JPH10173939A - Method and device for color information conversion - Google Patents

Abstract

PROBLEM TO BE SOLVED: To convert the color value of a specific color into a color material quantity suitable for color reproduction by generating a color chart, performing color value-to-color material quantity conversion with a color material quantity value corresponding to the color value, and making it correspond to the production standard of a black quantity. SOLUTION: When a conversion system is structured, a color chart generating means 33 generates a color chart 42 and a colorimetric means 46 selects a color material quantity CMYK value 60 in the color chart corresponding to a read-in color value Lab value 64 to structure a LabK-CMY value converting means 48. Here, the Lab value 62 read in by a color value input means 54 is inputted to a Lab-CMYK value converting means 50. The Lab-CMYK value converting means 50 generates a color value LabK of a specific color according to the production standard of a black quantity. Further, it is converted into a corresponding CMYK value 68 through the LabK-CMY value converting means 48. Then a color material quantity output means 56 outputs a CMYK value image.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a color information conversion method and apparatus, and more particularly, to a color information conversion method and apparatus for converting a color information value of a predetermined color into an optimal amount of color material so that the predetermined color can be reproduced. About.

[0002]

2. Description of the Related Art In recent years, a so-called color design for examining colors to be created has been performed before actually creating a copy of an original or an object of many colors. A simulation device that reproduces each color of a document or an object and displays it on a display device, or a print that reproduces each color of a document or an object is used to examine the colors to be created. A color reproduction device such as a hard copy device is known. In this color reproduction device, the color separation values of a large number of originals and objects are converted into device values of a display device, for example, TV signals and digital RGB output values (hereinafter referred to as RGB values), or color materials provided in a hard copy device. Color material amount (hereinafter CMYK)
Value) to display or output. As a result, it is possible to display or output a reproduced color of a multi-color original or a copy of an object.

However, the above-mentioned color separation values depend on the characteristics of an apparatus for specifying the color of a document or an object and the characteristics of an apparatus for converting the color separation values into device values. For example, if a light source or a colorimeter for irradiating a document or an object to specify a color is different, the color of the document or the object has a different color separation value.
Further, if the conversion characteristic to the device value is different, the color of the document or the object is reproduced in a different color. Therefore, conventionally, the color to be reproduced does not depend on the characteristics of the device for specifying the color,
Regardless of the device, in order to express a value representing the same color as the color of a document or the like, the color is once converted to a device-independent color information value (intermediate expression method). Such a color as a device-independent color information value is referred to as CIE (Commission I
CIE X defined by nternationalede I'Eclairage)
There are color values such as YZ values (hereinafter referred to as XYZ values) or uniform perceived color space CIE L ^* a ^* b ^* values (hereinafter referred to as Lab values).

Using color values such as XYZ values or Lab values as color information values, color separation values such as RGB values and CMYK values representing the color of a document or an object are converted into color values, and then color correction is performed. , So-called device-independent color reproduction. As described above, by matching the color values independent of the device, it is possible to reproduce the color with a stable accuracy even for different color reproducing devices for a multi-color original or the like.

[0005]

A color value such as an XYZ value or a Lab value as a so-called device-independent color information value has three values each as an element.
It is determined by the dimensional color space. On the other hand, in a color reproduction device such as a hard copy device, a color value is determined by a color material amount of a color material provided in the hard copy device, usually a color material amount of each color of CMYK (hereinafter, C
MYK value). This CMYK
The value is determined in a four-dimensional space having CMYK colors as elements.
However, conventionally, conversion between a three-dimensional color value and a three-dimensional color material amount (CMY value, RGB value, etc.) can be performed by a least square method, a lookup table method with interpolation, a neural network method, or the like. However, from the three-dimensional color value,
Since the conversion into the dimensional color material amount (CMYK value) is essentially one-to-many and usually not uniquely determined, the amount of K had to be determined by some method.

[0006] Conventionally, UCR, GCR, etc. have been performed, but CMYK obtained by conversion by these methods.
The color reproduced by the value and the color indicated by the color value before conversion are
Colorimetric matching was not guaranteed. In view of the above facts, the present invention provides a color information converter that can determine the amount of CMYK color material determined to reproduce an arbitrary predetermined color in a hard copy device from color values while performing colorimetric constant matching. The aim is to obtain a method and a device.

[0007]

Means for Solving the Problems The first invention is a cyan,
3 representing the predetermined color to be input to a color reproduction device that reproduces a predetermined color by combining and outputting the four color materials of magenta, yellow, and black in predetermined color material amounts, respectively
In a color information conversion device that converts color information values of a dimensional color system into respective color material amounts of cyan, magenta, yellow, and black, a color chart is output by the color reproduction device and color measurement is performed (cyan). , Magenta, yellow, and black) by rearranging the elements of the relationship between the color material amount values and the color values to treat the relationship between (color value, black) and (cyan, magenta, yellow) Color information conversion means for converting any (color value, black) into (cyan, magenta, yellow) based on the three-dimensional color information values of cyan, magenta, yellow, and black. This is a color information conversion device for converting into a color material amount value.

In the second invention, the color information converting means converts (color value, black) to (cyan, magenta, yellow).
The color information conversion device according to the first invention, wherein the color information conversion device is a multilayer feed-forward type neural network learned to convert the color information into a color information. The third invention represents the predetermined color to be input to a color reproduction device that reproduces a predetermined color by combining and outputting four color materials of cyan, magenta, yellow, and black with a predetermined color material amount. In a color information conversion method for converting color information values of a three-dimensional color system into respective color material amounts of cyan, magenta, yellow, and black, the color information is obtained by outputting a color chart by the color reproduction method and performing colorimetry ( By rearranging the elements of the relationship between the color material amount values and the color values (cyan, magenta, yellow, black), it is treated as the relationship between (color value, black) and (cyan, magenta, yellow). Any (color value, black) based on the relationship
A color information conversion method capable of converting color information into (cyan, magenta, yellow), and converting three-dimensional color information values into four color material amount values of cyan, magenta, yellow, and black. Color information conversion method.

According to a fourth aspect of the present invention, in the color information conversion method, (color value, black) is replaced with (cyan, magenta, yellow)
The color information conversion method according to the third invention, wherein the color information conversion method is a multilayer feed-forward type neural network that has been learned to convert the color information into a color information.

[0010]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Hereinafter, an embodiment of the present invention will be described in detail with reference to the drawings. In the following description, the color values are CIE XYZ, CIE L * a * b *,
CIE L * u * v * etc. can be used. In this embodiment, CIE L * a * b * (Lab value) is used. From the spectral distribution (obtained by measurement), the following equation (2) is obtained from the CIEXYZ tristimulus value shown in the following equation (1). Can be derived using

[0011]

(Equation 1)

X, Y, Z: tristimulus values in the XYZ system Xn, Yn, Zn: tristimulus values in the XYZ system of a perfect diffuse reflection surface As shown in FIG. 40
Is a device main body 10 including a microcomputer, a colorimetric device 24 for reading color values of each patch of a color chart, an external storage device 26 for transferring image data and the like to and from an external device, A keyboard 28 for inputting commands and commands, a monitor 30, and a printer interface 32 to which various hard copy devices can be connected. The device body 10 is a CP
U12, ROM 14, RAM 16, memory 18 for storing color information of a color chart, etc., input / output device (hereinafter referred to as I / O) 20 for exchanging data and the like between the main body and other devices, and these And a bus 22 connected so that data and commands can be input and output. The ROM 14 stores a processing program described later. The colorimetric device 24 is a device for outputting colors as data expressed by numerical values, irradiates an object (original document) with light, and measures the intensity of reflected light or transmitted light from the object using the photoelectric principle. It is for measuring. Basically, it consists of a light source and a photometer.

FIG. 1 is a block diagram showing a schematic configuration of each function of a color conversion apparatus 40 according to the present embodiment. The color information conversion device 40 according to the present embodiment includes a color chart creation unit 44, a color measurement unit 46, a LabK-CMYK value conversion unit 48, a Lab-CMYK value conversion unit 50, a color value input unit 54, a color material It has a quantity output means 56, and each of these means is configured using the hardware resources shown in FIG. 2 and the software resources described later. The operation of the color conversion apparatus 40 according to the present embodiment is roughly classified into a conversion system construction and a color information conversion. When the conversion system is constructed, the color chart 42 created by the color chart creation unit 44 is read by the color measurement unit 46. The CMYK value 60 of the color chart corresponding to the Lab value 64 is regarded as the correspondence between the LabK value and the CMY value, and La
The bK-CMY value conversion means 48 is constructed.

As shown in FIG. 3, the LabK-CMY value conversion means 48 comprises a network 74 and a teacher 76. When the conversion system is constructed, the input side of the network 74 is connected together with the input side of the teacher 76, and the output side of the network 74 is connected together with the output side of the teacher 76. The correction signal output side of the teacher 76 is connected to the correction signal input side of the network 74. The network 74 receives the values of LabK and outputs the CMY values. The teacher signal TC corresponding to the input color value and the output signal OC corresponding to the output color separation value are input to the teacher unit 76, and the teacher unit 76 outputs the correction signal SC obtained from the difference between them to the network 74. . That is, when each value of LabK is input to the network 74, CMY
The value is output. At the time of learning, a known color value and a known color separation value corresponding to the color value are input to the teacher unit 76 as a teacher signal TC, and the output signal O corresponding to the output color separation value is input.
C is also input to the teacher unit 76. The teacher unit 76 outputs a correction signal SC obtained from a difference between the input signals to the network 74. After correcting the network based on the correction signal SC, the above process is repeated, and when the correction signal SC becomes sufficiently small, the weight coefficient of the network is output as the conversion system definition data.

An example of the neural network used in the LabK-CMY value conversion means 48 is as follows.
As shown in FIG. 4, an input layer composed of four input units I1, I2, I3, and I4 for inputting respective values of L, a, b, and K, and an intermediate layer composed of a large number of units Mq (q ≧ 1) And three output units U1, U2, U3 for outputting each value of CMY. In this embodiment, each unit of the intermediate layer and each unit of the output layer are connected to a bias unit. The unit of the intermediate layer is composed of neural network elements whose input / output relationship is represented by a sigmoid function, and the unit of the input layer is composed of neural circuit elements whose input / output relationship is linear. Further, the output layer unit is constituted by a linear neural circuit element as described above, but is not limited to this, and the input / output relationship has a sigmoid characteristic as in the intermediate layer unit. May be configured. In the LabK-CMY value conversion means 64, LabK
When a value is given as an input, a unit in the intermediate layer of the network outputs an output Hj according to the following equation (3).

[0016]

(Equation 2)

Here, W ^(h) _ij represents a coupling coefficient (ie, weight) of the i-th unit of the input layer and the j-th unit of the intermediate layer, and I _i is the output of the i-th unit of the input layer (ie, weight). Color information value), b _j represents a bias value supplied from the bias unit. Similarly, the output O _k of the unit of the output layer is expressed by the following equation (4).

[0018]

(Equation 3)

Where W ^(o) _jk represents the coupling coefficient of the j-th intermediate unit and the k-th output unit, and H _j is the output value of the j-th unit of the intermediate layer defined by equation (3). It is. Also, n and m are the numbers of units in the input layer and the intermediate layer, respectively. Therefore, by inputting the color values and the black amount to the units of the input layer, a value of at least three CMY as output O _k from the unit of the output layer are outputted. The characteristics of each unit in the input layer are such that the input is output as it is, and the coupling coefficient of the unit in the intermediate layer is between [0, 1] defined by the following equation (5). Is a characteristic represented by a sigmoid function that monotonically increases.

F (x) = 1 / (1 + e ^−x ) (5) The unit of the output layer is a characteristic represented by a sigmoid function or a linear function. However, the output layer may have an input from the bias unit. The coupling coefficients of the neural network are learned and corrected in advance so as to minimize the error between the output of each unit in the output layer and the teacher signal whose color separation value is known. That is, the conversion from the LabK value to the CMY value is learned. In the present embodiment, a feed-forward coupled neural network is used as a neural network, and there are various methods for learning the network. For example, a back propagation algorithm (Runmelher [Runmelher]
t, DE and McClelland, JL (Eds), "Parallel Distribut
ied Processing ", Exploration in the Microstructure
Vol 1, 2, MIT Press, Cammbridge (1989)])).

Next, the details of the neural network learning process in the LabK-CMY value conversion means 48 will be further described with reference to FIG. During this learning, CMY
K, forming an image of a color chart including 1296 color patches each having a combination of every 20%, measuring the formed image with a colorimeter to obtain a Lab value, and inputting the Lab value to an input teacher. In addition to the data, 1296 sets of CMYK values are used as output teacher data. First, at step 156, 1296 sets of color patches are printed by outputting the color chart stored in the memory 18 to the printer 32. In the next step 158, the 1296 sets of printed color patches are measured for color, and in the next step 160, each color value (Lab value) is obtained and stored. The above processing is the generation of teacher data for learning the neural network. In the next step 162, the color values (Lab values) of the plurality of printed color patches are used as input teacher signals, and the CMYK values of the color patches are input to the neural network as output teacher signals, and learning is performed in the next step 164. That is, learning is performed so as to minimize the error between the network output (output teacher signal) and the teacher signal (input teacher signal). The above processing is completed (step 16
6) After the neural network has been sufficiently learned, the structure and weight of the network are stored as conversion definition data in the external storage device 26 in step 168, and the construction of the conversion system is completed. As described above, in the network defined by the conversion definition data by learning, when the Lab value as the color value and the K value as the black color material amount are input, the color having the input Lab value is reproduced. Among the plurality of combinations of the respective color material amounts to be performed, the CMY values of the combination having the input K value are output.

FIG. 6 is a graph of a learning curve showing an error with the number of times of learning when the number m of hidden units in the neural network is 14. The vertical axis is 129
The average error ΔE of the six colors and the horizontal axis represent the number of times of learning, and are the results of learning from the same teacher data from six different initial values. It can be seen that ΔE converges to less than 2 in all the learnings starting from six different initial values. After constructing the conversion system as described above, the CM
Conversion of YK into four color material amounts becomes possible. As shown in FIG. 1, Lab value 62 read by color value input means 54 is converted to CMYK value 68 by Lab-CMYK value conversion means 50.
And output by the color material amount output means 56. The output destination of the color material amount output unit 56 is the printer interface 32 or the external storage device 26 in FIG.

As shown in FIGS. 7 and 8, Lab-CMY
The K value converting means includes a K amount range determining means 51, a CMYK value storing means 52, and a K amount selecting means 53. The K amount range determining means changes the K value from 0% to 100% in increments of 1% with respect to the given Lab value.
A K value is created, and this is used as LabK-CMY value conversion means 48.
To convert to CMY values and reproduce the given Lab value
101 MYK value sets are obtained and stored in the CMYK value storage means 52. Of these 101, those in which one of C, M, and Y is negative or exceeds 100% are out of the color reproducible range. The K amount selection unit 53 sorts only the ones within the color reproducible range from the set of 101 CMYK values stored in the CMYK value storage unit 52 in ascending order of the K value.

As a result, n of K1, K2,...
The following description is continued assuming that the individual has been obtained. The K amount selecting means 53 outputs the K1, K2,
.., Kn, one K value is selected from the n K values, and the selection method at that time depends on the equation (6) in this embodiment. Medium black content creation criteria is smallest when K = K ₁ If the black amount generating reference maximum when K = K _n black content creation criteria of small ··· (6) K = K n / 4 black content creation reference K = K _{2/2 When the} black amount generation standard is large K = K _{3 * n / 2 The} K amount selecting means 72 outputs a set of CMYK values having the K value selected as described above. Output.

At the time of color information conversion, the color information conversion device 40 converts the Lab image read by the image input means 54 into a CMYK image by the Lab-CMYK value conversion means 50 and outputs the CMYK image from the image output means 56. I do.

[0026]

As described above, according to the present invention, a CM which uses a three-dimensional color information value representing a predetermined color in a color reproducing apparatus.
In the color information value conversion method and apparatus for converting into a YK four-dimensional color material amount, since LabK-CMY value conversion is used, a color value of a predetermined color is converted to a predetermined color according to a generation standard of a black amount. It can be converted into CMYK color material amount values that can be accurately reproduced.

[Brief description of the drawings]

FIG. 1 is a block diagram schematically illustrating a configuration of each function of a color information conversion device according to an embodiment of the present invention.

FIG. 2 is a block diagram illustrating a schematic configuration of a color information conversion device according to an embodiment of the present invention.

FIG. 3 is a conceptual image diagram for explaining the operation of the neural network.

FIG. 4 is a conceptual image diagram for explaining a configuration of a neural network.

FIG. 5 is a flowchart showing a processing procedure when constructing a conversion system of the neural network.

FIG. 6 is a learning curve diagram showing a learning process when constructing a conversion system of a neural network.

FIG. 7 shows Lab-CMYK according to the embodiment of the present invention.
It is a block diagram which shows the outline of a function-specific structure of a value conversion means.

FIG. 8 shows Lab-CMYK according to the embodiment of the present invention.
It is a flowchart which shows the process of a value conversion means.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 10 Device main body 12 CPU 14 ROM 16 RAM 18 Memory 20 Input / output device (I / O) 22 Bus 24 Colorimeter 26 External storage device 28 Keyboard 30 Monitor 32 Printer interface 40 Color information conversion device 42 Color chart 44 Color chart creation means 46 colorimetric means 48 LabK-CMY value converting means 50 Lab-CMYK value converting means 51 K amount range determining means 52 CMYK value storing means 53 K amount selecting means 54 image input means 56 image output means 74 network 76 teacher section

Claims (4)

[Claims] 1. A color reproducing device which reproduces a predetermined color by combining and outputting four color materials of cyan, magenta, yellow, and black in predetermined color material amounts, and representing the predetermined color. In a color information conversion device for converting color information values of a dimensional color system into respective color material amounts of cyan, magenta, yellow, and black, a color chart is output by the color reproduction device and color measurement is performed. , Magenta, yellow, and black) are treated as a relationship between (color, black) and (cyan, magenta, yellow) by rearranging the elements of the relationship between the color material amount values and the color values. Based on (color value, black), (cyan,
Color information conversion means for converting the three-dimensional color information values into four color material amounts of cyan, magenta, yellow, and black. apparatus. 2. The color information converting means comprises: (color value, black)
The color information conversion device according to claim 1, wherein the color information conversion device is a multilayer feed-forward type neural network that has been learned so as to convert into (cyan, magenta, yellow). 3. A color reproduction device that reproduces a predetermined color by combining and outputting four color materials of cyan, magenta, yellow, and black in a predetermined color material amount, and representing the predetermined color. In a color information conversion method for converting color information values of a dimensional color system into respective color material amounts of cyan, magenta, yellow, and black, the color information is obtained by outputting a color chart by the color reproduction method and performing colorimetry (cyan). , Magenta, yellow, and black) are treated as a relationship between (color, black) and (cyan, magenta, yellow) by rearranging the elements of the relationship between the color material amount values and the color values. Based on (color value, black), (cyan,
Magenta, yellow), and a three-dimensional color information value is converted into four color material amount values of cyan, magenta, yellow, and black.
A color information conversion method characterized in that: 4. A color information conversion method comprising the steps of: (color value, black)
The color information conversion method according to claim 3, wherein the color information conversion method is a multilayer feed-forward type neural network that has been learned so as to convert the color information into (cyan, magenta, yellow).