JP2008103871A - Color difference operation system for multicolor printing plate printing - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To calculate color differences of higher precision by a color difference arithmetic system for multicolor printing plate printing, used for color matching based upon color differences between plate making data and printed matter, by making it possible to compare the plate making data with the printed matter printed in a different environment with the same tint. <P>SOLUTION: A color correcting filter 33 with RGB-corresponding weight coefficients obtained by making an NNW (neural network) 32 learn an object region of RGB reference data by using RGB color sample data as a teacher signal is generated, and RGB correction reference data are obtained by making the NNW 42 learn RGB reference data of the plate making data 13 by using the respective weight coefficients of the color correction filter 33. A color different arithmetic means 27 compares comparison areas of the RGB correction reference data with corresponding comparison areas of RGB colorimetric data to calculate and digitize respective color differences. <P>COPYRIGHT: (C)2008,JPO&INPIT

Description

  The present invention relates to a color difference calculation system in multicolor printing that is used when color matching is performed based on a color difference between plate making data and a printed matter printed by a multicolor printing press.

  In recent years, multi-color printing has become the mainstream, and in multi-color printing machines, it is the actual situation to perform color adjustment for color matching with plate-making data according to each printing machine due to its characteristics. As a technique for performing matching, it is known to calculate a color difference based on colorimetric data of a printed matter. In such a color difference calculation, a highly accurate color difference calculation is required by comparing the plate-making data to be compared with the printed matter printed by the printing machine based on the plate-making data with the same hue.

  Conventionally, the color of the created image data (plate-making data) and the colorimetric data obtained by capturing the printed matter printed on the basis of the image data with a camera are shades depending on the environment of the equipment used, the light source, etc. Therefore, when the same object is photographed with different cameras, images of different shades are obtained, and it is not preferable to compare these images for color difference calculation. Therefore, when performing color correction on image data or the like created in different environments, a profile is generally used for each device. By using this profile, images created by different devices can be compared with the same color.

  Incidentally, a system for performing color correction in a color copying machine using a neural network (hereinafter abbreviated as “NNW”) is disclosed in the following patent documents. The color correction processing in the patent document is for converting an input signal RGB from a color scanner of a printed document or a photographic paper photographic document into an output signal CMYK to a color copying machine. Conversion (RGB → Lab) is set to two NNW input models for printing originals and photographic paper originals, CMY signals (Lab → CMY) in conversion of the printer unit are set to NNW output models, and conversion of the printer unit The K signal (Lab → K) is used as a black generation model, and these are used as a basic configuration, and color correction is performed to eliminate color errors between a printed document or a photographic paper photograph document and a copy copied by a color copying machine. Is.

Sharp Technique No. 76, April 2000, pages 15 to 19, "Neural Network Applied Color Correction System"

  However, in order to prepare a profile in accordance with the environment as described above, there is a problem that equipment for creating a profile and knowledge of profile creation are necessary, and human skills must be relied upon. The above-mentioned patent document is for color reproduction by color correction to eliminate color error between a printed document or a photographic paper photographic document and a copy copied by a color copying machine. Since there is no significant difference in hue between colors, and no correction is made to match each hue, the method disclosed in this patent document is printed in a different environment. It cannot be applied to the correction for adjusting the hue when the image is processed.

  Therefore, the present invention has been made in view of the above problems, and a multicolor printing plate capable of comparing a plate-making data and a printed matter printed in a different environment without using a profile with the same hue and calculating a more accurate color difference. An object is to provide a color difference calculation system in printing.

  In order to solve the above-mentioned problems, the invention of claim 1 calculates the color difference for final color adjustment when performing multicolor printing using plate-making data created with each reference ink color. A color difference calculation system in multicolor printing that performs mechanical overall processing on an area and outputs color difference data, and converts the plate making data into a reference wavelength color to perform processing as RGB reference data Conversion means; imaging means for picking up a color sample and outputting RGB color sample data of a reference wavelength color; and colorimetric measurement of a reference color color RGB colorimetric data by measuring a multicolored printed material based on the plate-making data. A colorimetric means for outputting, and RGB color sample data obtained by imaging with the imaging means as a teacher signal, and a target area of the RGB reference data converted by the reference wavelength color conversion means, a neural network Color correction file creating means for performing processing for creating, as a color correction file, a weighting factor corresponding to each RGB based on the RGB color sample data among the output values obtained as a result of learning, and the RGB The RGB reference data of the plate making data converted by the conversion means is input to a neural network using the respective weighting factors of the created color correction file, and the RGB reference data corrected with the weighting factors as an output result Reference data correction means for performing correction reference data processing, a comparison area for RGB correction reference data obtained by the reference data correction means, and a comparison area for RGB color measurement data output from the color measurement means, respectively And a color difference calculation means for performing a process of comparing, calculating and digitizing each color difference. To.

  According to the present invention, RGB color sample data is used as a teacher signal, and color correction files of weight coefficients corresponding to the respective RGB obtained by learning the target areas of the RGB reference data by the neural network are created. RGB correction reference data is obtained by learning the RGB reference data of the data by using a weighting factor of each color correction file by a neural network, and the comparison area of the RGB correction reference data is made to correspond to the comparison area of the RGB color measurement data. By comparing each color difference and calculating and digitizing each color difference, the color difference calculation with the RGB colorimetric data is performed with the RGB correction reference data obtained by correcting the RGB reference data obtained by the neural network based on the color sample data. Therefore, plate making without using a profile It is possible to compare the printing product of printing by chromatography data with different environments in the same hue, in which it is possible to calculate a more accurate color differences.

Hereinafter, the best embodiment of the present invention will be described with reference to the drawings.
FIG. 1 is a block diagram of a color difference calculation system according to the present invention. In FIG. 1, the color difference calculation system 11 includes at least a processing unit 12, an imaging unit 13, and a color measurement unit 14, and appropriately includes a display unit 15. That is, the color difference calculation system 11 according to the present invention calculates the color difference for final color adjustment on the work memo area when performing multicolor printing using plate-making data created with each reference ink color. This is a mechanically integrated process for outputting color difference data.

  Here, the image pickup means 13 picks up a color sample and outputs RGB color sample data of a reference wavelength color, and a CCD camera, for example, is used. The colorimetric means 14 measures the color printed matter based on the plate-making data input to the processing means 12 and outputs RGB colorimetric data of the reference wavelength color. Here, a CCD line camera is used. Is done. The display unit 15 is, for example, an LCD or CRT for displaying an image such as plate-making data by the processing unit 12. The plate-making data 16 is a source for multi-color printing, and is a single whole image data obtained by synthesizing block data created with each reference ink color (CMYK). For example, a predetermined storage medium And input to the processing means 12.

  The processing means 12 is, for example, a configuration included in a control computer for a printing press, and inputs the plate-making data 16 for color difference calculation. The input / output means 21, control means 22, memory 23, reference An RGB conversion unit 24, a color correction file creation unit 25, a reference data correction unit 26, and a color difference calculation unit 27, which are wavelength color conversion units, are provided.

  The input / output unit 21 inputs RGB color sample data of the color sample imaged by the image pickup unit 13, RGB colorimetric data obtained by measuring the color of the printed material based on the plate-making data 16, and the plate-making data 16. Further, it is for transmitting predetermined display image data and the like to the display means 15 and is constituted by, for example, a physical electronic circuit or software for ensuring consistency in exchange of these data.

  The control means 22 controls the system as a whole. For example, a control program is stored in a physical CPU. The memory 23 serves as a work area for developing and executing the programs of each means, and temporarily stores corresponding data and the like during execution processing. For example, the memory 23 is a semiconductor memory and is virtually set. This is a concept including a storage area of a hard disk to be stored.

  The RGB conversion means 24 performs processing for converting the plate-making data 16 created with each reference ink color into a reference wavelength color to obtain RGB reference data. The RGB conversion means 24 has a program for this and is developed on the memory 23. Executed. The color correction file creating means 25 will be described in detail with reference to FIGS. 2 and 3. The RGB color sample data obtained by imaging by the imaging means 13 is used as a teacher signal, and RGB converted by the RGB converting means 24 is used. The target area of the reference data is learned by NNW (neural network), and among the output values obtained as a learning result, each RGB corresponding weight coefficient based on the RGB color sample data is created as a color correction file. It performs processing, and is provided with a program for this purpose, and is expanded on the memory 23 and executed.

  The reference data correction means 26, which will be described in detail in FIG. 4, inputs the RGB reference data of the plate making data 16 converted by the RGB conversion means 24 to the NNW using the respective weighting coefficients of the color correction file. A process for obtaining RGB correction reference data obtained by correcting the RGB reference data with the weighting coefficient as an output result is provided, and a program therefor is developed and executed on the memory 23.

  The color difference calculating means 27 will be described in detail with reference to FIG. 5, but the comparison area for the RGB correction reference data obtained by the reference data correction means 26 and the comparison area for the RGB color measurement data output from the color measurement means 14 are provided. Each color is compared and each color difference is calculated and digitized, and a program for this purpose is provided, which is developed on the memory 23 and executed. The digitized color difference data is output to, for example, a printing machine or a printing control unit that controls the printing machine.

  FIG. 2 shows a conceptual diagram of the color difference calculation process of FIG. The plate-making data 16 created with the reference ink colors CMYK is converted into RGB reference data by the RGB conversion means 24 and inputted to the color correction file creation means 25, while the color outputted by, for example, DDCP (Direct Digital Color Proof). A sample is imaged by the imaging unit 13, and RGB color sample data is input from the imaging unit 13 to the color correction file creating unit 25. The input RGB reference data and RGB color sample data are temporarily stored in the memory 23. In the color correction file creation means 25, the learning reference specifying means 31 scans the RGB reference data and the RGB color sample data, for example, specifies the same coordinates as the learning objects (described in FIG. 3A). ).

  Therefore, the NNW 32 of the color correction file creating means 25 learns the target area specified by the learning target specifying means 31 of the RGB reference data using the RGB color sample data as a teacher signal, and the output value obtained as a learning result Among these, weight coefficients corresponding to RGB based on the RGB color sample data are obtained, and these are performed for all the target areas specified by the learning target specifying means 31, and the obtained weight coefficients corresponding to RGB are obtained as color correction files. 33 (described with reference to FIG. 3B). This color correction file 33 is sent to the reference data correction means 26, where it is temporarily stored.

  On the other hand, the RGB reference data is input to the reference data correction means 26. In the reference data correction unit 26, the input RGB reference data is handled by applying the corresponding RGB corresponding weight coefficient of the color correction file 33 stored in the RGB reference data to the NNW 41 and learning. Are converted to output RGB values to which the weighting factor is applied. Based on the output RGB values, the image data creating means 42 creates RGB correction reference data and sends it to the color difference calculating means 27 (described in FIG. 4).

  In the color difference calculation means 27, the color measurement means 14 measures the color of the printed matter based on the plate making data 16 and inputs RGB color measurement data (imaging data) of the reference wavelength color. Then, the calculation object comparison unit 51 associates the comparison area of the RGB correction reference data with the comparison area of the RGB colorimetric data, compares the comparison areas with each other, calculates the respective color differences, and digitizes them. The color difference data is output to, for example, a printing machine or a printing control unit that controls the printing machine (described in FIG. 5).

  FIG. 3 shows a process explanatory diagram of the color correction file creation means of FIG. FIG. 3A is an explanatory diagram for identifying a learning target, and FIG. 3B is a conceptual diagram of learning by NNW. In FIG. 3A, the learning target specifying unit 31 stores the RGB reference data obtained by converting the plate making data 16 by the RGB converting unit 24 and the RGB color sample data obtained by capturing the color sample by the imaging unit 13 from the memory 23. Call, for example, a portion surrounded by coordinates (200, 200) and coordinates (300, 300) of RGB reference data and a portion surrounded by coordinates (200, 200) and coordinates (300, 300) of color sample data And the target portion of the same coordinates to be learned are specified, and the learning target is specified by sequentially scanning all the coordinate targets.

  Further, as shown in FIG. 3B, the NNW 32 of the color correction file creating unit 25 employs an error back propagation learning method (back propagation method), and as an output value and a teacher signal of the output layer. This is a learning method in which prepared values are compared and the weights of the links between the layers are corrected by using the error, whereby the output value is optimized to obtain the desired color value.

  Here, for example, 255 is divided by the denominator for the RGB value of the RGB reference data as the input value, and converted to a value of 0 to 1. Also, the number of units in the input layer is set to 3 for inputting each RGB value, the number of units in the output layer is set to 3 for the hidden layer, and the number of units in the output layer is set to 3 for outputting each RGB value. The number of hidden layers and the number of nodes are not limited to this example. For example, the number of hidden layers may be 3 and the number of units may be 30.

  That is, first, a random number is given as a real number between -1 and 1 between units. When a value obtained by dividing the RGB value of the reference data by 255 is input to the NNW, calculation is performed in each layer, and output values (0.902, 0.863, 0.824) are once obtained. Since this output value is affected by the weight, it is not always this value. Between the units, the output value approaches the RGB value (200, 200, 200) of the RGB color sample data that is the teacher signal (the teacher signal takes (0.784, 0.784, 0.784)). Correct the weight. By repeating the above input and learning, the output value is (0.872, 0.843, 0.814), (0.852, 0.829, 0.794) each time the weight is corrected. The output value is updated and finally the output value approximates the target RGB value (200, 200, 200), and there is almost no change in weight.

  Then, these learnings are scanned with the coordinates of the respective data as shown in FIG. 3A, and the values of the respective coordinates are set as the input value and the teacher signal, respectively, so that the coordinates (200, 200) and the coordinates ( 300, 300), the learning of the color of the part enclosed is performed. Note that the learning end condition is that the error from the teacher signal is 0.01 or less or that the learning is performed up to a predetermined number of times of learning. After completion of the learning, the weighting coefficient data corresponding to the RGB of the NNW 32 is created as the color correction file 33 and is sent to the reference data correction means 26 here. The color correction file 33 may be stored in the memory 23 and called by the reference data correction unit 26.

  Next, FIG. 4 shows a process explanatory diagram of the reference data correction unit of FIG. In FIG. 4, the NNW 41 provided in the reference data correction means 26 is a hierarchical network having an intermediate layer. Here, the layer configuration, the number of units, and the like are the same as those in FIG. That is, in the NNW 41, an input value is obtained by applying the weight of the color correction file 33 as a weight between units using a value obtained by dividing the RGB value of the RGB reference data obtained by RGB conversion of the plate-making data 16 by 255. The RGB values of (255, 255, 255) are corrected to output RGB values of (200, 200, 200) (output values are (0.784, 0.784, 0.784)).

  Then, the image data creation means 42 creates image data based on the corrected output RGB values, thereby creating RGB correction reference data, which is output to the color difference calculation means 27.

  Next, FIG. 5 shows an explanatory diagram of color difference calculation processing in the color difference calculation means of FIG. 5A and 5B are diagrams for explaining the color difference calculation by the color difference calculation means 25 in detail. The color difference calculation means 27 is RGB colorimetric data (reference wavelength color) measured by the colorimetry means 14 of the RGB correction reference data corrected by the reference data correction means 26 and a multicolor printed material based on the plate making data 16. Imaged data) and the respective color differences are calculated to obtain digitized color difference data.

  That is, in FIG. 5 (A), in the color difference calculation means 27, first, the calculation object comparison means 51, as shown in FIG. 5 (B), the RGB correction reference data 61 created by the reference data correction means 26, The comparison areas are made to correspond to the RGB colorimetric data 62 acquired by the colorimetric means 14 (for example, coordinates (X1, Y1)).

  RGB correction reference data Lab value 61 (X1, Y1) and colorimetric RGB data Lab value obtained by converting the RGB data of the corresponding comparison areas (X1, Y1) into a common color space (for example, Lab color space). By calculating the color difference as 62 (X1, Y1), digitized color difference data is obtained. Similarly, each RGB value for each other comparison area is converted into a Lab value, and the color difference is obtained by calculating the color difference in all the comparison areas and digitizing it. The digitized color difference data is output to, for example, a printer main body or a print control computer.

  In this way, by making the RGB correction reference data obtained by correcting the RGB reference data obtained by the neural network based on the color sample data, the plate-making data and the colorimetric data of the printed matter printed in different environments have the same hue. Therefore, the comparison for color matching between the two can be performed with the same hue without using a profile, and the color difference can be calculated with higher accuracy. .

  The color difference calculation system of the present invention is suitable for a system that calculates a color difference based on colorimetric data of a printed material when performing color adjustment in multicolor printing using a printing plate.

It is a block block diagram of the color difference calculation system which concerns on this invention. It is a processing conceptual diagram of the color difference calculation of FIG. It is processing explanatory drawing of the color correction file preparation means of FIG. It is processing explanatory drawing of the reference | standard data correction means of FIG. It is explanatory drawing of the color difference calculation process in the color difference calculation means of FIG.

Explanation of symbols

DESCRIPTION OF SYMBOLS 11 Color difference calculation system 12 Processing means 13 Imaging means 14 Color measurement means 16 Plate making data 24 RGB conversion means 25 Color correction file creation means 25
26 Reference data creation means 27 Color difference calculation means 31, 41 Learning target specifying means 32, 42 Neural network (NNW)
33 Color correction file 51 Calculation target comparison means

Claims (1)

When performing multicolor printing using plate-making data created with each standard ink color, the color difference calculation for final color adjustment is performed mechanically in the work memo area, and the color difference data is A color difference calculation system for outputting multi-color printing plates,
A reference wavelength color converting means for converting the plate making data into a reference wavelength color to perform RGB reference data;
Imaging means for imaging a color sample and outputting RGB color sample data of a reference wavelength color;
Colorimetric means for measuring the color printed matter based on the plate making data and outputting RGB colorimetric data of a reference wavelength color; and
Using RGB color sample data obtained by imaging with the imaging means as a teacher signal, the target areas of the RGB reference data converted by the reference wavelength color conversion means are learned by a neural network, and obtained as learning results. Among the output values, color correction file creating means for performing processing for creating weight coefficients corresponding to each RGB based on the RGB color sample data as a color correction file;
The RGB reference data of the plate making data converted by the RGB conversion means is input to a neural network using the respective weighting coefficients of the created color correction file, and the RGB reference data is corrected with the weighting coefficients as an output result. Reference data correction means for performing processing for obtaining the corrected RGB correction reference data;
A process of comparing the comparison area of the RGB correction reference data obtained by the reference data correction means with the comparison area of the RGB color measurement data output from the color measurement means, and calculating and digitizing each color difference. Color difference calculation means for performing,
A color difference calculation system in multicolor printing printing, characterized by comprising:

Images