JPH07174631A - Skin colorimetric method and spectral reflectance spectrum estimation method - Google Patents

Abstract

(57) [Summary] [Purpose] To make skin color measurement easy and accurate.
The tristimulus values X, Y, Z of the skin are obtained based on the output values of the color video camera, and the obtained tristimulus values X, Y,
The spectral reflection spectrum can be accurately estimated based on Z. [Arrangement] A predetermined observation object is photographed by a color video camera in advance to obtain output values R, G, B of the camera, while the observation object is color-measured by a spectrocolorimeter and a tristimulus value X is obtained. , Y, Z are obtained, and a conversion matrix of the camera output values R, G, B and the tristimulus values X, Y, Z obtained by the spectrocolorimeter is used as at least the camera output values R, G, B. It is determined by multiple regression analysis as a matrix using up to the second order term. Using this transformation matrix, the camera output values R, G, B for any skin can be converted into tristimulus values X, Y, Z.
Convert to. Furthermore, the spectral reflectance spectrum of the skin is estimated from the obtained tristimulus values X, Y, and Z by the principal component analysis method.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a skin colorimetric method and a spectral reflection spectrum estimation method. More specifically, the present invention obtains output values R, G and B for red (R), green (G) and blue (B) by photographing skin with a color video camera. The present invention relates to a skin colorimetric method for obtaining tristimulus values X, Y, Z based on output values, and a method for estimating a spectral reflection spectrum of skin based on the obtained tristimulus values X, Y, Z.

[0002]

2. Description of the Related Art Conventionally, cosmetics such as foundations and lipsticks have been applied to the skin to obtain a desired color. The appearance color of the cosmetic and the application after the cosmetic is actually applied to the skin. Since it is significantly different from the color, it is difficult for a person who wears makeup to estimate the applied color after applying the cosmetic to the skin. Therefore, it is also difficult to select the most suitable cosmetic product to obtain the desired skin color.

Therefore, various skins are photographed in advance by a color video camera, the output values of R, G, and B at that time are recorded as the colorimetric results, and R when the specific makeup is applied to the skins. , G, B output values are recorded, and the texture in that case (how the observer feels) is recorded and stored as data before and after makeup. Then, for the skin for which a new cosmetic product is selected, the output values of R, G, B are similarly obtained by photographing with a color video camera, and the output values of R, G, B are accumulated. , The type and amount of makeup to be applied to the skin is determined.

[0004]

However, in the conventional color measurement method using a color video camera, the R, G, B values used as the color measurement result are different when the person observes the skin. It is different from the tristimulus values X, Y, and Z that the color looks like and how it feels. Therefore, there is a problem that the skin color cannot be evaluated well by the output values of R, G, and B. Also, since the optical properties of skin and cosmetics are diverse, it is impossible to store the R, G, B values and texture of the skin when applying any cosmetics to all kinds of skin as data. Is. Therefore, there is a problem that a desired color or texture cannot be obtained even if the color of the skin is measured using a color video camera and a cosmetic is selected based on the color measurement result.

On the other hand, estimating the application color of the skin to which cosmetics are applied is important in developing cosmetics. However, in the conventional method, it is not possible to accurately estimate the coating color when arbitrary cosmetics are applied to arbitrary skin.Therefore, in the actual development process, trial production is carried out by making a prototype and then conducting an application test. However, there was a problem in that efficient product development was hindered because many process steps were required.

In order to solve such a problem, it is conceivable to accurately measure the tristimulus values X, Y and Z by using a spectrocolorimeter to measure colors without using a color video camera. However, the spectrocolorimeter cannot measure the color of multiple points at the same time, which causes the following problems. That is, since a face to which makeup is applied generally includes a plurality of areas having different textures such as a forehead, a cheek, a nose, a lips, a jaw, and a nape of the neck, when measuring colors, multipoints belonging to each area are measured. Will be required. For this reason, in the case of colorimetry using a spectrocolorimeter that is incapable of multicolor measurement at the same time, it is necessary to repeat the measurement separately for each measurement point. Therefore, the labor of measurement becomes very complicated, and the colorimetric method cannot be practically used.

Further, in color measurement using a color video camera, it is conceivable to design the video camera so that the image input sensitivity of the color video camera matches the color matching function of the human eye. However, this method is not technically easy.

The present invention is intended to solve the problems of the prior art as described above, and when the color of the skin is measured using the conventional color video camera, the tristimulus values X, Y and Z are accurately determined. In addition, it is possible to estimate the spectral reflection spectrum based on the tristimulus values X, Y, and Z, and thereby to accurately predict the coating color of arbitrary cosmetics applied to arbitrary skin. The purpose is to do.

[0009]

The inventors of the present invention have found that the output values R, G, B and tristimulus values X, Y, Z of a color video camera.
The conversion matrix of and can be obtained with high conversion accuracy by using at least quadratic terms of the output values R, G, B of the color video camera, and the tristimulus values X, Y can be obtained by applying the principal component analysis. , Z, it has been found that the spectral reflectance spectrum of the skin can be well estimated, and the present invention has been completed.

That is, according to the present invention, the skin is photographed by a color video camera to obtain output values R, G, B for red (R), green (G) and blue (B), and the output value R of the camera. , G,
In the method of measuring the color of the skin based on B, a predetermined observation object is photographed with a color video camera in advance to obtain output values R, G, B of the camera, while the observation object is spectroscopic colorimeter. The tristimulus values X, Y, and Z are obtained by colorimetric measurement using the conversion matrix of the camera output values R, G, and B and the tristimulus values X, Y, and Z obtained by the spectrocolorimeter. The output values R, G, B of at least the second-order terms are determined by multiple regression analysis, and this conversion matrix is used to determine the output values R of the camera for any skin.
Provided is a skin colorimetric method characterized by converting G and B into tristimulus values X, Y and Z.

The present invention also provides a method for estimating a spectral reflection spectrum of the skin by the principal component analysis method from the tristimulus values X, Y and Z obtained as described above.

The present invention will be described below in detail with reference to the drawings.

FIG. 1 is a diagram showing a general outline of the method of the present invention. As shown in the figure, in order to obtain the tristimulus values X, Y, and Z of an observation object such as a face in the present invention, first, in step A, the observation object is photographed using a color video camera, and R, The output values of G, B are obtained, and then the obtained output values of R, G, B are processed in step B to produce tristimulus values X, Y, Z.
Convert to. Further, in the present invention, when the spectral reflection spectrum of the observation object is obtained, the tristimulus values X, Y and Z obtained in the step B are converted into the spectral reflection spectrum in the step C. Hereinafter, each step will be described in detail.

As described above, the step A is a step of photographing the object to be observed using a color video camera and obtaining the output values of R, G and B. Here, when obtaining the output values of R, G, and B, only the information about the color is obtained as the output value without being influenced by the surface unevenness that is recognized as surface gloss or gloss. It is preferable. For that purpose, a polarization filter may be used in the image input unit according to a known method to cut off the surface reflected light of the observation object so that only the internal reflected light is received (for example, JP-A-2- No. 206426, Japanese Patent Application No. 5
-247523 specification).

A commercially available color video camera can be used.

In step B, the output values R, G, obtained in step A,
Convert B to tristimulus values X, Y, Z. In this case, if the output values R, G, B do not have a linear relationship with the incident luminance, the output values R, G, B are corrected so as to have a linear relationship with the incident luminance prior to conversion. Preferably. That is, the output values R ₀ , G ₀ , and B ₀ when the observation target is photographed by the color video camera are generally represented by the following equations (1) to (3), respectively.

[0017]

[Equation 1] (Where E (λ) is the emission spectrum of the light source, O (λ) is the spectral reflectance, and r (λ), g (λ), and b (λ) are the spectra of the lens, the ultraviolet cut filter, and the polarization filter, respectively. The spectral imaging characteristic, which is the product of the transmission characteristic, the spectral transmission characteristic of the color separation filter, and the spectral conversion characteristic of the image sensor is expressed.) However, in reality, various processing causes non-linearity. , Fb, the actual output values R, G, B are expressed by the following equations (4) to (6).

[0018]

[Equation 2] (In the formula, Kr, Kg, and Kb each represent a white balance coefficient.) Therefore, generally, output values R, G, and
B has no linear relationship with the incident luminance. However, in order to accurately convert the output values R, G, B into the tristimulus values X, Y, Z, it is preferable that the output values R, G, B have a linear relationship with the incident luminance. Therefore, the output value R,
It is preferable to correct G and B so that they have a linear relationship with the incident luminance. The correction method in this case is not particularly limited, and for example, output values R, G, and
The correction equation may be obtained by using the terms up to the second order of B.

In step B of the present invention, the output values R, G, B of the color video camera (preferably the output values R, G, B corrected as described above) are converted into tristimulus values X, Y, Z. As a method, at least the quadratic terms of the output values R, G, and B are used, and the conversion matrix is obtained. That is, the tristimulus values X, Y, and Z of the observation target are expressed by the following equations (7) to
It is expressed as (9).

[0020]

[Equation 3] (In the formula, K is a constant, Es (λ) is a radiation characteristic of a standard light source,
O (λ) is the spectral reflectance of the observation object, x (λ), y
(Λ) and z (λ) represent color matching functions, respectively. ) Thus, the tristimulus value X represented by the formulas (7) to (9),
Since the color matching functions of Y and Z are different from the color separation characteristics of the color video camera, the expressions (1) to (3) or (4) are used.
~ The output value R of the color video camera represented by (6),
It is different from G and B. Therefore, it is not possible to satisfactorily evaluate the texture, which is the actual appearance, by the output values R, G, B of the color video camera. In order to be able to satisfactorily evaluate the actual appearance texture based on the output values R, G, B of the color video camera, the output values R, G, B of the color video camera are set to the tristimulus values X, Y. , Z is required.

Up to now, the average color difference ΔE ^{* has been} obtained by linearly converting the output values R, G, B of the color video camera ^.
When ab = 4.6, the output values R, G, B are set to tristimulus values X, Y, Z.
(Ikeda, Abe, Higaki, Nakamichi, "Optimal Color Matching of Color Video Cameras", Journal of Television Society, 44 , 1750 (1990)). However, since the skin color is a color that humans have high ability to identify,
With such an average color difference value, it cannot be said that the conversion accuracy is sufficient for skin color.

Therefore, in the present invention, for example, as in the following equation (I),

[0023]

[Equation 4] Using up to the quadratic term of the output values of R, G, B, R, G,
A conversion matrix of the output value of B and the tristimulus values X, Y, and Z is obtained. More specifically, a specific video object is photographed with a color video camera to obtain its output values R, G, B, and a spectrocolorimeter also measures the color of the video object with tristimulus values X, Y. , Z. Then, a conversion matrix M for converting the output values of R, G, B into tristimulus values X, Y, Z is directly obtained by the tristimulus values X, Y, Z obtained by the conversion and the spectrocolorimeter. It is determined by multiple regression analysis so that the difference between the three tristimulus values X, Y, and Z is minimized. This makes it possible to obtain a conversion matrix with high conversion accuracy. When obtaining the transformation matrix in this way,
It is sufficient to use at least the second-order terms of the output values of R, G, and B, and higher-order terms may be used if necessary.

After obtaining the transformation matrix, an arbitrary skin is photographed by this color video camera and its output values R, G,
B is obtained and the obtained R, G, B values are transformed into tristimulus values X, Y, Z using the transformation matrix determined as described above. Thus, the output values R, G, B of the color video camera
By obtaining the tristimulus values X, Y, and Z from, it becomes possible to evaluate any skin according to the color and texture actually felt by the observer.

In step C of the present invention, the spectral reflectance spectrum of the skin is estimated from the tristimulus values X, Y and Z obtained in step B using the principal component analysis method.

That is, it is generally impossible to obtain the spectral reflection spectrum of the skin in the visible region of 400 to 700 nm from the three values of the tristimulus values X, Y and Z of the skin. However, when the spectral reflectance spectrum of the skin is estimated from the tristimulus values X, Y, and Z of the skin, if three main components are used, the cumulative contribution ratio is 98.1% for the forehead and 97.2 for the cheek. %, And the lips can be restored at 98.0% (J. Soc. Cosme
t. Chem. Japan, Vol 13, No. 1,7
(1979)). Therefore, in the present invention, the tristimulus value X obtained in the step B is utilized by utilizing these principal component analysis methods.
The spectral reflectance spectrum of the skin is estimated from Y and Z.

In this case, it is preferable to perform the principal component analysis based on the average color of the skin for which the spectral reflection spectrum is to be estimated, because the cumulative contribution rate up to the third principal component can be improved. For example, assuming that the spectral reflectance O (λ) is represented by the following equation (10), the principal component analysis is performed.

[0028]

[Equation 5] (In the formula, O ⁻ (λ) is the average value P ₁ (λ), P of the skin color.
₂ (λ) and P ₃ (λ) are main components, W ₁ , W ₂ , and
W ₃ is a principal component score (weight), respectively. ) Substituting this equation (10) into the above equations (7) to (9), the following equations (11) to (13) are obtained.

[0029]

[Equation 6] Tristimulus values X, Y by the principal component i (i = 1, 2 or 3),
Xi, Yi, and Zi are increments of Z, and average values are X ⁻ , Y ⁻ ,
If Z ⁻ , the above equations (11) to (13) are represented by the following equations (14) to (16).

[0030]

[Equation 7] If this is expressed by a determinant, the following expression (17) is obtained.

[0031]

[Equation 8] That is, the principal component scores W ₁ , W ₂ , W ₃ are tristimulus values X, Y,
Therefore, Z has the relation of the following expression (18).

[0032]

[Equation 9] Thus, any tristimulus values X, Y, it is possible to obtain the principal component score _{W 1,} W 2, _{W 3} relative to Z, resulting principal component score _{W 1,} W 2, _{W 3} Expression By substituting into (10), it becomes possible to estimate the spectral reflection spectrum O (λ).

As described above, the flesh color spectral reflection spectrum O (λ)
In the same way, the average color of the skin is appropriately determined and the principal component analysis is performed in the same manner, so that the spectral reflection spectrum of any skin, for example, the spectral reflection spectrum of the lips can be estimated with high accuracy. It will be possible.

The spectral reflection spectrum thus obtained is very useful in the cosmetic development process.
For example, by analyzing the spectral reflectance spectrum of the cosmetic and the spectral reflectance spectrum of the skin together, it is possible to satisfactorily estimate the coating color when the cosmetic is applied to the skin.

Further, by analyzing the spectral reflection spectrum thus obtained and the separately measured irregularity information of the observation object, for example, the darkness due to the color of the skin itself and the darkness due to the shadow can be accurately identified. become. Therefore, it is possible to favorably evaluate the comprehensive texture of the observation object.

[0036]

According to the color measuring method of the present invention, since the color is measured based on the output values R, G, B obtained by photographing the skin with a color video camera, it is possible to easily measure the color of the skin at multiple points simultaneously. Is possible.

Further, according to the colorimetric method of the present invention, the output values R, G and B of the color video camera are set to the tristimulus values X, Y and
When obtaining the conversion matrix to be converted into Z, the output value R of the color video camera,
G and B are obtained and the tristimulus value X,
Y and Z are obtained, and the output value R of the color video camera,
By using at least the quadratic terms of G and B, the conversion matrix of the camera output values R, G and B and the tristimulus values X, Y and Z obtained by the spectrocolorimeter is subjected to multiple regression analysis. Ask. Therefore, by using this conversion matrix, the output values R, G, B of the color video camera can be converted into tristimulus values X, Y, with high conversion accuracy for an arbitrary observation object.
It becomes possible to convert to Z.

Further, according to the method for estimating the spectral reflection spectrum of the present invention, the output value R of the color video camera,
Since the principal component analysis is applied to the tristimulus values X, Y, and Z obtained based on G and B to estimate the spectral reflection spectrum,
It is possible to obtain a spectral reflection spectrum without using a spectrocolorimeter.

[0039]

EXAMPLES The present invention will be described in detail below based on examples.

Example 1 [Correction of Incident Luminance Sensitivity Characteristic] As an observation object, nine achromatic color chips (Nos. 1 to 9) shown in Table 1 were prepared. Also,
As a color video camera, a high-definition still image camera (HC-1500 manufactured by Nikon Corporation) is used, white balance correction is performed with a barium sulfate plate, each of the above color charts is photographed, and the output value R of the camera at that time, G and B were calculated. In this case, the light source is a metal halide lamp (RD) with a color temperature of 5700K.
SMI, HMI light) was used, and an ultraviolet cut filter was used. In addition, a polarizing plate was attached to the light source and the front surface of the camera to remove the surface reflected light of the observation object and photographed.

In this photographing, the incident luminance on the camera changes depending on the color chart to be photographed, but the incident luminance (relative luminance) at this time and the obtained output values of R, G, B of the camera The relationship is shown in FIG.

[0042]

[Table 1] Tristimulus values color chart No. XYZ 1 65.40 66.70 75.80 2 55.00 56.30 65.60 3 47.20 48.30 56.80 4 27.80 28.50 34.40 5 19.00 19.50 23.80 6 15.00 15.40 19.00 7 11.70 12.00 15.00 8 6.65 6.81 8.55 9 4.49 4.58 5.65 As shown in FIG. 2, none of the R, G, and B output values had a linear relationship with the incident luminance. This is the above equation (4)-
It is considered to be due to the nonlinear functions fr, fg, and fb of (6). The output values R, G, B of the camera are accurately set to the tristimulus value X,
In order to convert into Y and Z, it is preferable that the output values R, G, and B have a linear relationship with the incident luminance, so that the R, G, and B of the following equations (19) to (21) are obtained. Fitting was performed with a quadratic function.

[0043]

[Equation 10] As a result of such fitting, as shown in FIG. 3, the corrected R ′, G ′, and B ′ showed a good linear relationship with the incident luminance. Also, each R after fitting
The correlation coefficient between ', G', B'and the incident luminance was 1.00.

[Tristimulus value X of output values R ', G', B ',
Conversion to Y, Z] As a sample of the skin color region, 39 color chips (Huns of Munsell color system H = 0YR to 10YR, lightness V
= 5 to 8 and saturation C = 2 to 5), photograph them with a color video camera, and output values R'and G'as described above.
, B '. On the other hand, a spectrophotometer (Minolta product,
CM1000) was used to determine tristimulus values X, Y, and Z (C light source, 2 ° field of view) for each of the 39 color chips. Then, as in the following expression (22), a conversion matrix of the camera output values R ′, G ′, B ′ including the quadratic terms and the tristimulus values X, Y, Z obtained by the spectrocolorimeter. M _(i) was determined by multiple regression analysis.

[0045]

[Equation 11] As a result, the following transformation matrix M _(i) was obtained. However, the following shows the transposed matrix M _(i) ^T of M _(i).

[0046]

[Equation 12] For each color chart, the output values R ′, G ′, B ′ of the camera are converted into tristimulus values X, Y, Z using this conversion matrix M _(i).
To L ^* a ^* b ^* (CIE197
6L ^* a ^* b ^* ), and L ^* a ^* b ^* based on the tristimulus values X, Y, and Z obtained by the spectrocolorimeter, and the color difference (ΔE ^* ab) between them. It was The result is shown in FIG. Here, the average color difference was 0.60. The maximum color difference was 1.9. From the results of this example, according to the method of the present invention, the tristimulus value X, from the output value of the camera,
It can be seen that Y and Z can be converted accurately.

Comparative Example 1 When obtaining the conversion matrix from the output values R, G, B of the camera to the tristimulus values X, Y, Z, the output value R'of the camera
, G ', B'up to the first-order terms were used, and the matrix M _(ii) of the following equation (23 ₎ was obtained in the same manner as in Example 1.

[0048]

[Equation 13] As a result, the following transformation matrix M _(ii) was obtained. However, the following shows the transposed matrix M _(i) ^T of M _(i).

[0049]

[Equation 14] For each color chart, the output values R ′, G ′, B ′ of the camera are converted into tristimulus values X, Y, Z using this conversion matrix M _(ii).
To obtain L ^* a ^* b ^* based on it, while
L based on tristimulus values X, Y, Z obtained by the spectrocolorimeter
^{* A *} b ^* was obtained, and the color difference (ΔE ^* ab) between them was obtained. The result is shown in FIG. Here, the average color difference is 0.
99, the maximum color difference was 2.3. It can be seen that the conversion accuracy is inferior to that of the embodiment.

Comparative Example 2 In obtaining the conversion matrix from the camera output values R, G, B to the tristimulus values X, Y, Z, R, G before correcting the incident luminance sensitivity characteristic as the camera output value. , B, and the first-order terms of the output values R, G, B of the camera are used in the same manner as in Example 1 to obtain a matrix M _(iii) of the following equation (24).

[0051]

[Equation 15] As a result, the following transformation matrix M _(iii) was obtained. However, the following shows the transposed matrix M _(iii) ^T of M _(iii).

[0052]

[Equation 16] For each color chart, the output values R ′, G ′, B ′ of the camera are converted into tristimulus values X, Y, using the conversion matrix M _(iii) .
Converted to Z, it obtains the ^L * ^a * ^{b *} based, whereas, spectral colorimetric tristimulus values obtained by the device X, Y, determine the ^L * ^a * ^{b *} based on the Z, both of the color difference ( ΔE ^* ab) was determined. The result is shown in FIG. Here, the average color difference was 2.00 and the maximum color difference was 6.4. It can be seen that the conversion accuracy is inferior not only to Example 1 but also to Comparative Example 1. Therefore, in order to improve the conversion accuracy, it is effective to correct the incident luminance sensitivity characteristic for the output value of the camera.

Example 2 54 females in their twenties to fifties were subjected to observation at two locations on the cheek and mouth, a total of 108 locations, and the spectral reflection spectrum of each of them was measured by a spectrocolorimeter. In Figure 7,
The color distributions in the CIE1976L ^* a ^* b ^* color space of these observation points are shown. This color distribution corresponds to H = 2YR to 8YR, V = 5 to 7 and C = 2 to 5 in the Munsell color system.

Principal component analysis was performed based on this colorimetric data. FIG. 8 shows the cumulative contribution rate of each principal component. From the figure, it can be seen that the spectral reflectance spectrum of the skin can be estimated with a cumulative contribution rate of 99.5% up to the third principal component. In addition,
FIG. 9 shows the spectrum of each principal component obtained by this principal component analysis.
Shown in.

Next, each observation target portion was photographed with a high-definition still image camera (HC-1500 manufactured by Nikon Corporation), and its output values R, G, and B were used to determine the tristimulus value X, in the same manner as in Example 1.
Y and Z were obtained, and the spectral reflection spectrum of the observation target portion was estimated based on the above-mentioned principal component analysis result. The result is shown in FIG. For reference, FIG. 10 also shows the spectral reflection spectrum measured by a spectrocolorimeter. From this, it can be seen that the spectral reflection spectrum estimated by the principal component analysis almost completely matches the actually measured spectrum.

[0056]

According to the present invention, it is possible to obtain the tristimulus values X, Y, Z of the skin to be observed by using a color video camera, and thus the skin can be accurately measured in color. Becomes In addition, the obtained tristimulus values X, Y, Z
It is possible to accurately estimate the spectral reflection spectrum based on the.

[Brief description of drawings]

FIG. 1 is a schematic diagram of the method of the present invention.

FIG. 2 is a relationship diagram of incident luminance and output values R, G, B in a color video camera.

FIG. 3 is a relationship diagram between incident luminance and corrected output values R, G, and B of incident luminance sensitivity characteristics in a color video camera.

FIG. 4 is a distribution diagram of color differences (ΔE ^* ab) between tristimulus values X, Y, and Z obtained from output values of a color video camera and X, Y, and Z obtained by a spectrocolorimeter in Examples. Is.

FIG. 5 is a distribution diagram of color difference (ΔE ^* ab) between tristimulus values X, Y, and Z obtained from output values of a color video camera and X, Y, and Z obtained by a spectrocolorimeter in a comparative example. Is.

FIG. 6 is a distribution diagram of color differences (ΔE ^* ab) between tristimulus values X, Y, and Z obtained from output values of a color video camera and X, Y, and Z obtained by a spectrocolorimeter in a comparative example. Is.

FIG. 7 is a color distribution diagram of an observation target portion in the example.

FIG. 8 is a relationship diagram between the main component of the spectral reflection spectrum and its cumulative contribution rate in the example.

FIG. 9 is a spectrum of each main component of the spectral reflectance spectrum of skin.

FIG. 10 is a spectral reflection spectrum estimated in the example and a spectral reflection spectrum measured by a spectrocolorimeter.

Claims (5)

[Claims] 1. Output values R, G for red (R), green (G), and blue (B) obtained by photographing skin with a color video camera.
In the method of obtaining B, and measuring the color of the skin based on the output values R, G, B of the camera, the output value R,
G and B are obtained, on the other hand, the tristimulus values X, Y and Z are obtained by measuring the color of the observed object with a spectrocolorimeter, and the output values R and
G, B and tristimulus values X, Y, Z obtained by the spectrocolorimeter
The transformation matrix of and is determined by multiple regression analysis as a matrix using at least the quadratic terms of the output values R, G, B of the camera, and using this transformation matrix,
A colorimetric method for skin, characterized by converting output values R, G, B of a camera for arbitrary skin into tristimulus values X, Y, Z. 2. Output values R, G, G of the camera so that the output values R, G, B of the camera have a linear relationship with the incident luminance.
2. The skin colorimetric method according to claim 1, wherein B is corrected and a conversion matrix of the corrected camera output values R, G, B and tristimulus values X, Y, Z obtained by a spectrocolorimeter is obtained. . 3. The skin colorimetric method according to claim 1, wherein output values R, G and B based on the internal reflection light of the skin are obtained when the skin is photographed by a color video camera. 4. The color video camera output values R, G, B are converted into tristimulus values X, Y, Z by the method according to claim 1, and principal component analysis is performed from the obtained tristimulus values X, Y, Z. A method for estimating a spectral reflectance spectrum by estimating the spectral reflectance spectrum of the skin. 5. The spectral reflection spectrum estimation method according to claim 4, wherein the spectral reflection spectrum of the skin is represented by a linear sum of the average color of the skin and the first to third principal components when the principal component analysis is performed. .