JPH04332262A - Color converter - Google Patents

Abstract

(57) [Summary] This bulletin contains application data before electronic filing, so abstract data is not recorded.

Description

[Detailed description of the invention]

0001

[Industrial Application Field] The present invention is applicable to applications in which color images and color signals are input and arbitrary color conversion is performed in real time, such as color hard copy devices, color display devices, color television camera devices, and color recognition devices. The present invention relates to a color conversion device used in devices and the like.

0002

Conventionally, in image processing of monochrome images, the information held by one pixel of an image is one-dimensional information called brightness (density), and brightness conversion is performed as so-called gamma curve conversion.
LUT (lookup table) for various nonlinear curves
It was possible to convert it in real time by writing it in . Even if the image being handled is a color image, in applications where color conversion is performed in real time, it is treated as three monochrome images: an R (red) plane, a G (green) plane, and a B (blue) plane, each of which has an independent It was often converted using LUT. However, in this type of processing, the color conversion that can be handled is essentially one-dimensional processing, and R'=hR(R), G'=hG(G), B'=
Only color conversion in the form hB(B) is possible.

[0003] In color image processing, the information held by one pixel is three-dimensional information (R, G, B), and color conversion in the original sense is a three-dimensional conversion R'=f that puts these together.
R(R,G,B) G'=fG(R,G,B) B
'=fB(R,G,B).

For example, HLS conversion, which converts a color expressed as (R, G, B) into hue H, brightness L, and saturation S, is a technology that has become important in color image processing in recent years.
As in H=H (R, G, B), one output is a function of three inputs and belongs to the three-dimensional transformation described above. However, when trying to convert these using a general-purpose table, assuming that one color is an 8-bit signal, the conversion per color is 16 (Mbytes).
e) Requires additional memory capacity. Therefore, conventionally, there is a need for hardware that can perform three-dimensional color conversion for any arbitrary color conversion in a general-purpose manner and in real time. Therefore, at present, new hardware is designed for each purpose in real-time color image processing equipment.
A general-purpose real-time color conversion device has not been realized.

However, for color correction of color hard copies and color scanners, the color space is divided into a plurality of color spaces, a plurality of pieces of color correction information located at the vertices of the color space are selected, and a color signal is weighted and interpolated and output. There is an example of an interpolation method (Japanese Patent Publication No. 58-16180). In this example, a unit cube, which is a basic solid in a three-dimensional color signal space, is set for the interpolation process, and the unit cube is divided into a plurality of tetrahedra to simplify the interpolation calculation.

By using this as a color conversion device, it is possible to perform free nonlinear color conversion at high speed while maintaining the gradation of an image, such as changing the color of a specific color in a color space. However, in the example of the previous patent, the memory device used for interpolation stores the output value for the input value (the value of the second signal system in the patent), so when finding the interpolated value for the input value, The four coefficients had to be multiplied by the known values at the corresponding four tetrahedron vertices and each had to be added, and there was a restriction that the memory device addresses formed a cube.

[0007]

[Problems to be Solved by the Invention] First, in order to configure a high-speed color conversion device, interpolation calculations for color conversion values must be performed at high speed, and multiplication and addition necessary for interpolation must be performed in parallel. There is a need. However, in the interpolation method of the above patent, there are four multiplications to be performed in parallel, so four multiplications are required. In actual hardware, there is a desire to reduce the number of multipliers to three for simplicity.

Second, in the above patent, since the table memory stores the second signal system, which is the output value to be obtained for the first signal system, which is the input value, the second signal system has It is necessary to allocate the dynamic range as is to the storage bit width of the table memory. for example,
Even when the value of the second signal system does not change much compared to the first signal system, the second signal system is forced to store a very large numerical value as its absolute value. This does not make effective use of memory, and cannot be said to be the best method given the limited storage capacity.

Third, in an actual color space, it is difficult to say that each axis reflects human visual characteristics equally, and in a normal red, green, and blue color space, the green axis is particularly visually sensitive. There are results that are important. Therefore, if the total number of bits is constant, interpolation accuracy can be increased by allocating more bits to the green axis, thereby increasing the overall interpolation accuracy. From this point of view, it is desirable that the unit interpolation section of the color space be a rectangular parallelepiped, but in the above patent, the unit interpolation section was limited to a cube.

The present invention solves the above-mentioned problems of the prior art, and provides a color conversion device that can process any color conversion in real time whose input and output are three-dimensional or more than arbitrary dimensions using an interpolation method without discontinuities. The purpose is to provide

[0011]

[Means for Solving the Problems] In order to achieve the above object, the present invention inputs a set of three-dimensional or arbitrary-dimensional color signals, and assigns a different number of bits to each input signal, or assigns the same number of bits to each input signal. Split into upper signal and lower signal,
A table memory for storing color conversion value difference values selected in the set of upper signals, a table memory for storing color conversion value reference values selected in the set of upper signals, and a weight selected in the set of lower signals. coefficient providing means; means for selecting a plurality of unit tetrahedrons obtained by dividing a unit cube or unit rectangular parallelepiped in a color space selected by the upper level signal; and color conversion values at a plurality of vertices of the unit tetrahedron. calculation means for performing three-dimensional or arbitrary dimension interpolation of a color space from the difference value and color conversion value reference value and the weighting system value; and calculation function setting means for checking the calculation value of the calculation means. It was established. The purpose is to change the value stored in the memory from the value of the second signal system at each vertex of the tetrahedron to the second signal system difference value at one reference vertex of the tetrahedron. Next, the interpolation method is calculated using three interpolation coefficients and the second signal system difference values at the three vertices.
The method is changed to multiplication of times and addition of the second signal system reference value. By this measure, the number of multipliers required for interpolation can be reduced to three.

At the same time, the second signal system difference value storage memory stores signal difference values, which have small absolute values, and as a result, the memory can be used efficiently. Further, each axis of the input color space is divided into its upper and lower signals with different bit allocations, and the basic unit of interpolation is a rectangular parallelepiped in the color space, thereby performing interpolation that matches human visual characteristics.

[0013]

[Operation] The present invention uses the above structure to change the value stored in the memory from the value of the second signal system at each vertex of the tetrahedron to the second signal system difference value at one reference vertex of the tetrahedron. Next, the interpolation method is changed to three multiplications of the three interpolation coefficients and the second signal system difference values at the three vertices, and addition to the second signal system reference value. By this measure, the number of multipliers required for interpolation can be reduced to three.

At the same time, the second signal system difference value storage memory stores signal difference values, which have small absolute values, and as a result, the memory can be used efficiently. Further, each axis of the input color space is divided into its upper and lower signals with different bit allocations, and the basic unit of interpolation is a rectangular parallelepiped in the color space, thereby performing interpolation that matches human visual characteristics.

[0015]

[Example] The concept of the present invention will be explained below.

First, the signal interpolation method described in the above-mentioned Japanese Patent Publication No. 16180/1980 will be explained. For simplicity, a case where the input space is one-dimensional is shown. In Figure 2, the output value (P) at any point P included in the unit interpolation section AB
is the output value (A), (B) at A, B and two coefficients a,
Using b, (P) = a × (A) + b × (B) (Equation 1
) However, a+b=1
(Formula 2) is obtained. In this case, two multiplications are required. Therefore, another interpolation formula, (P) = (A) + b [ (B) - (A)] (Equation 3)
will be introduced. In other words, the output value reference value (
If A) and the output value difference value [(B)-(A)] are accumulated, multiplication can be performed in one interpolation using the coefficient b for (B) as shown in Equation 3.

FIG. 3 shows a case where the same thing is extended to three dimensions. The basic interpolation interval in three dimensions is a tetrahedron ABCD, and the output values at each vertex are (A) (B) (C) (D
). In the signal interpolation method of the patent, the coefficients a, b,
Using c, d, the interior point (P) is expressed as (P)=a
×(A)+b×(B)+c×(C)+d×(D) (
Equation 4) was used, but if left as is, four multiplications would be required. Therefore, the reference vertex among the four vertices is set to A, the output value reference value (A) is stored in A, and the difference value between the original output value and the output value reference value is stored in B, C, and D. )-(A)] , [
(C)-(A)], [(D)-(A)], (P)=(A)+b[(B)-(A)]
+c [ (C)
-(A)] +
d[(D)-(A)] (Formula 5
) can be interpolated using three multiplications.

The present invention can perform free color conversion at high speed by using a simple interpolation method as shown in Equation 5 above.

Next, a method of dividing the input color space into tetrahedrons according to the present invention will be explained. input color (x, y, z)
As an 8-bit numerical value expressed as (x1
, y1, z1), and the lower signal is (x2, y2, z2)
It is expressed as Assuming that 3 bits of each signal are taken as the upper signal, this upper signal is the unit solid to which the input color belongs in the input color space created by (x, y, z), in this case the unit cube is 512 (= 8x8x8), and the lower signal has the role of indicating the existence 1 in the unit solid selected by the upper signal. Figure 4 shows how the entire color space is divided into unit solids, in this case unit cubes, and Figure 5 shows how to divide one unit solid into six unit tetrahedra (units 0 to 5). Indicated. Which of the unit tetrahedrons a given color signal belongs to is determined by the (x2, y2, z2) no relational expression as shown in Table 1.

[0020]

[Table 1]

Now, once the unit tetrahedron to which the input color belongs is determined, the interpolation formula becomes (Equation 1).

[0022]

[Math 1]

Therefore, as shown in FIG. 6, Δ is the volume of the unit tetrahedron, and Δi is the vertices P1, P2, P3, P of the unit tetrahedron.
A new 4 is created by connecting 3 of 4 and the input color point P.
Create small tetrahedrons, and represent the volume of the small tetrahedron that does not include point Pi, and Di is the color conversion value difference value at point Pi,
O1 indicates the color output value at the reference vertex P1.

[0024] If the above-mentioned upper signal and lower signal division is performed with different bit allocations for each color, the color space can be divided into a set of rectangular parallelepipeds, and then interpolation can be performed using almost the same procedure. For example, input color signals are converted into upper signals (x1, y1, z1) to lower signals (x2, y2, z) with non-uniform precision.
2) Divide into 2). Here, x1, y1, z1 are 3, 4,
2 bits, and x2, y2, and z2 are 5, 4, and 6 bits. In this case, the unit cube is 32, 16, 6 in each axis direction.
4, and 8, 16, and 4 of these are lined up in each axis direction. This bit allocation is just one example, but the upper signal has the role of selecting one unit cube out of 512 (=8 x 16 x 4) in the color space created by (x, y, z). The lower signal indicates the position within the unit cube selected by the upper signal. FIG. 7 shows how the entire color space is divided into unit cubes, and FIG. 8 shows how one unit cube is divided into six unit tetrahedra (units 0 to 5).

[0025] Specific examples of the present invention will be described below.
This will be explained with reference to FIG. FIG. 1 is a block diagram of a color conversion device according to an embodiment of the present invention. In FIG. 1, only the portion where the output signal R' is generated from the input signals (R, G, B) is shown. Input signal (101~1
03) is gamma conversion table memory (104-106)
Gamma conversion is performed by Through this gamma conversion, it is possible to select in what color space the grid points storing the color conversion values are set. For example, when a log curve for density conversion is set in the gamma conversion table memory, the lattice point spacing becomes denser in the shadow part of the image due to the characteristics of the density conversion, resulting in better interpolation characteristics in the shadow part. The gamma-converted input signal is represented by (x, y, z). The input signals (125 to 127) are upper signals (x1, y1, z1) and lower signals (x2, y
2, z2). These three types of signals are collectively represented by (107) and (108) in FIG. 1, respectively. The upper signal (107) and lower signal (108) may have non-uniform bit division. The lower signal group (108) is input to weighting coefficient providing means (109-111). With these three weighting coefficient providing means, the lower order signals (x2, y2, z
When 2) is input, one unit tetrahedron (unit 0 to unit 5) to which it belongs within the unit cube is determined, and weighting coefficients W2 to W4 for using the four vertices are provided. This weighting coefficient is originally a number between 0 and 1, but this is the signal that is output as a number between 0 and 255 (128 to 13
0). The lower signal group (108) is also input to the unit tetrahedron selection means (132), and one of the unit tetrahedrons to which it belongs is selected from six units 0 to 5.
It is output as a unit selection signal (133) of 3-bit numerical value up to 5.

On the other hand, the upper signal group (107) and the unit selection signal (133) are connected to the address bus interface (11).
3) via table memory for storing color conversion value difference values (
115 to 117). Similarly, the upper signal group (1
07) is a table memory (118) for storing color conversion value reference values.
) is also entered. This is achieved by setting one end point of the diagonal axis of the unit solid at the reference vertex of the six unit tetrahedra in the division method into the unit tetrahedron shown in Figure 8. Because units can share a reference vertex (
This is because the unit selection signal (133) that can be input to the input signal (118) can be omitted.

The address interface connects to the address bus (114) when downloading data from the host computer to the table memory for storing color conversion value difference values and the table memory for storing color conversion value reference values, and when executing color conversion. The upper signal (107) is switched. When downloading data, the table memory for storing color conversion value difference values and the table memory for storing color conversion value reference values are
All combinations of the upper signal (107) and unit selection signal (133) are generated as addresses, and the color conversion value reference value O1 and color conversion value difference value D2-D4 at the unit tetrahedron reference vertex determined at that time. is calculated and the data is loaded. Next, when executing color conversion, the upper signal (10
7) and the unit selection signal (133), the color conversion value reference value O1 signal (137) and the color conversion value difference value D2 to D4 signals (134-136) at the four vertices are output. Ru. The data bus interface (119) is
The direction of data flow is changed when downloading data to the table memory for storing color conversion value reference values and the table memory for storing color conversion value difference values and when executing color conversion. Weighting coefficients W2 to W4 and color conversion value reference value O1 for three vertices
and the color conversion value difference value D2-D4, the multiplier (121) and the adder (122) execute (Equation 1). However, since the weighting coefficient is multiplied by 255 from the original value, the adder output is finally shifted to the right by 8 bits and a calculation of 1/255 is performed to obtain an 8-bit color signal. The obtained color conversion interpolated value (138) generates an output value R' (124) via an output LUT (123). Various nonlinear curves can be loaded into the output LUT depending on how this color conversion device is used.

Note that in this embodiment, the bit precision, bit allocation, etc. are only one example, and other numerical values may be used. In addition, in this embodiment, since the calculation data is complex, multipliers (121-0 to 121-0 to
121-2) and adders (122-0 to 122-2).
39).

[0029]

[Effects of the Invention] As described above, the present invention has three inputs and three outputs.
It is possible to process any color conversion of any dimension or any dimension or more in real time using an interpolation method without discontinuities.

At this time, since the conventional three-dimensional color signal interpolation method stores the output value for the input value in the memory contents,
There was a problem that multiplication was used four times in the interpolation formula, but in the present invention, since the reference signal and the difference signal of the output color signal are stored in memory, the number of multiplications in the interpolation formula can be reduced to three. It has the great advantage of being able to

Furthermore, since the dynamic range of the numerical values stored in the output value difference value memory becomes smaller, the effect of reducing the storage bit width or the effect of effectively utilizing the bit width can be expected. Furthermore, by dividing the input color space into unit cubes, there is an advantage that the table memory can be effectively used for visual characteristics.

[0032] Even when the calculation formula for color conversion changes, the color conversion value storage table can be rewritten, and since the interface with the host computer is fully equipped, it is possible to respond immediately. It also has versatility that is not found elsewhere, and is extremely effective in the field of color image processing.

[Brief explanation of the drawing]

FIG. 1 is a block diagram of a color conversion device according to an embodiment of the present invention.

[Figure 2] Conceptual diagram showing the conventional one-dimensional interpolation method

[
Figure 3: Diagram showing the memory contents at the vertices of a unit tetrahedron in conventional three-dimensional interpolation

[Figure 4] Division diagram that divides the entire conventional color space into unit solids

[Figure 5] Division diagram showing how to divide a unit solid into unit tetrahedrons (units 0 to 5)

[Figure 6] Division diagram for dividing a unit tetrahedron into small facets

[Figure 7]
Partition diagram that divides the entire color space into unit cubes

[Figure 8] Division diagram showing how to divide a unit rectangular parallelepiped into unit tetrahedrons (units 0 to 5)

[Explanation of symbols]

104 Gamma conversion table memory (red) 105
Gamma conversion table memory (green) 106
Gamma conversion table memory (blue) 109 Weighting coefficient W2 providing means 110 Weighting coefficient W3 providing means 111 Weighting coefficient W4 providing means 113 Address bus interface 115 Table memory for storing color conversion value difference value D2 (rewritable) 116 Color conversion value difference value D3 storage table memory (
(Rewritable) 117 Color conversion value difference value D4 storage table memory (
(Rewritable) 118 Table memory for storing color conversion value difference value O1 (
(Rewritable) 119 Data bus interface 121-0
Multiplier 121-1 Multiplier 121-2 Multiplier 121-3 Multiplier 122-0 Adder 122-1 Adder 122-2 Adder 123 LUT for output value 132 Unit tetrahedron selection means 139 Arithmetic function setting means

Claims (2)

[Claims] Claims: 1. A set of three-dimensional or arbitrary-dimensional color signals is input, and each input signal is divided into upper and lower signals with a different number of bits allocated or with equal bit allocation,
A table memory for storing color conversion value difference values selected in the set of upper signals, a table memory for storing color conversion value reference values selected in the set of upper signals, and a weight selected in the set of lower signals. coefficient providing means; means for selecting a plurality of unit tetrahedrons obtained by dividing a unit cube or unit rectangular parallelepiped in a color space selected by the upper level signal; and color conversion values at a plurality of vertices of the unit tetrahedron. calculation means for performing three-dimensional or arbitrary dimension interpolation of a color space from the difference value and color conversion value reference value and the weighting system value; and calculation function setting means for checking the calculation value of the calculation means. A color conversion device comprising: 2. When the input signal is three-dimensional, the unit solid specified by the upper level signal is a plurality of units 4 that share the diagonal axis.
The tetrahedron is divided into tetrahedrons, and the color conversion value difference value is stored at one vertex position forming the diagonal axis, and the color conversion value difference value is stored at the remaining vertex positions of each tetrahedron. The color conversion device according to claim 1.