JP2018106558A - Method for measuring color of object - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a method for measuring color of an object, which can acquire and generate color information usable to reproduce a color closer to a color perceived by a person from a photographed image.SOLUTION: In a method for measuring the color of an object including acquiring an image of the object, acquiring RGB values from the image, converting the RGB values to XYZ values, and converting the XYZ values to Labvalues, the conversion from the RGB values to the XYZ values is performed by using a conversion formula [XYZ]=M[SS... S], [Mis a 3×n transformation matrix, S, S, ..., Sare input values based on the RGB values acquired from the image, and n=8 to 14].SELECTED DRAWING: Figure 1

Description

  The present invention relates to a method for measuring the color of an object.

  In recent years, with the increase in remote communication and electronic commerce, it is required to reproduce colors that are faithful to the actual colors of objects. For this purpose, it is necessary to acquire and share color information that can reproduce colors closer to colors perceived by human eyes.

As means for obtaining color information relatively easily, it is known to obtain an RGB value by taking an image using an RGB colorimetric imaging means (camera or the like). Here, in the RGB color system, a negative part occurs in the color matching function, which causes inconvenience when calculating the color value. Therefore, after converting the RGB value obtained from the image into an XYZ value in which a negative part does not occur in the color matching function, it is converted into another color system (L ^* a ^* b ^* color system, etc.) Often used (for example, Patent Document 1).

JP 2010-159230 A

  When converting RGB values obtained from an image into XYZ values, it is known to perform linear conversion using a matrix. However, in fields such as cosmetics, containers, clothing, paints, etc. where color differences greatly affect product evaluation, conventional color information acquisition and generation methods are sufficient to reproduce colors close to the perceived color. In some cases, this was not possible.

  In view of the above points, according to one aspect of the present invention, the color of an object can be obtained by generating color information that can reproduce a color closer to the color perceived by human eyes from a captured image. It is an object to provide a method for measuring.

In order to solve the above problems, a method for measuring the color of an object according to an embodiment of the present invention acquires an image of the object, acquires RGB values from the image, and converts the RGB values to XYZ values. And converting the XYZ values into L ^* a ^* b ^* values, and converting the RGB values into the XYZ values using a conversion formula

[Wherein M _{3 × n} is a 3 × n transformation matrix, and S ₁ , S ₂ ,..., S _n are input values based on RGB values acquired from the image, and n = 8 ~ 14]
To do.

  According to an embodiment of the present invention, color information that can reproduce colors closer to colors perceived by human eyes from captured images can be acquired and generated.

FIG. 6 is a flow diagram illustrating a method according to an aspect of the present invention. 1 is a schematic view of an apparatus for carrying out an embodiment of the present invention.

  FIG. 1 shows a flow diagram of a method for measuring the color of an object according to one aspect of the present invention. FIG. 2 shows a schematic diagram of an apparatus for carrying out the method according to the present embodiment. In FIG. 2, the measurement object 14, the light source 10 for irradiating the measurement object 14 with light, the imaging means 12 for taking an image of the measurement object 14, and the data obtained by the imaging means 12 are processed. The calculation means 18 for this is shown.

  FIG. 2 is a view of the apparatus as viewed from above, and the light source 10, the imaging means 12, and the measurement object 14 are arranged substantially horizontally. The imaging means 12 is provided in front of the measurement object 14, and the light source 10 is arranged so that light from the light source 10 is irradiated at an angle with respect to the surface of the measurement object 14. Yes. The light source 10 may be arranged so that light is irradiated at an angle different from that shown in the figure as long as it does not interfere with the imaging of the measurement object 14 by the imaging means 12.

  Note that it is preferable to use a light source 10 having high color rendering properties. Specifically, the average color rendering index Ra obtained based on JIS Z 8726 of the light source 10 is preferably 90 or more, and more preferably 95 or more. Further, the light source 10 is preferably arranged at a distance of 50 cm or less from the measurement object 14.

  In the method according to this embodiment, first, an image of an object whose color is to be measured is acquired by the imaging device 12 (S1). The imaging device 12 is not particularly limited as long as it can acquire color information of an object. However, it is preferable to use a cooled CCD camera because there is little noise. For example, a BU-55C manufactured by BITRAN is used. Can be used.

  When a camera is used as the image pickup means 12, it is preferable to remove so-called dark noise at the time of image pickup. Specifically, an image (dark image) is acquired with the camera lens capped, and then each RGB channel of the camera can be adjusted so that the pixel value of the dark image becomes zero.

  Furthermore, it is preferable to perform luminance unevenness correction to obtain an image with uniform brightness. Specifically, an image (gray image) is taken on a gray plate (gray card) with a reflectance of 18% under an LED light source, and each channel of RGB is adjusted so that the pixel values of the entire screen become a predetermined value. . For example, if the camera has a pixel value (luminance) range of 0 to 65535, the acquired gray image is adjusted to have a pixel value of 11796.

  Further, in addition to the above-described dark noise removal and luminance unevenness correction, correction using a color chart can also be performed. For the color chart, Color Checker Passport (24 colors) manufactured by X-rite can be used. After the color chart is imaged to obtain an image (color chart image), the dark noise removal and the luminance unevenness correction are performed, the luminance of the color chart image is finally less than 0.18 times, preferably 0. Adjust to 15 times or more and less than 0.18 times. Since the reflectance of the gray image is 18%, it is originally set to 0.18 times, but by providing a buffer, color information closer to the apparent color can be obtained.

  The measurement object 14 has a measurement object surface 14a on the side directed to the imaging means 12, and an object whose color is to be measured can be arranged or applied to the measurement object surface 14a. Although all objects having color can be used as the measurement object 14, the method according to the present embodiment uses the surface on which a colored composition such as cosmetics or paint is applied as the measurement object, and provides color information. It is preferably used for obtaining. In particular, the subtle color difference greatly affects the selection and evaluation of products and is suitably used in cosmetics that require highly accurate color measurement. Especially, it can use suitably for the measurement of the color of cosmetics, such as a lipstick, an eye shadow, a blusher, and a foundation, especially a lipstick and an eye shadow.

  In addition, when the target object which performs a color measurement is a semi-solid with a small projected area like cosmetics, such as a lipstick, it is preferable to apply a target object to the measurement target surface 14a. In that case, it is possible to use a material in which pseudo skin is prepared, pseudo skin is applied to the measurement target surface 14a, and cosmetics are applied thereon. Moreover, the face of the person to whom cosmetics were given can also be made into a measuring object.

After taking an image using the apparatus of FIG. 2 (S1), RGB values are acquired from the taken image (S2). The resulting RGB values is finally (equal to the amount of the color value variation and human feel ^change) perceptual high homogeneity L ^* a * ^{b *} values (CIE1976L ^* a ^* b ^* However, in order to avoid the inconvenience at the time of color value calculation as described above, the RGB values are once converted into XYZ values (CIE 1931 XYZ color system) that do not depend on the device (S3).

  In the present embodiment, the following matrix conversion formula is used for conversion from RGB values to XYZ values.

[M _{3 × n} is a 3 × n transformation matrix, and S ₁ , S ₂ ,..., S _n are input values based on RGB values acquired from the image, and n = 8 ~ 14]
Is used.

  Regarding conversion from RGB values to XYZ values, it is typically known to perform conversion using a 3 × 3 conversion matrix (matrix) with R, G, and B values as input values. . However, in this embodiment, it is possible to acquire color information (high color reproducibility) that can reproduce colors closer to the color recognized by human eyes by using the above conversion matrix different from the conventional one. It becomes.

The input values S ₁ , S ₂ ,..., S _n may include multi-order terms, and may include secondary terms, for example. Further, R ² , G ² , and B ² may be included as input values, and RG, GB, and BR may be included.

Further, it is preferable that the third order term includes a product obtained by multiplying each value of R, G, and B, that is, RGB. Moreover, it is preferable to include the cube value of each value of R, G, and B, that is, R ³ , G ^3, and B ³ . This is considered to be because correction of the dark portion RGB value is emphasized by providing the cube term.

The input value preferably includes all 0th to 3rd order terms. In particular, the input values S ₁ , S ₂ ,..., _Sn are 1, R, G, B, R ² , G ² , B ² , RG, GB, BR, R ³ , G ³ , B ³ , It is preferable that the conversion matrix includes RGB and M is 3 × 14.

In addition, it is preferable that the input values S ₁ , S ₂ ,..., S _n do not include a fifth or higher order term, and more preferably do not include a fourth or higher order term.

In the generation of M _{3 × n} , software installed in the calculation means 18 shown in FIG. 2, for example, a personal computer or the like can be used. Specifically, in _order to generate M _{3 × n} , RGB values are acquired by imaging for a plurality of predetermined colors, and true values of XYZ values for the same colors are prepared. As the true value, for example, an XYZ value obtained from a measurement value of a spectral radiance meter evaluated that color information very close to human vision can be obtained can be used. The plurality of colors are preferably 24 colors or more, preferably 140 colors or more by using a color chart or the like.

Furthermore, M _{3 × n} can be obtained by introducing the RGB value (obtained by imaging) obtained for each color and the true value of the XYZ value into the above equation (1) and performing optimization. The optimization can be performed using software on the personal computer 18, for example. For example, MATLAB software (registered trademark), which is optimization software of MathWorks, or solver function of Office Excel (registered trademark) of Microsoft Corporation can be used.

Based on the equation (1) including M _{3 × n} obtained as described above, the RGB values acquired in S2 are converted into XYZ values (S3). Further, the XYZ value is converted into an L ^* a ^* b ^* value (S4). In the conversion, a known conversion formula can be used.

  In recent years, it is possible to obtain color information with high color reproducibility by measuring chromaticity and luminance using a spectral radiance meter. However, measurement with a spectral radiance meter is not suitable for practical use because it takes time. On the other hand, according to the present embodiment, since color information of RGB values can be used, an imaging means such as a camera can be used. Therefore, highly reproducible color information can be acquired easily in a short measurement time.

  Hereinafter, an embodiment of the present invention will be described in more detail based on examples and comparative examples.

  In the apparatus described with reference to FIG. 1, a color chart is arranged on the measurement target surface 14 a of the measurement target 14. As the color chart, Color Checker DIGITAL SG (140 colors) manufactured by X-rite was used. In addition, as the light source 10, Rich Color No. 11 (manufactured by Sharp Corporation, color rendering index Ra = 96) was used. As the camera 12, a cooled CCD camera manufactured by BITRAN, BU-55C (set at −10 ° C.) was used.

  Light from the light source 10 was applied to the color chart on the measurement target surface 14a, an image of the entire color chart was taken by the camera 12, and RGB values were obtained for all 140 colors of the color chart from the obtained image.

On the other hand, the color value of the color chart was also measured with a spectral radiance meter, and based on the measurement, an XYZ value was obtained for each color, and this was regarded as a true value (X ₀ , Y ₀ , Z ₀ ). Since the spectral radiance meter acquires color information based on the reflectance for each wavelength in the reflection spectrum of the measurement object, it is generally evaluated that color information very close to human vision can be obtained. In this study, a CS-2000 spectral radiance meter manufactured by KONICA MINOLTA was used as the spectral radiance meter.

  Next, a conversion matrix for converting RGB values into XYZ values was created. As a conversion formula for conversion from RGB values to XYZ values, the following formula was used.

Where M _{3 × n} is a 3 × n transformation matrix (matrix),

It is. In the matrix on the right side of Expression (2), the alphabets a, b, and c on the left side of each component are codes representing rows, and the numbers 1, 2,..., N on the right side are codes representing columns. Further, _{S 1,} S 2 in formula (1), ···, _{S n} is an input value based on the RGB values.

  Several types of transformation matrices in which n in Equation (1) was changed were created, and the degree of difference from the true value of XYZ was verified.

Specifically, the RGB values (obtained from the camera 12) and true values X ₀ , Y ₀ , Z ₀ (obtained from the measured values of the spectral radiance meter) respectively obtained for the color chart 140 colors as described above are obtained. The transformation matrix M _{3 × n} was obtained by inputting into the above equation (1) and further performing optimization. For optimization, Matlab (registered trademark), which is optimization software of MathWorks, was used.

Example 1
Similarly, n = 8 in equation (1), and the following equation was used as the transformation matrix.

  The above-mentioned optimization was performed, and the conversion matrix shown in the following table was obtained.

  Columns (a to c) in Table 1 represent rows in the above formula (2), and rows (1 to 8) represent columns in the above formula (2).

(Example 2)
Similarly, n = 11 in equation (1), and the following equation was used as the transformation matrix.

  The above-mentioned optimization was performed, and the conversion matrix shown in the following table was obtained.

(Example 3)
Similarly, in equation (1), n = 14, and the following equation was used as the transformation matrix.

  The above-mentioned optimization was performed, and the conversion matrix shown in the following table was obtained.

(Comparative Example 1)
In Equation (1), n = 3, and the following equation was used as the transformation matrix.

  The above-mentioned optimization was performed, and the conversion matrix shown in the following table was obtained.

(Comparative Example 2)
Similarly, n = 17 in equation (1), and the following equation was used as the transformation matrix.

  The above-mentioned optimization was performed, and the conversion matrix shown in the following table was obtained.

(Comparative Example 3)
Similarly, n = 20 in equation (1), and the following equation was used as the transformation matrix.

  The above-mentioned optimization was performed, and the conversion matrix shown in the following table was obtained.

In each example of Examples 1 to 3 and Comparative Examples 1 to 3, RGB values acquired from the camera 12 for each color of the color chart were converted into XYZ values using the obtained conversion matrix. Further, the difference between the XYZ values (X, Y, Z) obtained by the conversion and the above-mentioned true values (X ₀ , Y ₀ , Z ₀ ) is taken, and the absolute values of the differences are respectively expressed as E _X , E _Y , E _Z.

The larger the values of E _X , E _Y , and E _Z , the greater the distance from the true value (XYZ values measured by a spectral radiance meter), and the farther from the true value, the lower the color reproducibility. become. Conversely, the smaller the values of E _X , E _Y , and E _Z , the more accurate color information is obtained.

Since E _X , E _Y , and E _Z are obtained for each color of the color chart, the maximum value and the average value were obtained as representative values and compared. The results are shown in Table 7.

According to Table 7, in the case of Examples 1 to 3 (n = 8 to 14), the average of the maximum values of E _X , E _Y , E _Z is 5.5 or less, and E _X , E _Y The average of the average values of _EZ is 1.6 or less, and it can be seen that color information closer to the color information obtained using the spectral radiance meter is obtained. In particular, when a 3 × 14 matrix is used, the maximum values and average values of E _X , E _Y and E _Z are all low, and a result closer to the true value is obtained.

  As described above, according to the present embodiment, it is possible to acquire and generate color information that can reproduce colors closer to colors perceived by human eyes, and thus accurate color information can be shared. The method of this embodiment can be used for preparation of cosmetics, counseling, and the like.

DESCRIPTION OF SYMBOLS 10 Light source 12 Imaging means 14 Measurement object 14a Measurement object surface 18 Calculation means

Claims (4)

Get an image of the object,
Obtain RGB values from the image,
Converting the RGB values into XYZ values;
Converting the XYZ values into L ^* a ^* b ^* values,
Conversion from the RGB value to the XYZ value

[M _{3 × n} is a 3 × n transformation matrix, and S ₁ , S ₂ ,..., S _n are input values based on RGB values acquired from the image, and n = 8 ~ 14]
A method for measuring the color of an object using Wherein said _{S 1,} S 2, · · ·, is _{S n,} the R value of the RGB values, including G and B values each of the three squares, measuring the color of an object according to claim 1.   The method of measuring a color of an object according to claim 2, wherein the n is 14. 4. Acquiring the image includes capturing the image using a light source and an imaging means;
The method for measuring the color of an object according to any one of claims 1 to 3, wherein an average color rendering index Ra of the light source is 95 or more.

Images