JPH05225299A - Color conversion system - Google Patents

Abstract

PURPOSE:To obtain correspondence between color data in a computer and external color data, to substitute color information on an outward field with ideal color information in the computer, and to convert the color information in the computer into a natural color and output it. CONSTITUTION:This system consists of an image input part 1 which inputs external image information, a color conversion correspondence table 10 stored with the correspondence relation between external colors and internal colors in advance, a 1st color conversion part 11 which makes color characteristics of the external input image correspond to the ideal color characteristics in the computer by referring to the color conversion correspondence table 10, an image processing part 2 which performs image processing by receiving the output of the 1st color conversion part 11, and a 2nd color conversion part 12 which receives the output of the image processing part 2 and performs color conversion for approximation to external colors by referring to the color conversion correspondence table 10, and an image output part 3 which outputs the output of the 2nd color conversion part 12 as an image.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention replaces external color information with ideal color information inside a computer (apparatus), or color information inside the computer in various image processing systems, computer graphics systems, etc. that handle color images. The present invention relates to a color conversion system that provides a means for converting a color into a natural color.

[0002]

2. Description of the Related Art FIG. 13 is a conceptual diagram of the configuration of a conventional device.
The input image is converted into an electric signal by the image input unit 1 and sent to the image processing unit 2. The image processing unit 2 performs predetermined image processing. The result of the image processing is output as an image from the image output unit 3. Here, as the image output unit 3, for example, a printer or a graphic display is used.

By the way, generally, color data handled in a computer is expressed as coordinates in a three-dimensional space formed by collecting (R, G, B) three primary color signals of red (R), green (G), and blue (B). Often expressed. Since the R, G, B signals are adjusted according to the signal level inside the computer, the colors generally perceived by humans in the outside world and the ideal color data handled inside the computer do not match, and This causes a problem that the color of the image generated by graphics or the like lacks in naturalness, or it is difficult to identify the color of the color image input from the camera or the like.

When the color data of the outside world is handled inside a computer, for example, when identifying a color, a color that a human can perceive as "red" is also affected by the signal characteristics of the camera system, noise, etc. It will be different from the ideal "red" data stored internally, and correct color identification will not be possible.

Conversely, in computer graphics or the like, when color data is generated by a computer, it may be necessary to associate the (R, G, B) data with the color sample to be generated. Such association requires a great deal of effort because the operator must adjust each level of R, G, and B by trial and error while looking at the color sample. It would be convenient if such a conversion map could be generated flexibly and quickly according to the input processing system and the processing target.

[0006]

The problems of the above-mentioned prior art can be summarized as follows. (1) When handling the color of a color image input from a camera or the like, it is not possible to perform proper image processing because it is out of alignment with the ideal color data inside the computer. (2) When the ideal color data inside the computer is output to a display or the like, it is not possible to display it in a natural color due to the influence of the characteristics of the display screen surface. (3) In order to solve such a problem, the input and output images are associated with the color data in the computer, but the work is performed by trial and error, which requires a great deal of labor.

The present invention has been made in view of the above problems, and the color information of the outside world is converted to the ideal color information of the inside of the computer by associating the color data inside the computer with the color data outside. It is intended to provide a color conversion system capable of converting the color information inside the computer into a natural color and outputting it to the outside.

[0008]

FIG. 1 is a block diagram showing the principle of the present invention. The same parts as those in FIG. 13 are designated by the same reference numerals. In the figure, 1 is an image input unit for inputting image information (test chart 14 here) from the outside, 10
Is a color conversion correspondence table in which the correspondence between external colors and internal colors is stored in advance, and 11 is the color conversion correspondence table 10
Referring to, the first color conversion unit 2 for correlating the color characteristics of the image input from the outside with the ideal color characteristics inside the computer receives the output of the first color conversion unit 11. An image processing unit 12 for performing image processing by receiving the output of the image processing unit 2 and referring to the color conversion correspondence table 10 to perform a color conversion for approximating an external color. ,
An image output unit 3 outputs the output of the second color conversion unit 12 as an image.

[0009]

In order to automatically or semi-automatically associate the color characteristic of the image input from the outside with the ideal color characteristic in the computer, the test chart 14 having the predetermined color characteristic is input from the camera or the like. Then, the computer stores the color correspondence table in the previous color conversion correspondence table 10 based on the color characteristics known in advance by the computer. As a result, the external color information can be replaced with the ideal color information inside the computer, and when the internal color information is output to the outside, a natural color can be reproduced by the same operation.

[0010]

Embodiments of the present invention will now be described in detail with reference to the drawings.

FIG. 2 is a configuration block diagram showing an embodiment of the present invention. The same parts as those in FIG. 1 are designated by the same reference numerals. In the figure, the color conversion correspondence table 10 is composed of a correspondence table creation unit 20 that creates a color conversion correspondence table and a correspondence table storage memory 22 that stores the created table. A conversation processing unit 13 displays the image data sent from the image output unit 3 and performs conversation processing with the operator. First and second color conversion unit 1
Reference numerals 1 and 12 have a function as a color conversion filter for adjusting the external color and the internal color.

The operation of the thus constructed apparatus will be described below.

The features of the present invention are as follows. Determining which of the ideal color data inside the computer the colors read from outside correspond to. Obtain the colors between the colors attached to the test chart 14 by interpolation. The operator can interactively correct the finished color.

FIG. 3 is a flow chart showing the color conversion procedure according to the present invention. The sequence shown in the figure is for the color conversion unit 1.
1 or 12 is shown. Here, the color conversion unit 1
The operation of No. 1 will be described. The color conversion unit 11 reads the corresponding table from the correspondence table storage unit 22 (S
1) Input color data of each pixel of the input image (S2). This color data is color-converted by the table (S3), and the converted color data is output (S4). This output image is input to the next image processing unit 2.

FIG. 4 is a flow chart showing a procedure for creating a correspondence table (hereinafter referred to as a learning procedure) according to the present invention. First, a test chart image is input from the image input unit 1 (S1). The input test chart image enters the correspondence table creating unit 20. The correspondence table creating unit 20 automatically cuts out each color sample portion from the image-inputted test chart (S2). Then, the color data of each pixel in the sample is stored in the correspondence table storage memory 22 as a learning sample (S3).

Further, with respect to the colors not existing on the test chart, learning samples are automatically generated and added by interpolating the learning sample data already stored (S4). Using this learning sample data, the color conversion unit 11 refers to the correspondence table storage memory 22 and automatically adjusts it so that it has the same input / output characteristics as the learning sample (S5). The weight data of the learned color conversion unit 11 is stored in the correspondence table storage unit 22 as a correspondence table (S6).

FIG. 5 is a block diagram showing a detailed configuration example of the embodiment of the present invention. The same parts as those in FIG. 2 are designated by the same reference numerals. As the image input unit 1, for example, a scanner 3
A general-purpose image input device such as 1 or camera 32 is used. Three
Reference numeral 3 is a storage memory for temporarily storing image data. The correspondence table creating unit 20 includes a test chart automatic cutout unit 21, a test chart template storage memory 2
6, a learning sample data creation unit 23, a learning sample data storage memory 27, a saturation / hue calculation unit 29, a sample interpolation unit 24, a learning unit 25, and a working memory 28. 22 is the above-mentioned correspondence table storage memory. The operation of the apparatus configured as described above will be described below.

Test chart template storage memory 2
6 stores the color sample of what color name is presented at which position on the test chart, and the contents are as shown in FIG. (A) shows the configuration of the test chart, and (b) shows an example of the test chart template. In FIG. 10A, the horizontal axis represents the hue order and the vertical axis represents the saturation order. Then, a test chart having a predetermined size is arranged in the area where the hue and the saturation intersect.

In the figure, each color name Cname (Munsell index, etc.) and the coordinates of two points (X ₁ ,
Y ₁ ), (X ₂ , Y ₂ ). Any color sample may be arranged on the test chart, but here, it is assumed that the color samples are arranged at a constant hue interval and saturation interval like the Munsell index. Regarding the color name, it is also possible to use ideal (R, G, B) data corresponding to each color name instead. Here, it is assumed that there are N color samples on the test chart.

FIG. 6B is a diagram showing an example of the test chart template, and there are N samples as described above. The color names are Cname1, Cname2, ...
There are N names up to CnameN. The ideal color data and the chart position of each color sample are shown by X and Y coordinates.

The test chart automatic cutout unit 21 cuts out a sample on the test chart image input based on the information in the template storage memory 26. In this case, the coordinate position actually stored in the template storage memory 26 may be displaced.

Therefore, as shown in FIG. 7, a range expanded by the periphery D of the coordinates of the template is searched. In the figure, A indicates an ideal position and B indicates an actual position. As shown in (b) of FIG. 6, the coordinates of the ideal position are (X
₁ , Y ₁ ), (X ₂ , Y ₂ ). The part surrounded by the broken line in the figure is the search range Q enlarged by D in the vertical and horizontal directions from the ideal position. The color data (R, G, B) is checked for all the pixels within the search range Q, and clustering is performed according to the shortest distance.

FIG. 8 is a flowchart showing a procedure for cutting out a color sample. This procedure is to recognize the pixels included in this cluster as sample pixels if they are clustered within a certain range in the (R, G, B) space and the number of pixels is M or more. Furthermore, the size of the rectangle (| X1
The values of −X2 | and | Y1-Y2 |) are applied, and pixels outside the rectangle are excluded from the target.

First, the test chart automatic cutting section 21
Reads the position (X ₁ , Y ₁ , X ₂ , Y ₂ ) from the template by referring to the template storage memory 26 (S
1). Next, within the search range (X ₁ -D, Y ₁ -D, X ₂ +
All the images in D, Y ₂ + D) are read (S2). next,
As an initial setting, the cluster K = 0 is set (S3).

Next, for pixel i and pixel j, when i belongs to cluster K, the distance {(Ri-Rj) ² + (Gi-Gj) ² + (Bi-B
j) ² } ^1/2 <T (T is a threshold value), the pixel j is included in the cluster K (S
4). Next, the cluster K is updated by 1 (S5).

Next, the number of pixels in the cluster K is M (threshold value).
If it is above, this cluster is set as a sample data area (S6). Then, if the cluster K is the sample data area, the positions of the pixels in K are measured, and the minimum rectangle size (SX, SY) surrounding K is obtained (S7). Next, if SX <| X1-X2 | and SY <| Y1-Y2 |, all pixels in K are regarded as learning sample data. Otherwise, | X1-X2 |, | Y1-Y
The 2 | rectangle is brought to the center of (SX, SY), and the pixels within this rectangle are regarded as learning sample data (S8).
Next, the rectangular position indicating the area of the learning sample data is stored in the template storage memory 26 (S9).

The learning sample data creation unit 23, for the pixels in the sample locations searched by the test chart automatic cutout unit 21, each color data (R, G, B) and the color name (color data obtained from the template storage memory 26). ) Cn
learning sample data storage memory 27 with ame set
To store.

In the saturation / hue calculation unit 29, the color data (R, G, B) is converted into (luminance Y, saturation C, hue H) or vice versa (Y, C, H) is converted into (R, G). G, B) are converted into coordinates, and the following publicly known calculation formulas are used for convenience.

[0029]

[Equation 1]

FIG. 9 is a diagram showing the relationship between saturation and hue.
In the figure, the horizontal axis represents RY and the vertical axis represents BY. Hue H is the rotation angle from the R-Y axis
Indicates the distance from the origin at that angle to (RY, BY).

The sample interpolating unit 24 also generates learning sample data for colors other than the samples presented on the test chart. FIG. 10 is a flowchart showing the processing procedure of the sample interpolation unit 24.

First, the average value of (Y, C, H) is calculated for the learning sample data existing on the test chart (S1). Then, the calculated average value is used as the working memory 28.
(S2). The above operation is repeated N times.

FIG. 11 is a diagram showing the state of the working memory 28. The horizontal axis indicates the hue order, and the vertical axis indicates the saturation order. Color names are arranged for each type of Y, C, and H. saturation·
When the hue data is arranged in the working memory, the contents of the working memory have the same saturation and saturation as the sample arrangement on the test chart.
The order of hue is used, and the saturation and hue coordinates are written in the table. This saturation / hue coordinate value is represented by the average value for each learning sample of each color name.

Then, the sample interpolating unit 24 interpolates the sample based on the working memory 28. Here, a case of interpolating between the color names Cname1 and Cname2 will be described. Saturation hue coordinates (C1, H1) written in the location of Cname1 on the working memory, and Cname
Assume a line segment connecting between the two saturation hue coordinates (C2, H2). For the color name Cname3 that is just in the middle between Cname1 and Cname2, the saturation hue coordinates are obtained as follows.

[0035]

[Equation 2]

Similarly, a color internally dividing between Cname1 and Cname2 into m: n is obtained as follows.

[0037]

[Equation 3]

Since the (R, G, B) data cannot be restored from the saturation and the hue, the interpolation data obtained by the equation (5) are internally divided points of the luminances Y1 and Y2 of Cname1 and Cname2. Is calculated, and it is used as the virtual brightness Y3, and the inverse calculation of the expressions (1) to (3) is performed. The set of (R, G, B) and color name of the interpolation data thus obtained is added to the learning sample data storage memory 27.

In the learning section 25, when the color data on the input side is input by this learning sample, a conversion filter which outputs the learned color name (or color data) is constructed. A known back propagation learning method or the like can be used as a means for automatically configuring the conversion filter. The weight of the network learned in this way is stored in the correspondence table storage memory 22.

A network in the opposite direction to the network created by the above method is also created. This is to learn by inverting the data on the input / output side of the learning sample data, and the principle is exactly the same as above. The weight of this inverse transformation network is also stored in the correspondence table storage memory 22.

As described above, the color conversion unit 11 outputs the color name (or the internal side (R, G, B)) when the (R, G, B) value on the external side is input. The network weight data is read from the correspondence table storage memory 22. Similarly, the inverse color conversion unit 12 configures a neural network that outputs (R, G, B) on the external side when the color name on the internal side (or the internal (R, G, B)) is input. Then, the weight data is read from the correspondence table storage memory 22.

FIG. 12 is a block diagram showing an example of the configuration of the color conversion section 11, showing a neural network. The neural network is composed of an input layer, an intermediate layer and an output layer. Color data input from outside is RE,
If GE and BE, these color data are I1, I2, I
Enter the input layer as 3. The output Mi of the intermediate layer is {f (x) = 1 / (1 + exp (-
x))},

[0043]

[Equation 4]

It is represented by On the other hand, the output Oi of the output layer is

[0045]

[Equation 5]

It is represented by

The conversation processing 13 is composed of a normal color graphic display or the like, and is equipped with an instructing mechanism (mouse, keyboard, etc.) by which an operator can instruct a position on the screen and an instruction. The conversation processing unit 13 reads the conversion table (network weight) from the correspondence table storage memory 22, converts appropriate color data according to the instruction of the operator, and displays the result on the screen. If the operator sees the result and determines that it is not appropriate, the operator adds the set of color data and color name (or internal color data) at that time to the correspondence table storage memory 22 as a learning sample.

Learning sample data creating section 23 and learning section 2
When the execution instruction is given from the operator through the conversation processing unit 13, the No. 5 executes the same process as the automatic learning process described above. In the conversation processing unit 13, the test chart image is displayed on the display as it is by using only the learning sample correction function, and the operator cuts out the sample while watching the test chart image to create the learning sample. Is also possible.

[0049]

As described above, according to the present invention, in various image processing systems, computer graphics systems, etc. that handle color images, the external color information is replaced with ideal color information inside the computer, or the computer It is possible to convert the internal color information into a natural color.

That is, according to the present invention, a color conversion correspondence table for converting external and internal color information can be created almost automatically. By using this color conversion correspondence table, when handling colors of a color image input from a camera or the like, there is no deviation from ideal color data inside the computer, so image processing such as color identification can be simplified. When the internal ideal color data is output to a display or the like, it is possible to display in natural colors. As described above, according to the present invention, the correspondence between the color data inside the computer and the color data outside is substituted, and the color information in the external world is replaced with the ideal color information inside the computer,
It is possible to provide a color conversion system capable of converting color information inside a computer into a natural color and outputting it to the outside.

[Brief description of drawings]

FIG. 1 is a principle block diagram of the present invention.

FIG. 2 is a configuration block diagram showing an embodiment of the present invention.

FIG. 3 is a flowchart showing a color conversion procedure according to the present invention.

FIG. 4 is a flowchart showing a procedure for creating a correspondence table according to the present invention.

FIG. 5 is a block diagram showing a detailed configuration example of an embodiment of the present invention.

FIG. 6 is a diagram showing an example of contents of a test chart template storage memory.

FIG. 7 is a diagram showing a search range for color samples.

FIG. 8 is a flowchart showing a procedure for cutting out color sample data.

FIG. 9 is a diagram showing a relationship between saturation and hue.

FIG. 10 is a diagram illustrating a processing procedure of a sample interpolation unit.

FIG. 11 is a diagram showing a state of a working memory.

FIG. 12 is a block diagram illustrating a configuration example of a color conversion unit.

FIG. 13 is a conceptual diagram of a configuration of a conventional device.

[Explanation of symbols]

 1 image input unit 2 image processing unit 3 image output unit 10 color conversion correspondence table 11 color conversion unit 12 color conversion unit

─────────────────────────────────────────────────── ─── Continuation of the front page (51) Int.Cl. ⁵ Identification number Office reference number FI technical display location G09G 5/00 X 8121-5G 5/02 9175-5G

Claims (6)

[Claims] 1. An image input unit (1) for inputting image information from the outside, a color conversion correspondence table (10) for preliminarily storing a correspondence relationship between an external color and an internal color, and the color conversion. A first color conversion unit (11) for associating a color characteristic of an image input from the outside with an ideal color characteristic inside the computer with reference to the correspondence table (10); An image processing unit (2) that receives the output of the conversion unit (11) and performs image processing, and an output of the image processing unit (2) that refers to the color conversion correspondence table (10) It is composed of a second color conversion unit (12) that performs color conversion to approximate a color and an image output unit (2) that outputs the output of the second color conversion unit (12) as an image. Automatically correlate the color characteristics of the input image with the ideal color characteristics in the computer. In order to perform semi-automatically, a test chart (14) whose color characteristics are determined in advance is input from a camera or the like, and the color correspondence table is converted to a previous color conversion correspondence table ( 10) A color conversion device characterized in that it is stored. 2. A table of characteristics opposite to those recorded in the color conversion correspondence table (10) is automatically created in order to associate an ideal color characteristic in the computer with the image output section (3). The color conversion device according to claim 1, wherein the color conversion correspondence table (10) is stored. 3. The color conversion according to claim 1, wherein in order to automatically create the color conversion correspondence table (10), the position of the corresponding color sample is automatically searched for based on predetermined coordinates. apparatus. 4. In order to semi-automatically create the color conversion correspondence table (10), a test chart image is displayed on the display, and the operator displays the position on the chart and its color name (or ideal 2. The color conversion device according to claim 1, wherein a color data value) is designated. 5. Correspondence between external color data and internal color data is automatically obtained in the form of a primary conversion matrix or a neural network (non-linear conversion) based on the color sample obtained using the color chart. The color conversion device according to claim 1, wherein the weight data is stored in the color conversion correspondence table (10) as a correspondence table. 6. A conversion color is simulated using the correspondence table obtained by the learning means according to claim 5, is displayed on a display, and an operator indicates a portion where the correspondence is determined to be inappropriate, 2. The color conversion device according to claim 1, wherein the sample is modified and learning is repeated again.