JP2010124168A - Method of calculating color conversion matrix - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a method of calculating color conversion matrix, which reduces time until derivation (optimization) of color conversion matrix. <P>SOLUTION: The method of calculating color conversion matrix has a step for performing colorimetry on color patches to acquire a first color value, and a step for converting input color signals acquired by photographing color patches using a color conversion matrix, the matrix elements of which are unknown, to acquire a second color value. In addition, the method also has a step for calculating a color conversion matrix through a least squares method using the first and second color values. In this step, a first color conversion matrix is calculated through a least squares method not using lightness components but using only color difference components. <P>COPYRIGHT: (C)2010,JPO&INPIT

Description

  The present invention relates to a method for calculating a color conversion matrix used, for example, when performing color correction on a digital signal output from a color imaging device.

  In general, there is a gap between the color space of an image input system such as an imaging device and the color space of an image output system such as an image display device or a printer. In addition, the RGB signal obtained by the image input device depends on the imaging characteristics of an optical system such as a photographic lens, and a light receiving sensor such as a color filter and an imaging element, and thus there is a difference between the actual color and the color of the input signal. Therefore, when an image captured by the imaging apparatus is displayed on a monitor or the like, the color of the reproduced image is different from the color of the original subject. Further, such a difference in hue varies depending on the type of output device.

  Therefore, for example, there is a technique for approximating a reproduced color to an original color by color correction processing using a color conversion matrix (Patent Document 1).

In the technology according to Patent Document 1, in the L ^* a ^* b ^* space, L ^*, the square sum of the difference between a ^* and b ^* each actual color signal and a color signal of a captured image of the components of the smaller As described above, the color conversion matrix (linear matrix) is optimized using the iterative solution method of the least square method. In addition, a unit matrix is used as an initial value when the color conversion matrix is derived.

JP-A-2005-45446

  However, in the case of the derivation (optimization) method of the color conversion matrix according to Patent Document 1, there is a problem that a large amount of calculation is required for derivation of the color conversion matrix, and it takes a long time to derive the color conversion matrix. there were.

  Further, the technique according to Patent Document 1 has a problem that the reproducibility accuracy of the hue and saturation is lowered.

  SUMMARY OF THE INVENTION An object of the present invention is to provide a color conversion matrix calculation method that can shorten the time required until the color conversion matrix is derived (optimized). It is another object of the present invention to provide a color conversion matrix calculation method capable of improving the reproducibility of hue and saturation.

  In order to achieve the above object, a color conversion matrix calculation method according to claim 1 of the present invention is a color conversion matrix calculation method for calculating a color conversion matrix used for color correction for a color signal. (A) preparing a plurality of color patches whose chroma and hue change stepwise; (B) obtaining a first color value by measuring the color patch; C) obtaining an input color signal obtained by imaging the color patch by using the color conversion matrix having an unknown matrix element to obtain a second color value; and (D) the first Calculating the color conversion matrix by obtaining the unknown by a least square method using one color value and the second color value, and the step (D) includes: D-1) without using a brightness component, by the using only the chrominance components of least squares method, calculating a first of the color conversion matrix, a.

A color conversion matrix calculation method according to a second aspect of the present invention is the color conversion matrix calculation method according to the first aspect, wherein the step (B) is performed for a plurality of the color patches. Each is a step of obtaining the first color value, and the step (C) is a step of obtaining the second color value for each of the plurality of color patches, and the step (D-1). Is a step of calculating the first color conversion matrix by the least square method so that ΣωiΔEi is minimized. Here, Σ is a sum of the number of the plurality of color patches, i is an identifier of the color patch, ωi is a weighting constant, and the weighting constant ωi is equal to the number of the color patches. The sum is 1, and ΔEi is {(the first color difference component of the first color value) − (the first color difference component of the second color value)} ² + {(the first color difference component) Second color difference component of color value) − (second color difference component of the second color value)} ² .

  The color conversion matrix calculation method according to claim 3 according to the present invention is the color conversion matrix calculation method according to claim 1, wherein the step (D) includes (D-2) a plurality of the above-described steps. In the color patch, obtaining a first lightness average value that is an average value of lightness components of the first color value; and (D-3) an input color signal obtained by imaging the color patch, A step of obtaining a third color value by performing conversion using the first color conversion matrix; and (D-4) an average value of brightness components of the third color value in the plurality of color patches. A step of obtaining a second lightness average value; and (D-5) dividing the first lightness average value by the second lightness average value to obtain a value obtained by dividing the first color conversion matrix. By multiplying each constituent matrix element Calculating a second of the color conversion matrix, a further.

  A color conversion matrix calculation method according to a fourth aspect of the present invention is the color conversion matrix calculation method according to the first aspect, wherein the step (D-1) includes (D-1-1). ) Selecting any three color patches; (D-1-2) obtaining the first color values in the three color patches; and (D-1-3) the three colors. A step of obtaining the second color value from an input color signal obtained by imaging the color patch, and (D-1-4) the first obtained in the step (D-1-2). Obtaining the unknown so that the color value is equal to the second color value obtained in step (D-1-3); and (D-1-5) step (D-1). -4) is used as the first initial value by the least square method, And calculating a first color conversion matrix, a.

  According to a fifth aspect of the present invention, there is provided a color conversion matrix calculation method according to the fourth aspect, wherein the three color patches are red, blue and green. .

  A color conversion matrix calculation method according to a sixth aspect of the present invention is the color conversion matrix calculation method according to the fourth aspect, wherein the three color patches are cyan, magenta, and yellow. .

  In the color conversion matrix calculation method according to the first aspect of the present invention, the first color conversion matrix is calculated by the least square method using only the color difference component without using the brightness component.

  Therefore, since the lightness component is not added in the process using the least square method, the time until the solution of each matrix element constituting the first color conversion matrix is converged is shortened. In addition, since the lightness component is not taken into account, the weighting of the color difference component can be increased during the first matrix calculation. Therefore, it is possible to obtain a color conversion matrix with high accuracy of optimization of hue / saturation components.

  In the color conversion matrix calculation method according to the second aspect of the present invention, a solution of each matrix element constituting the first color conversion matrix is obtained from a function including the weighting constant ωi.

  Therefore, it is possible to make a difference in correction accuracy between a certain color and another color. Therefore, it is possible to calculate a color conversion matrix that improves the correction accuracy for a specific color compared to other colors.

  In the color conversion matrix calculation method according to the third aspect of the present invention, the second color conversion matrix is obtained only by simple four basic calculations.

  Therefore, the second color conversion matrix considering the color difference component and the brightness component can be obtained in a short time using the first color conversion matrix.

  According to a fourth aspect of the present invention, when the first color conversion matrix is derived, the initial initial value is determined based on the color values of any three color patches. is doing.

  In this way, the initial initial value is determined based on the input color signal that varies depending on the imaging device, not the fixed value. Therefore, it is possible to determine the initial initial value in consideration of the sensitivity characteristics for each imaging device.

  In the color conversion matrix calculation method according to claims 5 and 6 of the present invention, red, blue and green, or cyan, magenta and yellow are selected as the three color patches. .

  In this way, by selecting a three-color patch having a substantially different hue, the time until the optimization of the first color conversion matrix can be reduced compared to selecting a three-color patch having a similar hue. It can be shortened further. In addition, in the solution for optimizing the first color conversion matrix, the time required to reach a solution that can be determined to be sufficiently converged is shorter than when a unit matrix is selected as the initial value. be able to.

  Hereinafter, the present invention will be specifically described with reference to the drawings showing embodiments thereof.

<Embodiment 1>
FIG. 1 is a diagram showing a system overview of a color conversion matrix calculation method and a color conversion method according to the present invention.

  An imaging and color conversion matrix calculation system 100 illustrated in FIG. 1 includes an imaging device 10, a color conversion matrix calculation device 20, and an output device 30.

  The imaging device 10 is a device that can image a subject. In the imaging apparatus 10, an analog signal that is a captured image signal is output from the imaging element. The analog signal is converted into a digital signal by an A / D converter. The digital signal is subjected to digital processing and correction processing in the imaging device 10. For example, color separation processing for generating RGB signals for digital signals, white balance correction processing for adjusting gains of R (red), G (green), and B (blue) according to white reference values, color correction processing , And a gradation correction process for adjusting the gradation so as to cancel out the γ characteristic of the output device 30.

  The reproduced color output from the output device 30 and the original color of the subject imaged by the imaging device 10 generally do not match. Therefore, the color correction process is performed in order to bring the reproduced color close to the original color of the subject. The color correction process is a process of converting the color gamut of the primary color RGB signal that is the color signal of the first color space to be input into the color gamut of the sRGB signal that is the color signal of the second color space defined by the sRGB standard. It is. The color correction process is performed using a color conversion matrix.

  For example, the color conversion matrix is obtained (optimized) in cooperation with the color conversion matrix calculation device 20 at the final stage of the manufacturing process of the imaging device 10. The color conversion matrix is a 3 × 3 matrix and includes nine matrix elements. Here, the values of nine matrix elements constituting the color conversion matrix are determined according to the color sensitivity characteristics of the imaging system of each imaging device 10.

  Next, a color conversion matrix calculation method in the color conversion matrix calculation apparatus 20 will be described.

  First, a color checker including a plurality of color samples (color patches) whose saturation and hue change in stages is prepared. As the color checker, for example, a Macbeth color chart (registered trademark) as shown in FIG. 2 can be adopted.

  The Macbeth color chart shown in FIG. 2 includes 24 color patches P1 to P24. Here, the color patches P1 to P18 are chromatic colors, and the color patches P19 to P24 are achromatic colors. In the following description, a Macbeth color chart is used as the color checker.

  Next, the color patches P1 to P24 constituting the Macbeth color chart are accurately measured with a colorimeter or the like. The colorimetric values of the color patches P1 to P24 of the Macbeth color chart may be known. If the colorimetric values of the color patches P1 to P24 of the Macbeth color chart are not known, the color patches P1 to P24 are measured under the same illumination environment as that for imaging the Macbeth color chart by the imaging device 10 described later. By the color measurement, Cme1 to Cme24 are obtained as the first color values for the color patches P1 to P24.

  Here, as will be described later, when calculating the color conversion matrix, only the color patches P1 to P18 that are chromatic colors are basically used. The achromatic color patches P19 to P24 are used for gray gradation correction, white balance correction, gain adjustment, and the like.

Colorimetric signals (RGB signals) that are the colorimetric results of the color patches P <b> 1 to P <b> 24 are input to the color conversion matrix calculator 20. Here, if the input data is not an RGB signal, for example, an XYZ signal, an L ^* a ^* b ^* signal, a spectral reflectance, or the like, it is converted into an RGB signal by a known conversion formula ( For example, see formulas (3) and (4) described later).

  Next, the Macbeth color chart is imaged as a subject using the imaging device 10. The imaging by the imaging device 10 is performed under the same illumination environment as that at the time of color measurement. As described above, the imaging apparatus 10 can obtain the input color signal (RAW data) by imaging the color patches P1 to P24.

  After the imaging, the imaging apparatus 10 transmits the input color signal (RAW data) to the color conversion matrix calculation unit 20. Here, the input color obtained from the input color signal (RAW data) is defined as Cin (R (red) in, G (green) in, B (blue) in).

  Further, the color conversion matrix M before the optimization (derivation) process is defined as the following expression (1) using nine unknown matrix components (matrix elements) m1 to m9.

  In the color conversion matrix calculation device 20, the input color Cin is converted using the color conversion matrix M whose matrix elements are unknown, thereby correcting the colors Ces (R (red) es, G (green) es. , B (blue) es). It can be understood that the color value obtained using the color conversion matrix M whose matrix element is an unknown number is the second color value. The relationship of Ces = M · Cin is established between the input color Cin and the correction color Ces. The following equation (2) is a matrix representation of the relational expression.

Here, in the matrix calculation, colors are handled by RGB signals. However, in the evaluation of the color coincidence, it is handled by the L ^* a ^* b ^* signal. This is because the Lab color space is a uniform color space with good correlation between human color perception and coordinate distance. Note that L ^* is a lightness component, and a ^* and b ^* are color difference components indicating hue and saturation. Further, a ^* is a first color difference component, and b ^* is a second color difference component.

The conversion from the RGB signal to the L ^* a ^* b ^* signal is realized by known formulas shown in the following formulas (3) and (4).

Here, Expression (3) is a conversion expression for converting the RGB signal into the XYZ signal. Expression (4) is a conversion expression for converting an XYZ signal into an L ^* a ^* b ^* signal. When evaluating the degree of color matching, the RGB signal is converted into an L ^* a ^* b ^* signal by RGB → XYZ → L ^* a ^* b ^* conversion.

In the formula (3), the viewing angle with respect to the eyes of the standard observer is 2 °, and the standard light CIE-D ₆₅ is used as illumination light.

In the formula (4), if X / 95.047 ≦ 0.008856, f (X) = X ^1/3 is set. If 0.008856 <X / 95.047, then f (X) = 7.787X + 16/116. If Y / 100.000 ≦ 0.008856, then f (Y) = Y ^1/3 . If 0.008856 <Y / 100.000, f (Y) = 7.787Y + 16/116. If Z / 108.883 ≦ 0.008856, f (X) = Z ^1/3 . If 0.008856 <Z / 108.883, then f (Z) = 7.787Z + 16/116.

  Although not specifically mentioned below, when color signal conversion processing is necessary, the above-described equations (3) and (4) are used.

After obtaining the first and second color values and converting the first and second color values from an RGB signal to an L ^* a ^* b ^* signal, the first color value and the second color value are obtained. The color conversion matrix M is calculated (optimized) by obtaining an unknown number of matrix components m1 to m9 by the least square method used.

  In the present embodiment, for the 18 colors of the color patches P1 to P18, all of the correction colors Ces (second color values) predicted from the input color Cin are the corresponding target colors Cme (and the first color). The color conversion matrix M is optimized so as to substantially match. Specifically, matrix elements m1 to m9 are obtained by the iterative solution of the least square method such that the sum of squares of 18 colors of the color difference ΔE between the correction color Ces and the target color Cme is equal to or less than an allowable error.

  For a predetermined color of the color patch, the color difference ΔE between the correction color Ces and the target color Cme is defined by the following equation (5) and evaluated in the Lab color space.

Here, in Expression (5), the L ^* a ^* b ^* signal of the correction color Ces is (L ^* es, a ^* es, b ^* es), and the L ^* a ^* b ^* signal of the target color Cme is (L ^* me, a ^* me, b ^* me).

Here, it should be noted that in the color conversion matrix calculation method according to the present invention, the color conversion matrix is calculated by the least square method using only the color difference components a ^* and b ^* without using the lightness component L ^*. is doing. Here, the L ^* a ^* b ^* signal (space) is mentioned here, but the same argument holds for any space composed of one lightness signal and two color difference signals.

  The color conversion matrix obtained by the least square method that does not use the brightness component is referred to as a first color conversion matrix. It can be seen from the above that the first color conversion matrix is derived by a least square method on a two-dimensional plane stretched with only color difference components.

  Here, in order to distinguish the color differences ΔE corresponding to the 18 color patches P1 to P18 of the chromatic color, the number of the color patch is attached as a subscript and expressed as ΔE1, ΔE2,. If a parameter indicating a color patch number (color patch identifier) is i (i = 1, 2,..., 18), each color difference is represented by ΔEi. If the parameter indicating the order of the matrix elements m1, m2,..., M9 is j (j = 1, 2,..., 9), each matrix element is represented by mj. The correction color Ces is obtained by applying the color conversion matrix M to the input color Cin. Therefore, the color difference ΔEi can be regarded as a function ΔEi (mj) having the matrix element mj as a parameter.

  In the least square method, the first and second color values are obtained for 18 chromatic colors, and the color difference ΔEi (i = 1, 2,..., 18 obtained from the first and second color values for 18 chromatic colors. ) Is defined as a function φ as shown in the following equation (6).

Here, Σ in equation (6) is the sum of the number of color patches P1 to P18 for 18 chromatic colors. ΔEi for each color in Equation (6) is defined by Equation (5). Therefore, ΔEi shown in Expression (6) is {(first color difference component of the first color value: a ^* ) − (first color difference component of the second color value) for each of the 18 chromatic colors. : A ^* )} ² + {(second color difference component of the first color value: b ^* ) − (second color difference component of the second color value: b ^* )} ² .

  Then, the matrix element mj (j = 1, 2,..., 9) is optimized so that the function φ is equal to or less than a predetermined threshold. The conditions for minimizing the function φ are given by nine formulas shown in the following formula (7).

  However, equation (6) is a nonlinear equation, and it is impossible to directly find the solution of the matrix element mj.

  Therefore, a matrix element serving as an initial initial value (first starting point) is given, and a matrix element serving as an approximate solution is obtained using equations (6) and (7) under the initial initial value. Then, using the obtained approximate solution as a new starting point, a new approximate solution of the matrix element is obtained by Expressions (6) and (7). In this manner, the iterative solution method of the least square method is repeatedly performed until the function φ reaches the threshold value or less. When the function φ reaches the threshold value or less, the approximate solution at that time is assumed to have converged to the solution, and the converged solution is determined as the values of the matrix elements m1 to m9 of the first color conversion matrix.

  In the optimization of the color conversion matrix using the function φ, the color difference between the correction color Ces and the target color Cme varies substantially equally for each color. However, when it is desired to improve the color matching accuracy of a specific color by limiting to a specific color such as a skin color, weighted constants ωi (i = 1, 1) respectively set for 18 colors of chromatic color patches P1 to P18. 2,..., 18) is multiplied by the color difference ΔEi and weighted. The following equation (8) shows a function φ that takes into account the weighting constant ωi.

  The first color conversion matrix is calculated by the method of least squares so that ΣωiΔEi shown in Expression (8) is minimized. In the case of such a calculation method, it is possible to perform more faithful color reproduction for a specific color having a larger weight by setting the weighting constant ωi. That is, when there is a specific color for which the reproducibility is particularly desired, the ratio of the weighting constant ωi corresponding to that color is increased.

  Here, the sum of the weighting constants ω1 to ω18 is 1 as shown in the following equation (9).

  As can be seen from the above description, Σ in equations (8) and (9) is the sum of the number of color patches P1 to P18 for 18 chromatic colors.

  In addition, the solution processing by the least square method for each function φ described above is known as disclosed in Patent Document 1. Therefore, the specific description of the solution processing by the least square method is omitted here.

The color conversion matrix calculation device 20 obtains each value of the matrix elements m1 to m9, which are optimal solutions, by the iterative solution of the least square method (that is, optimizes the first color conversion matrix). The first color conversion matrix is optimized from the viewpoint of only the color difference components a ^* and b ^* . In addition, since the first color conversion matrix that is the optimum color is determined for the color difference components a ^* and b ^* , the color conversion matrix calculation device 20 uses the lightness component L ^* to subsequently determine the first color conversion matrix. To obtain a second color conversion matrix which is a final color conversion matrix.

The color difference components a ^* and b ^* correspond to the ratio of the R, G, and B values in the R, G, and B spaces, and the lightness component L ^* corresponds to the magnitudes of the R, G, and B values. Therefore, by using the lightness component L ^* , the first color conversion matrix is adjusted as described later to obtain the second color conversion matrix.

First, the color conversion matrix calculation device 20 extracts lightness components L ^* mei for 18 colors from the color measurement results (which can be grasped as the first color values) for the chromatic color patches P1 to P18 in the Macbeth color chart. To do. Here, “me” indicates a colorimetric value. Further, i is an identifier of each chromatic color, i = 1, 2,. As described above, the color conversion matrix calculation apparatus 20 has already been input with the color measurement result of the Macbeth color chart.

Next, as shown in the following equation (10), a first lightness average value Lav1, which is an average value of the lightness components L ^* mei for the 18 colors, is obtained.

  On the other hand, the color conversion matrix calculation device 20 converts the input color signal (RAW data) using the first color conversion matrix. The conversion process is performed for 18 chromatic colors. Here, the above-described input color signal (RAW data) is obtained when a Macbeth color chart is imaged using the imaging device 10 under the same illumination environment as during colorimetry. A value obtained by converting the input color signal (RAW data) using the first color conversion matrix is referred to as a third color value.

Next, the color conversion matrix calculation device 20 extracts lightness components L ^* esi from the third color values for 18 colors of the chromatic color patch. Here, “es” indicates a color value after the conversion process using the first color conversion matrix. Further, i is an identifier of each chromatic color, i = 1, 2,.

Next, as shown in the following equation (11), a second lightness average value Lav2 that is an average value of the lightness components L ^* esi for the 18 colors is obtained.

  The color conversion matrix calculation device 20 obtains the second color conversion matrix Maf using the results of the above equations (10) and (11) and the first color conversion matrix. Here, it is assumed that the second color conversion matrix Maf, which is a 3 × 3 matrix, is expressed by the following equation (12).

  Specifically, the color conversion matrix calculation device 20 divides the first lightness average value Lav1 by the second lightness average value Lav2. Then, the color conversion matrix calculation device 20 multiplies each matrix element constituting the first color conversion matrix by the value obtained from the division result. The values after the multiplication are the matrix elements maf1 to maf9 constituting the second color conversion matrix Maf. In other words, the color conversion matrix calculation device 20 determines each matrix element mafk constituting the second color conversion matrix Maf using the following equation (13). Here, k is an identifier of a matrix element, and k = 1, 2,.

  Here, in Expression (13), mk (k = 1, 2,..., 9) is each matrix element of the first color conversion matrix. The second color conversion matrix composed of the matrix elements mafk obtained by Expression (13) is optimized in consideration of not only the color difference component but also the brightness component. The second color conversion matrix can be grasped as a color conversion matrix that matches the RGB signals obtained by photographing with the RGB signals obtained by colorimetry.

  The color conversion matrix calculation device 20 stores and sets the second color conversion matrix in a memory (not shown) provided in the imaging device 10. Then, after the second color conversion matrix is set, when the imaging device 10 captures an object, a digital signal processing circuit (not shown) disposed in the imaging device 10 uses the second color conversion matrix. The digital color correction process is performed on the captured image signal. By executing the color correction process, the output device 30 realizes color reproduction that is extremely equal to the colorimetric value.

  As described above, in the color conversion matrix calculation method according to the present embodiment, the first color conversion matrix is calculated by the least square method using only the color difference component without using the brightness component. That is, in the derivation of the first color conversion matrix, unnecessary brightness components are not taken into account when evaluating hue and saturation.

  Therefore, since the lightness component is not added in the process using the least square method, the time until the solution of each matrix element constituting the first color conversion matrix is converged is shortened. Note that the second color conversion matrix stored in the imaging device 10 is obtained by simple calculation using only the brightness component as described above. Therefore, the time until the derivation of the color conversion matrix finally stored in the imaging device 10 can be considerably shortened.

  In the technique according to Patent Document 1, the color conversion matrix is optimized by the least square method from a function that includes not only the color difference component but also the brightness component. Therefore, the color conversion matrix calculation method according to the present embodiment can reduce the time until the color conversion matrix calculation (optimization) compared to the technique according to Patent Document 1, and colorimetrically measure RGB signals obtained by photographing. It is possible to derive a color conversion matrix that matches the RGB signal with higher accuracy.

  Further, in the color conversion matrix calculation method according to the present embodiment, the first color conversion matrix is calculated using only the color difference component without using the lightness component, so that the hue / saturation component is optimized. A color conversion matrix with high accuracy can be obtained. This is because the weighting for the color difference component can be increased as much as the lightness component is not included in the first color conversion matrix calculation.

  Further, in the color conversion matrix calculation method according to the present embodiment, as shown in the equation (8), a solution of each matrix element constituting the first color conversion matrix is obtained from the function φ taking into account the weighting constant ωi. May be.

  In this case, it is possible to make a difference in correction accuracy between a certain color and another color. Therefore, it is possible to calculate a color conversion matrix that improves the correction accuracy for a specific color compared to other colors.

  In the color conversion matrix calculation method according to the present embodiment, the second color conversion matrix that is finally set and stored in the imaging apparatus 10 is obtained using Expression (13).

  Since equation (13) is only simple four arithmetic calculations, the second color conversion matrix considering the color difference component and the lightness component can be obtained in a short time using the first color conversion matrix.

<Embodiment 2>
In the first embodiment, the optimization of each matrix element constituting the first color conversion matrix has been described by the iterative solution method of the least square method. In the optimization calculation, the number of iterations until the solution converges differs depending on the value of the first initial value (first starting point in the calculation). Therefore, if a value that is as close as possible to the convergence solution is adopted as the initial value, the number of loops of the iterative solution is also reduced, and as a result, the calculation time until the optimization of the first color conversion matrix can be shortened. .

  In the present embodiment, a flow until the initial initial value in the iterative solution is determined will be described.

  First, any three color patches are selected from the chromatic color patches P1 to P18 from the subject (Macbeth color chart). Here, it is desirable that the color patch is selected from three colors having greatly different hues. For example, it is desirable to select three color patches of red, blue and green, or three color patches of cyan, magenta and yellow, or three color patches of blueish green, orange and magenta. .

  Next, the first color values, which are the colorimetric values described in the first embodiment, are input from the outside to the color conversion matrix calculation device 20 for the selected three colors.

  It is assumed that red, blue, and green in the Macbeth color chart are selected as the three color patches. In this case, the first color value (Rmer, Gmer, Bmer) which is the color measurement result of red, the first color value (Rmeg, Gmeg, Bmeg) which is the color measurement result of green, and the blue color measurement result. The first color value (Rmeb, Gmeb, Bmeb) that is the color result is input to the color conversion matrix calculation device 20.

  Further, as described in the first embodiment, in the color conversion matrix calculation process, the imaging device 10 captures a Macbeth color chart. The imaging device 10 acquires an input color signal (RAW data) by the imaging, and transmits the input color signal (RAW data) to the color conversion matrix calculation unit 20. Here, the input color signal (RAW data) naturally includes the input color signal (RAW data) for the three colors.

  The color conversion matrix calculation device 20 implements using the color conversion matrix M (see Equation (1)) in which the matrix elements m1 to m9 are unknown from the input color signals (RAW data) of the three colors. Each of the second color values described in the first embodiment is obtained.

  It is assumed that red, blue, and green in the Macbeth color chart are selected as the three color patches. In this case, the following three second color values are obtained using the relationship of Expression (2). An input color signal (Rinr, Ginr, Binr), which is an imaging result of a red color patch, is converted using a color conversion matrix M whose matrix elements are unknown numbers, thereby obtaining a first second color value (Resr). , Gesr, Besr). Also, the second color value of the second color value is obtained by converting the input color signal (Rinb, Gimb, Binb), which is the imaging result of the blue color patch, using the color conversion matrix M whose matrix elements are unknown. (Resb, Gesb, Besb). Further, by converting the input color signal (Ring, Ging, Bing), which is the imaging result of the green color patch, using the color conversion matrix M whose matrix elements are unknown, the third second color value is obtained. (Resg, Gesg, Besg).

  Next, the color conversion matrix calculation device 20 configures the color conversion matrix M so that the first color values for the three colors are equal to the second color values for the three colors. Find matrix elements.

  Specifically, the color conversion matrix calculation device 20 creates nine simultaneous equations as shown in Expression (14).

  From the nine simultaneous equations of Expression (14), the color conversion matrix calculation device 20 calculates values of matrix elements m1 to m9. The color conversion matrix M composed of the values of the matrix elements m1 to m9 obtained based on the solution of the simultaneous equations changes the reproduction colors of red, green and blue to the actual colors of the actual Macbeth color chart. Can be matched exactly.

  Next, the color conversion matrix calculation device 20 determines the value of the solution result of the simultaneous equations as the first initial value in the first color conversion matrix calculation process. Then, the color conversion matrix calculation device 20 executes the first color conversion matrix optimization process by the iterative solution of the least square method using the first initial value (first starting point).

  As described above, in the color conversion matrix calculation method according to the present embodiment, the first initial value is determined based on the color values of any three color patches when the first color conversion matrix is optimized. Has been decided.

  Therefore, compared with the cited reference 1 which has set the unit matrix as the first initial value, it has the following effects. That is, the sensitivity characteristic of the imaging device 10 varies depending on the individual imaging device 10. Therefore, when the unit matrix is adopted as the initial initial value, the number of iterations of the iterative solution until the optimization of the color conversion matrix differs depending on the type of the imaging device 10 or the like.

  On the other hand, as can be seen from the above, in the present embodiment, the initial initial value is based on the colorimetric values of any three color patches and the input color signal that is the imaging result of these color patches. The value is determined. In other words, in the present embodiment, the initial value is determined based on the input color signal that varies depending on the imaging device 10, not the fixed value between the different imaging devices 10.

  Therefore, in the present embodiment, the initial initial value can be determined in consideration of the sensitivity characteristics for each imaging device 10 and the like. Therefore, even if the type of the imaging device 10 is different, the number of iterations of the iterative solution until the optimization of the first color conversion matrix can be made equal.

  In addition, it is desirable to select three color patches having greatly different hues. For example, three color patches include red (R), blue (B) and green (G), or cyan (C), magenta (M) and yellow (Y), or blueish green, orange and magenta. Is preferable from the viewpoint of shortening the time until the first color conversion matrix is optimized.

  The time until the optimization of the first color conversion matrix can be further shortened, compared to the case where the hues of the three colors are not greatly different from each other.

  Various color combinations were set, and the time until optimization of the first color conversion matrix was considered. As a result, when RGB, CMY, blueish green, orange, or magenta in the Macbeth color chart is selected as the three-color patch, a solution that can be judged to have sufficiently converged is reached. It was found that this time can be sufficiently shortened compared with the case where the unit matrix is selected as the initial initial value.

  For example, when the convergence time to a predetermined convergence value (convergence solution) was calculated, when the unit matrix was adopted as the initial initial value, the convergence time was 2.125 seconds. Further, when initial initial values were set from the three color patches of RGB in the Macbeth color chart, the convergence time was 0.516 seconds. When the initial values were set from the three color patches CMY in the Macbeth color chart, the convergence time was 0.968 seconds. Further, when initial initial values were set from three color patches of blueish green, orange, and magenta in the Macbeth color chart, the convergence time was 0.062 seconds.

  In addition, as a result of considering the convergence time until convergence to another convergence value different from the predetermined convergence value, as in the case of the above consideration result, when the initial value is derived, blueish green is used as a three-color patch. -It was confirmed that selecting orange and magenta is most preferable from the viewpoint of shortening the convergence time.

  Since the first initial value is determined as described above when deriving (optimizing) the first color conversion matrix, the initial initial value may be shifted at random within a certain range. .

  Further, in the present specification, a color conversion matrix for obtaining a color correction in accordance with the sRGB standard for the RGB signal obtained by the imaging system is obtained. However, the matrix calculation method of the present invention is not particularly limited to calculation of a color conversion matrix for the purpose of color correction. For example, it is also applicable to the calculation of color conversion matrix for mutual conversion of color signals in different color spaces (converting RGB signals to XYZ signals or CMYK signals for printing, converting complementary CMY signals to RGB signals, etc.) it can.

  Note that the imaging device 10 may be a digital video camera, a scanner, an electronic endoscope, or the like in addition to a digital camera.

  In this embodiment, the color conversion matrix calculation device 20 is separated from the imaging device 10. However, the color conversion matrix calculation device 20 described above may be mounted in the imaging device 10.

It is a block diagram which shows the outline of the system which can implement | achieve the color conversion matrix calculation method concerning this invention. It is a figure which shows the external appearance of the Macbeth color chart which consists of a several color patch.

Explanation of symbols

DESCRIPTION OF SYMBOLS 10 Imaging device 20 Color conversion matrix 30 Output device

Claims (6)

A color conversion matrix calculation method for calculating a color conversion matrix used in color correction for a color signal,
(A) preparing a plurality of color patches whose saturation and hue change in stages;
(B) obtaining a first color value by measuring the color patch;
(C) obtaining a second color value by converting an input color signal obtained by imaging the color patch using the color conversion matrix having an unknown matrix element;
(D) calculating the color conversion matrix by obtaining the unknown by a least square method using the first color value and the second color value, and
The step (D)
(D-1) including a step of calculating the first color conversion matrix by the least square method using only the color difference component without using the brightness component,
A color conversion matrix calculation method characterized by the above. The step (B)
Obtaining the first color value for each of a plurality of the color patches;
The step (C)
Obtaining the second color value for each of the plurality of color patches;
The step (D-1)
Calculating the first color conversion matrix by the least square method so that ΣωiΔEi is minimized;
here,
Σ is the sum of the number of the plurality of color patches,
i is an identifier of the color patch;
ω i is a weighting constant,
The sum of the plurality of color patches of the weighting constant ωi is 1,
ΔEi is {(first color difference component of the first color value) − (first color difference component of the second color value)} ² + {( ^second color difference of the first color value) Component) − (the second color difference component of the second color value)} ² .
The color conversion matrix calculation method according to claim 1. The step (D)
(D-2) obtaining a first lightness average value that is an average value of lightness components of the first color value in the plurality of color patches;
(D-3) obtaining a third color value by converting an input color signal obtained by imaging the color patch using the first color conversion matrix;
(D-4) obtaining a second lightness average value that is an average value of lightness components of the third color value in the plurality of color patches;
(D-5) By multiplying each matrix element constituting the first color conversion matrix by a value obtained by dividing the first lightness average value by the second lightness average value, A step of calculating a second color conversion matrix;
The color conversion matrix calculation method according to claim 1. The step (D-1)
(D-1-1) selecting any three color patches;
(D-1-2) obtaining the first color value in the three color patches;
(D-1-3) obtaining the second color value from an input color signal obtained by imaging the three color patches;
(D-1-4) The first color value obtained in step (D-1-2) is equal to the second color value obtained in step (D-1-3). Obtaining the unknown, so that
(D-1-5) calculating the first color conversion matrix by the least square method using the unknown obtained in step (D-1-4) as an initial initial value. ing,
The color conversion matrix calculation method according to claim 1. The three color patches are:
Red, blue and green,
The color conversion matrix calculation method according to claim 4. The three color patches are:
Cyan, magenta and yellow,
The color conversion matrix calculation method according to claim 4.

Images