JP2015012422A - Image processing apparatus, color change processing method and computer program - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide an image processing apparatus capable of easily performing predetermined color change processing and obtaining a high-quality result, a color change processing method and a computer program.SOLUTION: On the basis of a color change designation that designates N pairs of conversion source colors on a processing object image and conversion target colors on a reference image, color change processing means determines a function F that converts a color (x1, x2 and x3) appearing on the processing object image displayed in a three-dimensional color space into a color (y1, y2 and y3) after the color change processing so as to minimize a norms of a differential from identity conversion based on the color change designation as an n-th order polynomial ((n) is a natural number equal to or greater than 2) of variables x1, x2 and x3.

Description

The present invention relates to a technology in the field of color conversion in which the colors of an image are aligned in a plurality of images that are often required in a commercial print such as a product catalog.

  In commercial printouts such as product catalogs, when there are multiple photographic images of the same product, if the color is different due to differences in shooting conditions, the same product will look the same color by eliminating the color difference “Product color alignment” issues such as wanting to do, “Background color alignment” issue that wants to align the background color in product photos placed on the same page, one product image based on the product color There is a problem of “color change of product” that it is desired to generate an image showing a variation. In order to deal with these problems (hereinafter collectively referred to as “color change processing”), in the production of printed materials, image processing application software having a retouch processing function represented by Photoshop (registered trademark) is used. Manual processing is often performed one by one, but a method that requires skill and can be easily performed even by an inexperienced worker has been desired.

 Various attempts have been made to deal with such problems relating to color change processing. In Patent Document 1, an image input unit 1 that inputs a processing target image to be processed and a reference image to be referred to for correction processing, a processing target image input by the image input unit 1, and a reference image, Using the color space dividing unit 2 that extracts a processing target region that is a pair of color space conversion and a reference region corresponding thereto, and the processing target region and the reference region that are extracted by the color space dividing unit 2, Based on the processing result in the color space conversion unit 3 that converts the color of the processing target image based on the reference image, and the processing result in the color space conversion unit 3, a predetermined processing is performed on the processing target image input in the image input unit 1. An invention of an image correction system and an image correction method including an image output unit 4 that outputs a processing result image in a format is disclosed.

 In Patent Document 1, affine transformation is used based on the idea of shifting the frequency distribution in the color space to the frequency distribution. A process of converting from the partial color space to the partial color space is performed by applying an affine transformation function for each correspondence of the divided color spaces. However, the continuity of the color conversion function over the entire color space is not guaranteed, so if the color of the processing target image changes in a gradation or the target image has a wide color gamut, the color change becomes discontinuous. It is highly probable that an inconvenient result will occur.

By the way, color reproduction devices such as printers and monitor displays determine colors to be reproduced by three color variables such as (c, m, y) and (r, g, b), but the same (r, g, b) Even if the color is generated, if the display device is different, the stimulus perceived by the human visual system is different and is recognized as a different color. Therefore, a color space conversion function for associating (r, g, b) for each device with the CIE uniform color space, or for directly associating device-specific color spaces (when this function is defined as a linear polynomial of a color variable) There is often a need to obtain a color conversion matrix) which is a coefficient matrix. In particular, it is indispensable to obtain an appropriate color space conversion function in order to realize color management that reproduces (outputs) the same color so that it looks the same even if the devices are different.

Patent Document 2 discloses a method for generating a color conversion matrix for converting a first color space defined by a certain device into a second color space defined by another device. Constructing an nth-order matrix coefficient relating to the first color space of a plurality of color chart data sets, obtaining a pseudo inverse matrix for the constructed matrix, and further providing a color relating to the second color space of the color chart data set This is a method of obtaining a desired color conversion matrix by performing a step of generating a color conversion matrix from the product of a matrix composed of data and the pseudo inverse matrix.

JP 2007-281666 A Japanese Patent Laid-Open No. 2005-110089

  The present invention provides an image processing apparatus, a color change processing method, and a computer program that can easily perform a desired color change process and obtain a high-quality result for any image to be processed. This is the issue. In particular, as in the invention disclosed in Patent Document 1, it avoids the drawbacks of dividing a space and obtaining a conversion function for each subspace, and obtains a single continuous function defined over the entire color space. This presents a method for obtaining a simple and high-quality result image.

The first invention for solving the above-described problems is
A storage unit that records and holds each image data and color change instruction data of the reference image, the processing target image, and the processing result image; and a display unit that displays the reference image, the processing target image, and the processing result image;
Accepted and accepted a color change instruction by specifying one or more color change correspondences that are combinations of a conversion source color that is a specific color on the processing target image and a conversion target color that is a specific color on the reference image Color change instruction input means for recording color change instruction data in the storage unit;
Based on the color change instruction so that the conversion source color on the processing target image is converted into a conversion target color specified in correspondence with the color change, and the color gamut not specified as the conversion source color satisfies a global condition separately defined. A color change processing means for performing continuous color change processing of the processing target image to obtain a processing result image;
An image processing apparatus that performs a color change process of a processing target image based on a reference image including a display unit, an input unit, and a control unit that controls a color change processing unit,
The color change processing means reads the color change instruction data from the storage unit, and changes the color (x1, x2, x3) appearing on the processing target image displayed in the three-dimensional color space to the color (y1, a function F for converting to y2, y3) as an n-order polynomial (n is a natural number of 2 or more) of variables x1, x2, and x3,
The function F as a global condition determined separately is determined based on a condition that the norm of the difference from the identity transformation is minimized,
The color change instruction input means is configured to accept the specified number N corresponding to the color change by a number smaller than the number L of terms of the nth order polynomial, or the color change processing means has the number of terms of the nth order polynomial. An image processing apparatus configured to obtain a function F having a sufficiently large order so that L exceeds a specified number N.

In such an image processing apparatus, since the conversion function F for performing the color change conversion is defined as a single continuous function over the entire three-dimensional color space, the image to be processed has a wide color gamut or a gentle gradation portion. Even if it has, it is possible to obtain a color change processing result image with good quality. The continuous color changing process means a conversion process using a continuous function.

  Further, the problem of obtaining the coefficient matrix of F by the color change instruction due to the relationship between the specified number N of color change correspondences and the number of terms L as a linear polynomial of the function F is the problem of an underdetermined system simultaneous linear equation. It becomes a problem to find a solution. As a result, the function F that reliably copies the conversion source color to the conversion target color cannot be determined by only the color change instruction. However, if all the L color change correspondences are set in order to determine one function F, the added color change correspondence causes some specific colors on the processing target image to remain the same color. Even if it is a specification that seems to be appropriate, the condition is only a local condition, so the global form of the function F cannot be predicted, and there is a possibility that a defect will occur in a part of the converted image. It cannot be wiped out completely. Therefore, it is possible to determine the function F by setting only a small number N (<L) of colors to be changed and imposing global conditions to be satisfied by the function F. However, it is practically preferable. Alternatively, the color change processing means may be configured to obtain a function F in which the order is set sufficiently large so that the number L of terms of the nth order polynomial exceeds the specified number N.

  The global condition is that the sum of squares of differences between the elements of the identity matrix, that is, the coefficient matrix of the transformation I that does not change all colors in the color space and the coefficient matrix of the function F, is the minimum, ie, F−I. The condition that minimizes the norm of is adopted. Under this condition, the conversion source color for which the color change instruction has been issued moves to the conversion target color, and the conversion F by a continuous polynomial is uniquely determined so that all colors that do not correspond to the conversion source color remain at the original color as much as possible.

In the image processing apparatus according to the first invention, the conversion source color and the conversion target color are displayed as a set of three color variables (x1, x2, x3) in a three-dimensional color space. In addition, when the color change correspondence is specified, the color space of the conversion source color and the conversion target color are the same color space. Typically, this is possible by outputting the processing target image and the reference image on the same output device such as the same monitor display or color printer and then associating the corresponding colors between the two images. However, if it is considered that the correspondence in the same color space should be taken, a reference image (output by the same output device) is not necessarily required to designate the color change correspondence. For example, for color change correspondence, specifying the conversion target color corresponding to the conversion source color as its own color is effective because the correspondence is specified in the same color space. Further, in practice, there is a case where it is desired to keep this color as it is in the color change process, so it is desirable that the color change instruction input means accepts such a designation.

The second invention for solving the above-mentioned problem is
After performing a color change instruction by designating one or more color change correspondences that are combinations of a conversion source color that is a specific color on the processing target image and a conversion target color to be obtained after conversion,
Based on this color change instruction, a function F for converting the color (x1, x2, x3) appearing on the processing target image displayed in the three-dimensional color space into the color (y1, y2, y3) after the color change process is obtained. A color change processing method for obtaining a color change processing result image by obtaining an n-th order polynomial (n is a natural number of 2 or more) of variables x1, x2, and x3 and converting the processing target image by this function F,
The specified number N corresponding to the color change when instructing the color change is specified to be less than the number L of terms of the nth order polynomial, or the order n is set so that the number L of terms of the nth order polynomial exceeds the specified number N. It is determined that the function F is set to be sufficiently large and the function F is a solution of the underdetermined simultaneous linear equations determined by the color change instruction, and the norm of the difference between the function F and the identity transformation is minimized. This is a characteristic color change processing method.

  A third invention of the present application for solving the problem is a computer program that causes a computer including a control unit, a display unit, and an input unit to function as the image processing apparatus according to the first invention.

  According to the image processing apparatus, the color change processing method, and the computer program according to the present invention, a single continuous function that satisfies all the color change instructions and is defined in the entire color space only by specifying the necessary color change instructions. 1 can be obtained, and there is a remarkable effect that even a non-skilled worker can easily obtain a desired color-change image.

1 is a schematic configuration diagram of an image processing apparatus 100 according to an embodiment of the first invention. 3 is a processing flowchart of the image processing apparatus 100. FIG. 6 is a display example during processing of the image processing apparatus 100. 6 is an explanatory diagram of a color change instruction process of the image processing apparatus 100. FIG. 4 is a display example of a processing result image of the image processing apparatus 100.

Hereinafter, embodiments of the present invention will be described with reference to the drawings as necessary. First, the procedure for determining the color conversion function F based on the color change instruction will be described.
First, preparation is made for mathematically describing the problem of determining the color conversion function F.
A three-dimensional color space is represented by C (bold character representing space). The color space can also be represented by a dimension larger than 3. For example, the color may be represented by an m-dimensional vector by the spectral intensity at each wavelength obtained by discretizing the wavelength of light in the visible region with m values at equal intervals. However, since three dimensions such as (r, g, b) are usually sufficient, C is a three-dimensional space. The conversion source color on the conversion target image is assumed to be x (∈ C) (x is a three-dimensional vector). The conversion target color on the reference image is y (∈ C) (y is a three-dimensional vector).

The actual color space takes various variables (color space axis) such as (r, g, b), (c, m, y), (L ^* , a ^* , b ^* ), but the following mathematical In the description, these are generalized and displayed as (x1, x2, x3) as the color space. Designating a conversion target color for the conversion source color is the same as designating a pair of coordinates on the same color space.

The color conversion instruction is specified as a set (x, y) of the conversion source color x and the conversion target color y. This set is called color change correspondence.
The color change correspondence (x, y) represents the intention of an instruction “convert color so that the conversion source color x moves to the conversion target color y”. Actually, the conversion target image is composed of a large number of pixels. It may be considered that each pixel constituting the image has a color specified on the color space by a three-dimensional vector as its attribute. According to this view, the color change correspondence (x, y) means “convert the color attribute of the pixel of the conversion target image exhibiting the conversion source color x so as to exhibit the conversion target color y”. In the following description, pixels are not mentioned one by one unless necessary. When referring to the conversion source color x, the meaning of one point x on the color space C and the meaning of several pixels on the conversion target image having the color attribute x are not distinguished unless otherwise specified (unless confused).

The color change instruction is for specifying N sets of the color change correspondences (the conversion source color and the conversion target color). In the present embodiment, the value of N is 1 or more, and a constraint (equation (1)) that is a natural number smaller than L described later is provided.
1 ≦ N <L (1)
In other words, the color change instruction is represented as a set of N sets corresponding to color change.
[(x (1), y (1)), (x (2), y (2)), .., (x (N), y (N))]

Consider a continuous conversion f from the color space C to the same color space C. It is assumed that an arbitrary point x in C 1 moves to a point y 1 in C by f. That is,
y = f (x) (2)
It is.

In the present embodiment, it is assumed that f 1 is expressed by an nth order polynomial from three variables to three variables. n may arbitrarily be a value of 2 or more. The system side may set a fixed value, or the user may be able to specify it every time he / she works. Based on n determined in this way, L is defined by the following equation (3). L is equal to the number of terms of the n-order polynomial of three variables.
L = (3 + n)! / (3! N!) (3)

  The value of n can be determined first, and then a color change instruction consisting of N 1 or more (L−1) or less corresponding groups of color change can be designated accordingly. Alternatively, when the number N of color-change-corresponding sets that must be specified is determined in advance, a sufficiently large value of n 1 may be selected so as to satisfy Equation (1).

Let us take the case of n = 4 as an example. According to equation (3), L is
L = (7!) / (3! 4!) = 35
Is calculated as
An arbitrary point x in the color space C is represented by a color component x = (x1, x2, x3) ^T (4)
Can be written down. T on the right shoulder represents a transposed matrix. Similarly, the point y in C is expressed as y = (y1, y2, y3) ^T (5)
Can be written down.

式（６）の右辺は３行３５列の行列と３５行１列の列ベクトルの積である。前者をＦで表し、後者をｕで表すことにする。すると式（６）は次のようにも表示できる。

ｙ ＝ Ｆｕ （６） The color conversion f can be written down with the following matrix representation.

The right side of Equation (6) is the product of a 3 × 35 matrix and a 35 × 1 column vector. The former is represented by F and the latter is represented by u. Then, equation (6) can be displayed as follows.

y = Fu (6)

Now, an example of m = 3 (the number of dimensions of the color space) and n = 4 has been described. However, expression (6) can also be expressed for general m and n by the same operation. At this time, x and y are column vectors composed of m rows, u is a column vector composed of L rows, and F is a matrix of m rows and L columns. It is easy to obtain u from x = (x1, x2, x3) ^T.

Color change instructions
[(x (1), y (1)), (x (2), y (2)), .., (x (N), y (N))]
X (1), x (2), ..., x (N)
U (1), u (2), ..., u (N) corresponding to each of u
It will be expressed as

Here, we want to obtain the continuous conversion f 1. In the present embodiment, since f is limited to a polynomial expression, obtaining f 1 is nothing other than determining the values of all the components of the matrix F. First, as a first condition to be satisfied by f, it is determined that all the color change correspondences specified by the color change instruction are satisfied. That is, all the expressions (7.1) to (7.N) (collectively, the expression (7)) are satisfied.
y (1) = Fu (1) (7.1)
y (2) = Fu (2) (7.2)
... ... ... ... ... ... ... ...
y (N) = Fu (N) (7.N)

これを、さらに、式（９）のように書き表すことにする。
Ｙ ＝ ＦＵ （９）
と書き表すことにする。このとき、Ｙ は３行Ｎ列の行列であり、Ｕは Ｌ行Ｎ列の行列である。（Ｆは３行Ｌ列の行列） If this is written horizontally, Expression (7) can be re-expressed as Expression (8).

This is further expressed as shown in Equation (9).
Y = FU (9)
Will be written. At this time, Y is a matrix of 3 rows and N columns, and U is a matrix of L rows and N columns. (F is a matrix with 3 rows and L columns)

  In equation (9) for F, the number of unknown variables is 3L and the number of equations is 3N. However, since N is defined as 1 ≦ N <L, it is an underdetermined system in which the number of equations is smaller than the number of unknown variables. At this time, F exists that satisfies the first condition, but it exists innumerably and is not uniquely determined.

  Therefore, the second condition (global condition) can be set so that F is uniquely determined. In general, it is often used as a condition for uniquely determining a solution of a simultaneous linear equation of an underdetermined system to select a solution having a minimum norm. It is known that the solution is uniquely determined by adding this condition. However, this condition is not always preferable for color conversion.

When specifying a color change instruction consisting of N pairs of conversion source color and conversion target color, the origin is as much as possible while maintaining the overall continuity for colors not specified as the conversion source color on the processing target image. It is not usually intended to shift to a color close to, but is usually intended not to change the original color as much as possible. In view of this intention, when the second condition is set as preferable in practice, it is considered that the condition is to minimize the norm of the difference between the transformation f and the identity transformation determined by the matrix F 1. . Here, the norm of transformation determined by a matrix can be defined by the square root of the sum of squares of all components of the matrix.

ｘから生成したｕ は、I により ｘ に移される。つまり、
ｘ ＝ Ｉｕ
である。色替え指示で指定されたすべての組について、これを書き並べると式（１１）となる。

これを、さらに、式（１２）で書き表すことにする。
Ｘ ＝ ＩＵ （１２）
一方、式（９）は、
Ｙ ＝ ＦＵ （９）
であった。式(９)から式（１２）を辺々差し引くことにより、式（１３）を得る。
Ｙ−Ｘ ＝ （Ｆ−Ｉ）Ｕ （１３） Hereinafter, a procedure for obtaining F so as to satisfy the second condition will be described. Let I be the matrix corresponding to the identity transformation. I is a matrix of 3 rows and L columns expressed by Expression (10). However, all the components in the area indicated by ... are 0.

u generated from x is moved to x by I. That means
x = Iu
It is. When all the groups designated by the color change instruction are written and arranged, Expression (11) is obtained.

This is further expressed by equation (12).
X = IU (12)
On the other hand, Equation (9) is
Y = FU (9)
Met. By subtracting equation (12) from equation (9) side by side, equation (13) is obtained.
Y−X = (FI) U (13)

Considering this as a simultaneous linear equation with all the components of (FI) as unknowns, it is a simultaneous linear equation of an underdetermined system. The second condition described above is to minimize the norm of (FI).
|| FI || min (14)
In other words. This is in the form of a commonly known norm minimization problem of simultaneous linear equations of underdetermined systems. It is known that the solution of this is uniquely determined (for example, Tetsuro Yamamoto “Matrix Analysis Note Pearl Theorem and Selection Problem”, Science (2013/1/21), or Yoshiomi Nakagami, Haruo Yanai “ Rectangle determinant ”, Maruzen Publishing (2012/12/15)).
That is,
F−I = (Y−X) U ⁺ (15)
It is.
Where U ⁺ is a Moore-Penrose generalized inverse for U. From Expression (15), F is obtained as Expression (16).

F = I + (Y-X) U ⁺ (16)

(Experimental example)
The colors of the points 1 to 3 in the reference image B (FIG. 4B) are the numbers 1 to 3 in the reference image B (FIG. 4B). In addition, the color designation (color changing instruction not to be changed) corresponding to its own color was performed for the color of the location of the circle and the color of the circled number 4 to the circled number 8 of the image A. N = 8.
As the color of the dotted circle numbers 1 to 8 in the image A, the average value of each (r, g, b) value in the vicinity of 5 × 5 pixels of the designated coordinate pixel on the image by the instructor is used. It was determined as the color of the part. The colors specified by this method of determining each of the circled numbers 1 to 8 on the image A are x (1),..., X (8), and the color of the corresponding portion of the corresponding image B, respectively. Each is y (1), ..., y (8),
x (1) = (185.960, 80.000, 76.320) ^T
x (2) = (198.240, 93.000, 90.240) ^T
x (3) = (150.240, 63.200, 64.120) ^T
y (1) = (146.360, 45.880, 46.480) ^T
y (2) = (183.840, 62.480, 63.280) ^T
y (3) = (107.800, 28.280, 28.480) ^T
x (4) = y (4) = (238.720, 206.760, 186.800) ^T
x (5) = y (5) = (198.600, 156.040, 137.960) ^T
x (6) = y (6) = (218.880, 181.120, 162.440) ^T
x (7) = y (7) = (253.080, 253.080, 253.080) ^T
x (8) = y (8) = (233.400, 232.400, 228.400) ^T
As a result of getting the color change instruction,
Y-X = (x (1) -y (1), x (2) -y (2), x (3) -y (3),
(0,0,0,0,0)
Get.
In this experiment, a sixth order polynomial was used. Therefore, L = 84. If the degree n of the polynomial is determined, u (1),..., U (8) are specifically determined from the above x (1),. Matrix U is defined. The Moore-Penrose generalized inverse matrix U ⁺ from U to U must be determined, which is determined by known computational procedures.
For example, U is not degenerate, ie
rank U = 8
After confirming that
U ⁺ = (U ^T U) ^-1 U ^T
U ⁺ can be obtained as However, (-1) on the right shoulder represents a normal inverse matrix.
Substituting YX, I, U ⁺ into equation (16), a coefficient matrix F of sixth-order polynomial transformation f is obtained. (Showing actual numerical values of the matrix U of 84 rows and 8 columns and U ⁺ of 8 rows and 84 columns is complicated and will be omitted.)
Further, as another method for solving a simultaneous linear equation of an underdetermined system under a constraint condition of norm minimization, an LQ decomposition method is also known and may be used.
A result image obtained by converting the image to be processed by the color change function f determined by F is obtained as shown in FIG. In FIG. 5C, the slightly dull red two-piece is changed in color in a direction in which the vividness increases as a result of matching with the color of the reference image, but the neck, arms, legs, etc. are not changed.

Next, an image processing apparatus equipped with a color change processing function based on the color change processing method described so far will be described. FIG. 1 is a schematic configuration diagram of an image processing apparatus 100 according to an embodiment of the first invention of the present application. The image processing apparatus 100 can be realized by connecting a display unit 120 and an input unit 130 to a general-purpose computer main body and mounting a program that functions as a color change instruction input unit and a program that functions as a color change unit. The image processing apparatus 100 includes a central processing unit (CPU) 101, a display interface 102, an input interface 103, and a storage device 106. These are connected by a bus 109. In addition to the image data of the processing target image and the reference image, the storage device 106 accepts a color change instruction and executes a program code 201 for executing an instruction for storing the color change instruction data in a predetermined area of the storage device 106. Program code 202 for executing the replacement process is stored. The program codes 201 and 202 are read from the storage unit 106 by the CPU 101 and are interpreted and executed by the CPU to drive and control the hardware resources of the image processing apparatus 100, thereby respectively changing the color change instruction input unit and the color change processing unit. Will function as. In addition, although not shown, the storage unit 106 also stores and holds an OS (operating system) code that provides a user with basic functions of a display unit and an input unit connected to the image processing apparatus 100. The display unit 120 is a monitor display that displays display data in the display memory of the display interface 102 in color. The input unit 130 is a keyboard and / or a mouse.

  FIG. 2 is a flowchart for explaining the processing flow of color change processing by the image processing apparatus 100.

First, the operator selects an image file of a processing target image on the image processing apparatus 100 and displays it on the display unit 120 (S1). Next, the operator selects an image file of a reference image on the image processing apparatus 100 and displays it on the display unit 120 (S2). At this stage, the processing target image (A) and the reference image (B) are displayed on the display unit 120 as shown in FIG.

Next, the color change instruction input unit 201 is activated, and the operator designates one point of the reference image corresponding to one point of the processing target image, thereby associating the coordinates on the color space represented by the point of the image. And stored in the storage unit 106 as one set of color change correspondences (S3). (Depending on the implementation mode, assuming that a predetermined number of pixels in the vicinity of the relevant location based on the display magnification is designated as one designated point on the screen, the color of the designated point is set on the color space of several relevant pixels. (In the following description, it will be expressed as “color represented by a specified point”).

Color change correspondence is basically to associate a certain point on the reference image with a certain point on the image to be processed and associate the colors displayed by each point on the color space, but as a variation, It may include a designation for preventing a color of a certain point on the processing target image from being changed by the color changing process. In this case as well, there is no change in specifying one set of correspondences in the color space. The necessary number of color correspondence instructions are performed to obtain N sets of color instruction correspondences, and the color change instruction input (S3) is completed. The inputted N sets of color instruction correspondences are temporarily recorded in a predetermined area of the storage unit 106 in a predetermined format as color change instruction data.
As a further variation, a color patch may be displayed on the display unit, and the user may select a color from the color patch as a conversion source color or conversion target color designation means in the color change instruction.
Furthermore, the user can specify R, G, B or hue, brightness, and saturation to generate an arbitrary color, and the generated color can be specified as the conversion source color or conversion target color. Good.
For example, a lineup of products with the same shape and different colors, such as handbags, may be provided, but the image obtained by shooting one color product can be replaced with another color product. Sometimes you want to use it. In such a case, the color patch or the arbitrary color generation means is useful.

In FIG. 4, the circled number 1 on the processing target image is set to the collar color at the position of the circled number 1 in the reference image, and the rounded number 2 (cloth background color) on the processing target image is set to the circled number 2 on the reference image. Change the color of the position (slightly cuffs) to the color of the circled number 3 on the image to be processed (the part in the shadow of the cuffs) to the color of the circled number 3 position (cuffs darkened) on the reference image, Circle number 4 (neck), circle number 5 (right arm), circle number 6 (right foot highlight), circle number 7 (background light portion), circle number 8 (background dark portion) of the image to be processed A dot indicates a case where a color change instruction is given so as not to change the color change process explicitly.

When the input of the color change instruction is completed, the color change processing means 202 is activated, the color change instruction data is read from the storage unit 106, and the color change process is executed (S4). At this time, as the order n of the function F to be obtained, a default value, for example, n = 4 is set in advance. As a result, a default value of L (L = 35 by equation (3) when n = 4) is also set. Therefore, before entering the color changing process, it is checked whether N satisfies the expression (1). If not, n is incremented as necessary to satisfy equation (1). Thereafter, color change processing is performed. As described above, the content of the color changing process is as follows. The color-reversed matrix F is obtained by calculating the Moore-Penrose generalized inverse matrix U ⁺ for given X and U to U and substituting it into equation (16). And the color coordinate of each pixel of the processing target image is converted by this F.

The color change processing result image is displayed on the display unit (S5). If the operator is satisfied with the result image (down arrow in S6), the result image is saved and the color change process is terminated (S7). If there is dissatisfaction, the process returns to step S3 to redo the color change instruction. The order of the function n may be incremented by 1 and the step S4 may be performed again with the original color change instruction.

The image processing apparatus 100 as one embodiment of the image processing apparatus according to the first invention of the present application has been described above. However, the image processing apparatus according to the first invention of the present application is not limited to the above description. An apparatus variously modified without departing from the spirit of the present invention can also be an apparatus according to the technical scope of the image processing apparatus according to the first invention of the present application. For example, in FIG. 3-5, there is one reference image, but a plurality of reference images may be used. When the color patch is displayed on the display unit and the user selects a color from the color patch as the conversion source color or conversion target color designation means in the color change instruction, a plurality of reference images are used. It corresponds to. The user interface for instructing the color change correspondence may be anything as long as it can specify a set of two points on the same color space, and does not have to be a screen as shown in FIG.

As described above, the image processing apparatus and the color change processing method of the present invention may be useful for improving productivity in the industrial field where color image processing is performed, particularly in the field of production of commercial printed matter.

100 Image Processing Device 101 Central Processing Unit (CPU)
DESCRIPTION OF SYMBOLS 102 Display interface 103 Input interface 106 Memory | storage device 109 Bus | bath 120 Display part 130 Input part 201 Program code for functioning as a color change instruction input means 202 Program code for functioning as a color change processing means

これを、さらに、式（９）のように書き表すことにする。
Ｙ ＝ ＦＵ （９）
このとき、Ｙ は３行Ｎ列の行列であり、Ｕは Ｌ行Ｎ列の行列である。（Ｆは３行Ｌ列の行列） If this is written horizontally, Expression (7) can be re-expressed as Expression (8).

This is further expressed as shown in Equation (9).
Y = FU (9)
At this time, Y is a matrix of 3 rows and N columns, and U is a matrix of L rows and N columns. (F is a matrix with 3 rows and L columns)

Claims (4)

A storage unit that records and holds each image data and color change instruction data of the reference image, the processing target image, and the processing result image; and a display unit that displays the reference image, the processing target image, and the processing result image;
Accepted and accepted a color change instruction by specifying one or more color change correspondences that are combinations of a conversion source color that is a specific color on the processing target image and a conversion target color that is a specific color on the reference image Color change instruction input means for recording color change instruction data in the storage unit;
Based on the color change instruction so that the conversion source color on the processing target image is converted into a conversion target color specified in correspondence with the color change, and the color gamut not specified as the conversion source color satisfies a global condition separately defined. A color change processing means for performing continuous color change processing of the processing target image to obtain a processing result image;
An image processing apparatus that performs a color change process of a processing target image based on a reference image including a display unit, an input unit, and a control unit that controls a color change processing unit,
The color change processing means reads the color change instruction data from the storage unit, and changes the color (x1, x2, x3) appearing on the processing target image displayed in the three-dimensional color space to the color (y1, a function F for converting to y2, y3) as an n-order polynomial (n is a natural number of 2 or more) of variables x1, x2, and x3,
The function F as a global condition determined separately is determined based on a condition that the norm of the difference from the identity transformation is minimized,
The color change instruction input means is configured to accept the specified number N corresponding to the color change by a number smaller than the number L of terms of the nth order polynomial, or the color change processing means has the number of terms of the nth order polynomial. An image processing apparatus configured to obtain a function F having a sufficiently large order so that L exceeds a specified number N.   The image processing apparatus according to claim 1, wherein the color change instruction input unit is capable of specifying color change correspondence with the conversion target color corresponding to the conversion source color as its own color. A featured image processing apparatus. After performing a color change instruction by designating one or more color change correspondences that are combinations of a conversion source color that is a specific color on the processing target image and a conversion target color to be obtained after conversion,
Based on this color change instruction, a function F for converting the color (x1, x2, x3) appearing on the processing target image displayed in the three-dimensional color space into the color (y1, y2, y3) after the color change process is obtained. A color change processing method for obtaining a color change processing result image by obtaining an n-th order polynomial (n is a natural number of 2 or more) of variables x1, x2, and x3 and converting the processing target image by this function F,
The specified number N corresponding to the color change when instructing the color change is specified to be less than the number L of terms of the nth order polynomial, or the order n is set so that the number L of terms of the nth order polynomial exceeds the specified number N. It is determined that the function F is set to be sufficiently large and the function F is a solution of the underdetermined simultaneous linear equations determined by the color change instruction, and the norm of the difference between the function F and the identity transformation is minimized. Characteristic color change processing method. A computer program that causes a computer including a control unit, a display unit, and an input unit to function as the image processing apparatus according to claim 1.

Images