, and 2 directions, Vertex G is identified.

Therefore, in order to specify the pentagonal pyramid T formed by the vertices A, B, C, and G, the following signals are output from the output terminals at b and c of the ROM 17 to 2 bits of the 4-way counter 15. The primary output is based on the count output.

(0,0,0) (1,0,0) (L O,1) (1,1,1) This excess signal is supplied to adders 118R, 18G, and 18B and becomes the upper 6 bits of the signal. Added to vertex A,
B, C, and G are calculated, and the pentagonal pyramid T is specified.

That is, since the first base signal is (0, 0, 0), even if this is added to the reference grid point (0, 0, 0), the position does not move and the vertex A is specified.

The signal for the next distribution is (J, 0. O), so adding this to the reference grid point (0, 0, 0) gives (1,
0,0> ((8,0,0) at the quantization level), and vertex B is specified. For the next shot, the signal is (1,0
, 1), adding this to the reference grid point (0, 0, 0) yields (L O+ 1) ((8, 0, 8) at the quantization level), and vertex C is specified. Furthermore, since the signal for the next related part is (1°1.1), adding this to the reference grid point (0,00) (].1.1
) ((8, 8, 8) at the quantization level), and the vertex G
is specified.

Further, for example, when the coordinates of the interpolation point P are P (3, 1° 6), the pentagonal pyramid T including the interpolation point P becomes as shown in FIG. Therefore, vertex A.

In order to specify the pentagonal pyramid T formed by C, D, and G, the following distribution signals are sent from output terminals a + b * c of the ROM 17 to 2 of the 4-way counter 16.
The bits are sequentially output based on the bit count output.

(0, 0, 0) (1, 0, 1) (0, 0, 1) (1, 1, 1) Then, the signal for this drop is added 1118R, 18G,
18B and added to the upper 6 bit signals to calculate vertices A, C, D, and G, thereby specifying the triangular pyramid T.

Further, the signals for specifying the vertices of the pentagonal pyramid T outputted from the adders 18R, 18G, and 18B (see FIG. 5H) are sent to latch circuits 19R and 19G, respectively.

It is supplied to 19B and latched (1!11),
These latch circuits 19R, 19G, 19B have terminals 14
The reference clock CLK supplied to is supplied as a latch pulse.

The latch outputs from these latch circuits 19R, 19G, and 19B are color correct data forming means 20Y, 20M for yellow Y1, magenta M, cyan C, and black K.
20C, 20K, Y, M, respectively.
C and K image data are formed.

The color correction data forming means 20Y is configured as follows.

That is, the latch outputs from the latch circuits 9R, 19G, and 19B are supplied to the color reproducing ROM 21Y, and the corresponding vertices of a new pentagonal pyramid T' obtained when the pentagonal pyramid T is mapped onto a different color system are The coordinates of are referenced. In other words,
This color reproduction ROM 21Y stores Y image data corresponding to the vertices of the basic grid, and this is stored in the latch circuit 19R,
19G.

It is made to be referenced by the latch output from 19B.

Note that the color reproduction ROM includes, for example, the above-mentioned 100% UC.
Y, M by R method or improved 100% UCR method
, C, and K image data are stored.

The image data of Y (FIG. 5J) sequentially outputted from the color reproduction ROM 21Y is supplied to the multiplier accumulator 22Y, and the image data of Y (FIG. 5J) is supplied to the multiplier accumulator 22Y. Coefficients are sequentially supplied and multiplication and accumulation processing is performed. This multiplication accumulation! ! ! The reference clock CLK supplied to the terminal 14 is supplied to 22Y.

In this case, latch (ro) road! The lower 3 bits of the latch output signals from 2R, 12G, and 12B are the weighting coefficient ROM3.
0, and this weighting coefficient ROM 30 contains 4
A 2-bit count output from the path counter 16 is supplied, and the weighting coefficients determined by 7 based on the volume of a new pentagonal pyramid formed by the vertices forming the pentagonal pyramid T and the interpolation point P are sequentially referred to.

For example, in the case of the pentagonal pyramid T shown in FIG. 2, the following four orchid weighting coefficients are referred to.

V BCGP, V ACGP, V AB(,P
, V ABCP After the weighting coefficient output from the weighting coefficient ROM 30 is latched by the latch circuit 31, it is supplied to the multiplication and accumulation 1122Y, and the arithmetic processing as shown in the following equation is executed. The reference clock CLK supplied to the terminal 14 is supplied to the latch circuit a1 as a latch pulse.

P' (Y) = 1 /VABCG (V8CG
P-A'+VACGP-B'+VA8GP-C'
+VABCP-G') No. 51! IK indicates data that is being processed.

Here, M indicates multiplication processing, A indicates accumulation processing, and R indicates rounding processing.

The multiplication accumulator 22Y outputs the calculation result each time it is multiplied ("15, N), and this output is supplied to the latch circuit VIi23Y and latched. Then, this latch circuit 23Y
The result of the calculation at the end of the night is output as interpolated data Y (see P in the figure).

Here, in order to perform the above-mentioned arithmetic processing at a predetermined timing, a control circuit '#i40 consisting of a logic circuit is provided. The 2-bit count output (E, F in the figure) from the four-way counter 16 is supplied to the NAND circuit 41, and the output of this NAND circuit 41 (G in the figure) is supplied to the latch circuit #a43 and latched. . The latch output (shown in the same figure) from this latch circuit 43 is used as an accumulation pulse for multiplication accumulation 112.
Supplied to 2Y.

Further, the latch output from this latch circuit 43 is supplied to the latch circuit va44, and the latch output from this latch circuit 44 is supplied to the latch circuit 23Y as a second latch pulse (O in the figure). Latch circuits 43 and 44 have terminal 1
The reference clock CL I (supplied to 4) is supplied as a latch pulse.

Further, the phase of the output of the NAND circuit 41 is inverted by an inverter 42, and the output of the inverter 42 (M) in the figure is supplied to the accumulator 22Y as a rounding pulse.

By the way, as the color reproduction ROM21Y, for example, 25
A ROM with a capacity of 6 bits is used, and 32 bits between the minimum level and the maximum level of R9G, B image data are used.
Only the points are extracted, thereby 32X32X32=3
2768 points (DY image data is stored.

In this case, the R, G, and B image data are 8 bits and have 266 gradations, and the distribution of 32 points is, for example, as shown below, starting from 0 and dividing into 8's. , 8. 16. ◆・4- 240.248, which is divided into equal parts for a total of 32 points, and the 33rd point from 249 to 256 is not used, or 24
treated as a day.

The Y image data of each of these distribution points, that is, the vertices of the basic grid whose basic grid spacing is 8 quantization levels, is accurately calculated, and this calculated Y image data is used as the color reproduction RO.
Stored in M21Y.

Further, when the basic lattice spacing is 8 quantization levels, the total of the four weighting coefficients is 8X8kB/6=512/6, which is normalized to 256. Then, as the weighting coefficient ROM 30, an 8-bit general-purpose IC is used.
The sum of the four weighting coefficients is always 256 so that
In addition, the maximum value of the weighting coefficient is 266. For example, if the interpolation point P is at the same position as the vertex A in FIG. 1, the weighting coefficients V 8CGP, V ACGP
, V ABGP, V Al3CP are as follows. Values in parentheses are theoretical values.

V BCGP, V ACGP, V ABG
P, V A & CP (512/8 0
0 0) In this way, the sum of the four weighting coefficients is always 256, so even if it is not mentioned above, the multiplication cumulative e22Y
The division process of 1/V ABCG in the output 8
This is done with a bit shift. Then, as rounding processing in this case, for example, it is determined whether the 8th bit from the bottom is "1" or "0", and if it is "1", rlJ is added to the 9th bit. .

Although the explanation is omitted, the color correction data forming means 20M
, 200.20 as well as the color correction data forming means 20 described above.
It is configured on the same machine as Y, and interpolated data M, C, and K are output.

In this way, in this example, the interpolation process is performed using a pentagonal pyramid, which is a divided space with the minimum number of vertices, and only four multiplication and accumulation processes are required, compared to the conventional interpolation process using a cube. Processing time can be significantly reduced.

In addition, in this example, the basic lattice composed of vertices A-H is divided to include dividing lines parallel to the achromatic axis to form a total of six pentagonal pyramids. Therefore, Y, M, C stored in the color density fJI ROM
，! Even if the image data of () is based on the 100% NCR method or the improved 100% NCR method, interpolation errors can be reduced.

In other words, Y by 100% UCR method.

When performing interpolation processing using M, C, and K image data, the conventional interpolation processing using a cube
As shown in FIG. 6, the Y, M, C, and K image data are not smoothly interpolated and become wavy. In this figure, the input is 8-bit (0 to 255) R, G, and B image data, and the output is the converted Y, M, and C image data.
, K is image data. Gray wedge (R=G=8
), because it is! The number of image data is the largest. Here, the wavy portions are data points, and the curved portions are interpolated points.

Furthermore, even if the interpolation process is performed using a pentagonal pyramid, if the basic lattice is divided to include a dividing line perpendicular to the achromatic axis to form a pentagonal pyramid, the seventh As shown in the figure, the image data of Y, M, and C1I are not interpolated smoothly and become wavy.

On the other hand, in this example, the interpolation process is performed using a triangular pyramid, and the basic grid is divided so that the dividing line parallel to the achromatic axis is included. As shown in Figure 8, Y, M
, C, and K are smoothly interpolated.

The above can be explained by modeling as shown in FIG.

Here, it is assumed that for a two-dimensional input, only the height of the arrow is output. The output data changes in a v-shape. In this case, the height of the arrow in the valley of the figure 7 is reduced by performing UCR, and the height of the arrow becomes O according to the 100% UCR method.

Figure A is a model of interpolation processing using a cube, and V
When interpolating the valley part of a character, it is interpolated in an arched manner as shown by the broken line.

In addition, Figure B is a model of interpolation processing using a triangular pyramid formed to include dividing lines orthogonal to the achromatic axis with respect to the basic lattice. It is interpolated like a triangular wave.

On the other hand, C in the same figure is a model of interpolation processing using a triangular pyramid formed to include dividing lines parallel to the achromatic axis with respect to the basic grid. , it can be seen that interpolation is performed smoothly as shown by the broken line.

In addition, in the above example A, the Y, M, C, and K image data stored in the color reproduction ROM are 100% OC.
I mentioned that the improved 100% UCR method will be used to settle R, but Y, M, and C stored in ROM.

Even if the image data of K is obtained by the general UCR method, the present invention can be similarly applied and the same effects can be obtained.

Furthermore, in the above embodiment, four t! ! , correction data forming means 20Y to 20K are provided so that the interpolation process for Y~ is serially processed, but only one color correction data forming means is provided, and Y is stored in the color reproduction ROM. ,
M, C, and K image data are stored respectively, and the image data output from this color depth 7M ROM is Y,
Switch to M, C, K in order, then Y, M, C
, r( may be sequentially performed.

Furthermore, although the above-mentioned embodiment shows the case of 3 inputs, the present invention can be similarly applied to the case of 4 inputs or the like. For example, in the case of 4 inputs, interpolation processing will be performed using a 4-dimensional triangular pyramid (laid-out pyramid), but in this case, the 24 laid-out pyramids are divided into segments whose basic lattice is parallel to the achromatic axis. It will be divided and formed so that the line is included.

Effects of the Invention As described above, according to the present invention, interpolation processing is performed using a divided space with a minimum number of vertices, so that the interpolation processing time can be shortened. Further, since the divided space having the minimum number of vertices is divided so as to include the dividing line parallel to the achromatic color axis, interpolation errors can be significantly reduced.

1 to 31111 are diagrams for explaining one embodiment of the present invention, No. 4m is a configuration diagram showing one embodiment of the present invention, FIGS. 5 to 9 are diagrams for explaining the same, and FIG. Figures 10 to 12
The figure is a diagram for detailed explanation of the invention.

! 2R ~ 1213.1 15　・・ 18R ~ 188 1 20Y~20 Bōken 1Y 1 22 Y 1 40・・ 9R~19B, 23Y, 3 trouch circuit ・Quadratic counter ・Sorting ROM ・Adder ・Method for forming correct data ・Colorful Ill ROM ・Multiply accumulator ・Control circuit

Claims (1)

[Claims] (1) n color separation image data that can be input for color correction (
(n is an integer of 2 or more) is divided into a plurality of basic grids, and the image data of yellow Y, magenta M, cyan C, and black K are respectively divided into a plurality of basic grids, and the image data of yellow Y, magenta M, cyan C, and black K are respectively and a weighting coefficient for generating a weighting coefficient for each of the (n+1) Y, M, C, and K image data selected from the data storage means based on the input n color separation image data. and (n+1) of Y, M, C, and K selected from the data storage means based on the input n color separation image data.
Y, M corrected by multiplying and accumulating the image data and the weighting coefficients generated by the weighting coefficient generating means.
, C, and K, and the (n+1) pieces of image data of Y, M, C, and K are processed so that the basic lattice includes a dividing line parallel to the achromatic color axis. Y, M, C for the above n color separated image data located at (n+1) vertices of the divided space formed by dividing into
, K image data.