Estimation method of physiological pigment concentration of skin

Abstract

To estimate a physiological concentration of each pigment in a skin from an internally reflected light image of the skin without being affected by shadow, and to accurately form a simulation image of the skin with a prescribed pigment concentration.SOLUTION: A method for estimating a physiological pigment concentration in a skin from a pixel value of an internally reflected light image of the skin, includes: acquiring a relation between a non-shadow vector and the physiological pigment concentration by pre-setting a constraint condition for the pigment concentration; acquiring a pixel value of the internally reflected light image of the skin; acquiring a non-shadow vector obtained by excluding a shadow component from a color space vector of the pixel value; and estimating the physiological pigment concentration in the skin or an amount of change thereof from the non-shadow vector of the skin on the basis of the constraint condition and the relation.SELECTED DRAWING: Figure 2

Description

[Technical Field]

[0001] The present invention relates to a method for estimating physiological pigment concentration in skin using an internally reflected light image of the skin, and a method for estimating a non-shadow vector of internally reflected light in skin from the physiological pigment concentration in the skin to form a simulated image of the skin. [Background technology]

[0002] Generally, as shown in Figure 1, for the pixel value (R, G, B) of point P in the internal reflection light image of the skin, if the color space vector p(r, g, b) in Cartesian coordinates is defined as (-log(R), -log(G), -log(B)), then by selecting the basis vectors appropriately, the color space basis vectors (pigment component vectors) in oblique coordinates, that is, the melanin vector m, hemoglobin vector h, and shading vector l, and their coefficients (M, H, L), can be given by the following equation 1, and it is known that the melanin distribution can be obtained from the coefficient M, and the hemoglobin distribution can be obtained from the coefficient H.

[0003]

number

[0004] In the method for forming a pigment concentration image of skin described in Patent Document 1, the color space vector of the orthogonal coordinates of the internally reflected light image of the skin is expressed as a linear sum of pigment component vectors including a shading vector that forms oblique coordinates, and the pigment component vectors and shading vectors are sequentially optimized to suppress the influence of shading and each other's pigment concentration in the pigment concentration image.

[0005] This method can obtain a dye component image that suppresses the influence of shading and each other's dye concentration, allowing the relative level of dye concentration to be determined with high accuracy. However, because the obtained dye concentration is based on an image, it was not clear how it specifically corresponds to the actual dye concentration (hereinafter also referred to as physiological dye concentration) with units such as weight %.

[0006] Furthermore, it is difficult to determine the oxygenated hemoglobin concentration and the deoxygenated hemoglobin concentration separately using the method described in Patent Document 1. In fact, the document does not describe any specific methods for determining these concentrations separately.

[0007] On the other hand, Monte Carlo simulation can determine the relationship between pigment concentration and reflectance spectrum, and by taking into account the spectral characteristics of the camera and light source, the relationship between reflectance spectrum and pixel values ​​(R, G, B) can be determined. Using this method, physiological pigment concentration can be determined from the pixel values ​​(R, G, B) of the internal reflection light image of the skin (Non-Patent Document 1).

[0008] However, because this method is based on the premise that the illuminance is constant, when trying to determine the pigment concentration over a wide area such as the entire face, the pigment concentration cannot be accurately determined because the illuminance is not constant due to shadows caused by the shape. For example, if the illuminance is not constant due to shadows caused by the unevenness of the face, such as the nose or eye sockets, it is impossible to eliminate the effect of this on the pigment concentration. [Prior art documents] [Patent documents]

[0009] [Patent Document 1] Japanese Patent Publication No. 2020-202983 [Non-patent literature]

[0010] [Non-Patent Document 1] Nishidate I., et al. (2011) Noninvasive imaging of human skin hemodynamics using a digital red-green-blue camera. Journal of Biomedical Optics,16(8),086012. Summary of the Invention [Problem to be solved by the invention]

[0011] In contrast to the above-mentioned prior art, the object of the present invention is to enable estimation of the physiological concentration of individual pigments in skin, etc. from an internal reflection light image of skin, etc., without being affected by shadows (i.e., even if the illumination at each pixel is not uniform), and to accurately form a pigment concentration image of skin, etc., having a predetermined pigment concentration.

[0012] Here, skin, etc. refers to any biological tissue that contains multiple pigments and can be photographed, and includes not only the epidermis but also mucous membranes such as the lips and the mucous membrane inside the mouth. Hereinafter, "skin, etc." will also be referred to simply as "skin." [Means for solving the problem]

[0013] The inventors of the present invention have conceived the following: by obtaining a non-shadow vector, which removes the shadow component, from the color space vector of the pixel values ​​of an internally reflected light image of skin, and by setting constraints on the pigment concentration to obtain the relationship between the non-shadow vector and the physiological pigment concentration, it is possible to estimate two or more pigment concentrations that comply with the constraints using two variables, which remove the shadow information from the three RGB variables of the pixel values; therefore, it is possible to distinguish and estimate, for example, the oxygenated hemoglobin concentration and the deoxygenated hemoglobin concentration; and by using this relationship to estimate the non-shadow vector of the skin from the physiological pigment concentration of the skin, it is possible to form a simulation image when the pigment concentrations of oxygenated hemoglobin and deoxygenated hemoglobin are calculated based on the values ​​of the non-shadow vector, thereby completing the present invention.

[0014] That is, the present invention provides a method for estimating physiological pigment concentration of biological tissue including skin and mucous membrane (hereinafter simply referred to as skin) from pixel values ​​of an internal reflection light image of the skin, the method comprising: A constraint condition is set in advance for the dye concentration to obtain the relationship between the non-shadow vector and the dye concentration, Obtain pixel values ​​of the internal reflection light image of the skin; A non-shade vector is obtained by removing the shade component from the color space vector of the pixel value; Based on the constraints and the relationship, there is provided a method for estimating the physiological pigment concentration of the skin, or the amount of change thereof, using the non-shading vector of the skin.

[0015] The present invention also provides a method for generating a simulated image of skin based on physiological pigment concentrations of the skin, comprising: A constraint condition is set in advance for the dye concentration to obtain the relationship between the non-shadow vector and the physiological dye concentration. obtaining a physiological pigment concentration of the skin; estimating a non-shading vector of an internally reflected light image of the skin from the physiological pigment concentration of the skin based on the constraint and the relationship; A method for forming a simulated image of skin is provided, which forms a simulated image of skin based on the values ​​of the non-shadow vectors.

[0016] The present invention also provides a system for implementing the dye concentration estimation method, comprising: A system for estimating a physiological pigment concentration in skin from pixel values ​​of an internally reflected light image of the skin, the system comprising: A function for storing the relationship between the non-shadow vector and the dye concentration obtained by setting constraints on the dye concentration; a function of calculating a non-shade vector by excluding a shade component from a color space vector of a pixel value of an internally reflected light image of the skin when the pixel value is input; A function of calculating the physiological pigment concentration or the amount of change thereof in the skin using the non-shadow vector of the skin based on the constraint conditions and the relationship. The present invention provides a physiological dye concentration estimation system having:

[0017] The present invention also provides a system for implementing the simulation image forming method, which includes: A system for generating a simulated image of skin from physiological pigment concentrations of the skin, the system comprising a computing device and a display, the computing device comprising: A function for storing the relationship between the non-shadow vector and the dye concentration obtained by setting constraints on the dye concentration; a function for calculating a non-shadow vector of an internally reflected light image of the skin from a physiological pigment concentration of the skin when the physiological pigment concentration of the skin is input; A function that calculates a simulated image of the skin based on the calculated non-shadow vector value and outputs it to the display. The present invention provides a system for forming a simulated skin image, comprising: [Effects of the Invention]

[0018] According to the method for estimating physiological pigment concentration of skin of the present invention, a constraint condition is set in advance for the pigment concentration to obtain the relationship between the non-shading vector and the physiological pigment concentration, and based on this relationship, the physiological pigment concentration of any skin can be estimated from the pixel value of the internal reflection light image of that skin.

[0019] Therefore, according to the method for estimating physiological pigment concentration of skin of the present invention, it is possible to distinguish between the oxygenated hemoglobin concentration and the deoxygenated hemoglobin concentration contained in the skin and estimate them separately. Therefore, when reactive hyperemia occurs in the lips, it is possible to observe that the area where reactive hyperemia occurs becomes darker and the deoxygenated hemoglobin concentration decreases in the deoxygenated hemoglobin concentration change image obtained from images taken before and after the occurrence of reactive hyperemia.

[0020] Furthermore, the physiological pigment concentration estimation system of the present invention can implement the physiological pigment concentration estimation method of the present invention.

[0021] On the other hand, according to the method for forming a simulation image of skin of the present invention, it is possible to accurately form a simulation image of skin when the skin has a predetermined pigment concentration based on the relationship between the non-shadow vector and the physiological pigment concentration.

[0022] Furthermore, the skin simulation image forming system of the present invention can implement the skin simulation image forming method of the present invention. Using this system, for example, it is possible to generate a simulation image of lips with different lip colors, in which the physiological pigment density of the lips has been changed, or a simulation image in which the lips with the changed pigment density have been replaced with the original lips. This simulation image makes it possible to visually present to the user and a third party the change in the impression of the lips and face that accompanies a change in lip color based on the change in lip pigment density. [Brief explanation of the drawings]

[0023] [Figure 1] Figure 1 shows the color space vector of point P with pixel value (R, G, B) expressed by the color space vector (r, g, b) in orthogonal coordinates, and the melanin vector m, hemoglobin vector h, and shading vector l in oblique coordinates. [Figure 2] FIG. 2 is a diagram showing the uv plane perpendicular to the shading vector l in the color space shown in FIG. [Figure 3] FIG. 3 shows the relationship between the melanin concentration Cm, the total hemoglobin concentration Cth, and the oxygen saturation StO2 on the UV surface. [Figure 4] FIG. 4 shows the calculation results when the melanin concentration Cm=5 in FIG. 3 is extracted and the oxygenated hemoglobin concentration Coh and the deoxygenated hemoglobin concentration Cdh are set in a grid pattern. [Figure 5A] FIG. 5A is a frequency distribution diagram of oxygen saturation StO2 in the face in a field survey. [Figure 5B] Figure 5B is a frequency distribution diagram of oxygen saturation StO2 in the lips in the actual survey. [Figure 6A] FIG. 6A is a diagram showing the relationship between the melanin concentration Cm and the color space component r when a Monte Carlo simulation is performed while changing the melanin concentration Cm with the oxygenated hemoglobin concentration Coh=0.3%. [Figure 6B]FIG. 6B is a diagram showing the relationship between the melanin concentration Cm and the color space component g when a Monte Carlo simulation is performed while changing the melanin concentration Cm with the oxygenated hemoglobin concentration Coh=0.3%. [Figure 6C] FIG. 6C is a diagram showing the relationship between the melanin concentration Cm and the color space component b when a Monte Carlo simulation is performed while changing the melanin concentration Cm with the oxygenated hemoglobin concentration Coh=0.3%. [Figure 7A] FIG. 7A is a diagram showing the relationship between the oxygenated hemoglobin concentration Coh and the color space component r when a Monte Carlo simulation is performed while changing the oxygenated hemoglobin concentration Coh with the melanin concentration Cm=5%. [Figure 7B] FIG. 7B is a diagram showing the relationship between the oxygenated hemoglobin concentration Coh and the color space component g when a Monte Carlo simulation is performed while changing the oxygenated hemoglobin concentration Coh with the melanin concentration Cm=5%. [Figure 7C] FIG. 7C is a diagram showing the relationship between the oxygenated hemoglobin concentration Coh and the color space component b when a Monte Carlo simulation is performed while changing the oxygenated hemoglobin concentration Coh with the melanin concentration Cm=5%. [Figure 8A] FIG. 8A is a graph showing the relationship between the coefficient M of the melanin vector m and the melanin concentration Cm when the oxygenated hemoglobin concentration Coh is 0.3% and the deoxygenated hemoglobin concentration Cdh is 0%. [Figure 8B] FIG. 8B is a relationship diagram between the coefficient H of the oxygenated hemoglobin vector h and the oxygenated hemoglobin concentration Coh when the melanin concentration Cm=5% and the deoxygenated hemoglobin concentration Cdh=0%. [Figure 9A] Figure 9A shows the relationship between H' and H when the non-shadow component is (M', H') and the non-shadow component is (M, H) when the melanin concentration is Cm = 5% and the oxygen saturation is StO2 = 50% (i.e., Cdh = Coh) for the lips, assuming (M', H') and (M, H) when Cdh = 0%. [Figure 9B]Figure 9B shows the relationship between H' and M-M' when the non-shadow component is (M', H') and the non-shadow component is (M, H) when Cdh = 0% (i.e., Cm = 5%) and StO2 = 50% (i.e., Cdh = Coh) for the lips, assuming a melanin concentration of 5% and oxygen saturation of 50% (i.e., Cdh = Coh). [Figure 10] Figure 10 shows an internal reflection image of the lip where reactive hyperemia is occurring (a) and an image of the change in deoxyhemoglobin in the area where reactive hyperemia is occurring (b) before and after the onset of reactive hyperemia. [Figure 11A] FIG. 11A is a graph showing the relationship between the physiological melanin concentration of skin determined from the spectrum of a spectrophotometer and the melanin concentration estimated from an image using the method of the present invention. [Figure 11B] FIG. 11B is a graph showing the relationship between the physiological oxygenated hemoglobin concentration of the skin obtained from the spectrum of a spectrophotometer and the oxygenated hemoglobin concentration estimated from an image by the method of the present invention. [Figure 12A] Figure 12A is a diagram showing the relationship between the physiological melanin concentration of skin obtained from the spectrum of a spectrophotometer and the melanin concentration (M) estimated from an image using the method described in Patent Document 1 (however, the derivation of the pigment vector follows Example 1). [Figure 12B] Figure 12B is a diagram showing the relationship between the physiological oxygenated hemoglobin concentration of the skin obtained from the spectrum of a spectrophotometer and the oxygenated hemoglobin concentration (H) estimated from an image using the method described in Patent Document 1 (however, the derivation of the pigment vector follows Example 1). DETAILED DESCRIPTION OF THE INVENTION

[0024] The present invention will now be described in detail with reference to the drawings, in which the same reference numerals represent the same or equivalent elements.

[0025] [Outline of the method for estimating skin pigment concentration] In the method for estimating pigment concentration of the present invention, a constraint condition is set on the pigment concentration to obtain a relationship between the non-shading vector and the pigment concentration, and the physiological pigment concentration of the skin or its change amount is estimated from the non-shading vector of the pixel value of the skin using the relationship. By performing this operation for each pixel of the image, they can be visualized.

[0026] In the method for estimating pigment concentration of the present invention, the color space vector (r, g, b) of the pixel value (R, G, B) is expressed as a linear sum of the melanin vector m, hemoglobin vector h, and shading vector l in oblique coordinates as shown in the above-mentioned Equation 1. Based on this, the non-shading vector is considered to be the linear sum of the melanin vector m and the hemoglobin vector h, i.e., M m + H h.

[0027] The melanin vector m and the hemoglobin vector h can be determined by, for example, independent component analysis, and can also be determined as in specific example 1 described below.

[0028] Here, independent component analysis is a method of modeling the layer structure of the skin as a laminated structure consisting of an epidermal layer containing melanin as the main pigment component, a dermal layer containing hemoglobin as the main pigment component, and subcutaneous tissue containing other pigment components, and assuming that the distribution of melanin and the distribution of hemoglobin are independent, and that pigment component signals are emitted independently from each layer based on the Lambert-Beer law, and that these signals are mixed together to form an image signal, and then separating and extracting the pigment component signals of each layer from the image signal ("Image Analysis of Skin Color Using Independent Component Analysis and Its Application to Analysis of Age Spots", Journal of the Japanese Society of Cosmetic Technologists 41(3) pp. 159-166 (2007)).

[0029] In the method for estimating dye concentration of the present invention, (1) Obtain a non-shade vector from the color space vector of the pixel value; (2) By imposing constraints on the pigment concentration, we obtain the relationship between the non-shadow vector of the skin and the physiological pigment concentration; (3) On the other hand, pixel values ​​of an internal reflection image of the skin of the subject or the like are acquired, and a non-shade vector is acquired by removing the shading component from the color space vector of the pixel values. (4) Based on the constraints and relationships in (2), the physiological pigment concentration or its change amount is estimated using the non-shading vector of the skin in (3). The above items (1) to (4) will be explained below.

[0030] (1) Obtaining non-shading vectors In (1), a non-shade vector is obtained from the color space vector of the pixel value (R, G, B).

[0031] Therefore, first, the pixel value (R, G, B) is converted into a color space vector (r, g, b) using the following formula. Color space vector (r,g,b)=(-log(R),-log(G),-log(B))

[0032] The non-shade vector can be obtained from the color space vector (r, g, b) by separating the color space vector into a non-shade vector and a shade vector. In this separation, the non-shade vector does not contain any shade information. However, in cases where this separation is performed using a uv plane (described later), it is not essential that the shade vector does not contain any dye density information.

[0033] As a method for obtaining a non-shade vector that does not contain shading information from a color space vector (r, g, b), for example, the following (i) or (ii) can be performed. (i) After calculating the color space vector (r, g, b) = (-log(R), -log(G), -log(B)), calculate the coefficient M of the melanin vector m, the coefficient H of the hemoglobin vector h, and the coefficient L of the shading vector l in the oblique coordinates. Then, using only the coefficient M of the melanin vector and the coefficient H of the hemoglobin vector, the non-shading vector = M m + H h (a linear combination of only the pigment vectors).

[0034] (ii) Find the color space vector (r, g, b) = (-log(R), -log(G), -log(B)), then use the uv plane perpendicular to the shading vector l, as shown in Figure 2.

[0035] The uv plane can be calculated as the xy coordinates when the shading vector l is aligned with the z axis in an xyz Cartesian coordinate system, for example, using Rodrigues' rotation formula.

[0036] On the uv plane, the magnitude of the shading vector in the l direction is zero, so shading information is removed, and the position vector of any point on the uv plane is a non-shading vector.

[0037] In this specification, including FIG. 2, the following symbols are used to denote physiological dye concentrations. Cm: melanin concentration Coh: Oxygenated hemoglobin concentration Cdh: deoxygenated hemoglobin concentration Cth: Total hemoglobin concentration, Cth=Coh+Cdh StO2: Oxygen saturation in hemoglobin, StO2=Coh / Cth

[0038] (2) Obtaining the relationship between the non-shadow vector and physiological dye concentration In this invention, we set constraints on pigment concentration to obtain the relationship between the skin's non-shading vector and physiological pigment concentration. To obtain pixel values ​​(R, G, B) from pigment concentration, we first calculate the reflectance spectrum of a specified chromaticity concentration using Monte Carlo simulation, and then convert the reflectance spectrum into pixel values ​​using the camera's spectral sensitivity and the illumination spectrum. The method for obtaining non-shading vectors from pixel values ​​is as described in (1).

[0039] Here, Monte Carlo simulation is a method of setting up a calculated skin model, irradiating light onto the skin model and causing scattering and absorption to occur stochastically, tracing the trajectory multiple times, and probabilistically determining the reflectance from the results.

[0040] The spectral sensitivity of the camera and the spectrum of the illumination are determined in advance for the camera and illumination that are expected to be used.

[0041] In the present invention, the relationship between the non-shadow vector and the physiological pigment concentration is obtained by imposing constraints on the pigment concentration. Below, we will explain how to obtain this relationship when the non-shadow vector is obtained using the UV plane.

[0042] (2-1)UV surface A specific method for associating pigment concentration with the uv surface can be as follows, for example, when the melanin concentration Cm, total hemoglobin concentration Cth, and oxygen saturation in hemoglobin StO2 take the values ​​(%) in Table 1.

[0043] [Table 1]

[0044] In Table 1, Cm is the volume percentage of the concentration when the entire epidermis is made up of simple melanosomes, which is taken as 100%. Coh and Cdh are the volume percentages of the concentration when the entire dermis is made up of blood with a hematocrit of 45% and all hemoglobin is oxygenated or deoxygenated, which is taken as 100%, and Cth is their sum. Note that the units for Cm, Coh, Cdh, and Cth will be omitted below.

[0045] On the other hand, assuming a Canfield VISIA facial imaging device (camera: Canon EOS 5D Mark II) as the imaging device, the spectral sensitivity for each RGB of the device's camera and the illumination spectrum are determined in advance. The reflectance spectrum is determined by Monte Carlo simulation using a computing device, and converted to RGB using the previously determined spectral sensitivity of the camera and illumination spectrum. This is then converted to a color space vector (r, g, b) = (-log(R), -log(G), -log(B)), and then converted to UV coordinates, resulting in Figure 3. The effects of uneven lighting have been eliminated on the UV plane.

[0046] Specifically, the Monte Carlo simulation was performed under the following conditions: A two-layer skin model was used, with a 4.94 mm thick dermis layer underneath a 0.06 mm thick epidermis layer. The anisotropic coefficient in scattering was set to g = 0.93, and the scattering coefficient was (2 × 10 5 ×λ -1.5 +2×10 12 ×λ -4 ) / (1-g)[cm -1 ]. We assumed that melanin with a concentration of Cm was uniformly distributed in the epidermis, and oxygenated hemoglobin and deoxygenated hemoglobin with concentrations of Coh and Cdh were uniformly distributed in the dermis. All Monte Carlo simulations used below were performed in this manner.

[0047] Figure 3 shows the direction of increase in melanin concentration Cm and total hemoglobin concentration Cth, i.e., the direction of color change on the uv plane. It also shows that the total hemoglobin concentration Cth when oxygen saturation StO2 = 100% changes linearly on the uv plane within the actual range. Furthermore, the greater the distance d from the melanin vector, the higher the oxygenated hemoglobin concentration Coh. When oxygen saturation is close to 100%, both the changes in melanin concentration and total hemoglobin concentration are linear, so successful separation can be expected using conventional methods such as those described in Patent Document 1 and related publications.

[0048] Figure 4 shows the results of a Monte Carlo simulation in which the melanin concentration Cm=5 in Figure 3 was extracted and the oxygenated hemoglobin concentration Coh and the deoxygenated hemoglobin concentration Cdh were set in a grid pattern. Figure 4 reveals that the direction of increase in deoxygenated hemoglobin and the direction of increase in melanin are roughly opposite. The directions of increase in oxygenated hemoglobin, deoxygenated hemoglobin, and melanin are shown in the figure.

[0049] This tendency of change in pigment concentration appears regardless of the part of the skin as long as the skin model used in the Monte Carlo simulation is valid.

[0050] Furthermore, the direction in which the melanin concentration Cm increases on the uv plane and the direction in which the oxygenated hemoglobin concentration Coh increases also coincide with the direction in which the melanin vector m and hemoglobin vector h obtained by the method described in Patent Document 1 are projected onto the uv plane. In this respect, the method of the present invention is consistent with the method described in Patent Document 1 when oxygen saturation is high, demonstrating the validity of the method of the present invention.

[0051] The method for obtaining the relationship between the non-shading vector and physiological pigment concentration is not limited to Monte Carlo simulation, but may involve measuring the light reflection on simulated skin with controlled scattering, absorption, etc. However, Monte Carlo simulation is preferable due to its ease of use and high reproducibility.

[0052] (2-2) Constraints The constraint is an assumption about the pigment concentration in order to estimate the physiological concentration or its change of the pigment contained in the skin from the pixel values ​​of the internal reflection light image of the skin. It is preferable to make a practically reasonable assumption about any of the oxygen saturation (StO2), deoxygenated hemoglobin concentration (Cdh), oxygenated hemoglobin concentration (Coh), and melanin concentration (Cm).

[0053] For example, in the phenomenon of reactive hyperemia, when a cotton swab is pressed against the lips and then removed, blood flow increases compared to before the cotton swab was pressed against the lips. Since it can be assumed that the melanin concentration Cm does not change over such a short period of time, it is possible to set a constraint that the melanin concentration Cm remains constant, which makes it possible to estimate the amount of change in the deoxygenated hemoglobin concentration Cdh before and after reactive hyperemia.

[0054] Furthermore, as a practically reasonable assumption, a condition that the oxygen saturation level StO2 or the deoxygenated hemoglobin concentration Cdh is a constant may be set, so that once the melanin concentration Cm and the oxygenated hemoglobin concentration Coh are determined, the deoxygenated hemoglobin concentration Cdh is also determined automatically.

[0055] (2-3) Substantial and reasonable constraints As a specific example, first, in a field survey, the spectral spectra of the forehead, cheeks, and sides of the mouth of 154 people aged 20 to 50 were obtained using a spectrophotometer (Konica Minolta Japan, CM-2600d). In a separate field survey, the spectral spectra of the upper and lower lips of 45 people aged 20 to 30 were similarly obtained. These were then solved using Monte Carlo simulation as an inverse problem (optimization problem) of the problem of calculating spectral spectra from pigment concentrations, and the pigment concentrations of each part of each subject were determined. The previously described method (Monte Carlo simulation) was used to associate pigment concentrations with spectra.

[0056] The resulting oxygen saturation (StO2) was then used to calculate the frequency distribution for each skin region. The results are shown in Figures 5A and 5B. Figure 5A is a histogram of the facial data, which combines all data from the forehead, cheeks, and sides of the mouth, and Figure 5B is a histogram of the lip data, which combines data from the upper and lower lips.

[0057] From Figures 5A and 5B, it can be seen that although the oxygen saturation StO2 values ​​differ between the face and lips, the oxygen saturation StO2 for facial skin is close to 100%, so it is acceptable to approximate the oxygen saturation StO2 as 100% (i.e., StO2 = 1, Coh = Cth, Cdh = 0).

[0058] Similarly, since the lips are concentrated in the range of StO2 = 0.5 to 0.6, it can be seen that calculations can be performed assuming StO2 = 0.5 or 0.6.

[0059] (2-4) Obtaining the relationship between non-shadow vectors and physiological dye concentrations or their changes under constraint conditions The relationship between the non-shading vector and these physiological pigment concentrations under constraint conditions can be obtained by creating a lookup table of the melanin concentration Cm, the oxygenated hemoglobin concentration Coh, or the deoxygenated hemoglobin concentration Cdh relative to the coefficients of the melanin vector and the hemoglobin vector that make up the non-shading vector, or the uv coordinate. It can also be obtained by creating an approximate equation between the coefficients of these pigment vectors and the pigment concentrations. In this case, the pigments whose concentrations are to be kept constant are sequentially fixed under the constraint condition (2). While it is possible to perform sequential Monte Carlo simulations to estimate the pigment concentrations, it is preferable to use a lookup table or approximate equation to ensure calculation speed.

[0060] Furthermore, in creating a lookup table or approximation formula, in order to eliminate the influence of shading, as described above in (1)(i), a color space vector (r, g, b) of orthogonal coordinates (r, g, b) = (-log(R), -log(G), -log(B)) is calculated from the pixel values ​​(R, G, B), and then only M and H of the coefficients M, H, L of the oblique coordinates with m, h, and l as the basic vectors are used, or, as described in (1)(ii), the dye vector is considered on the uv plane.

[0061] (3) Obtaining the non-shadow vector of the internal reflection light image of the subject's skin The non-shadow vector can be obtained in the same manner as the non-shadow vector in (1).

[0062] (4) Estimating the pigment density of the skin or forming a simulated image of the skin After obtaining the relationship between the non-shading vector and the physiological pigment concentration, the physiological pigment concentration is estimated from the RGB pixel values ​​of the internally reflected light of the subject's skin based on that relationship. When imaging the estimated pigment concentration, color may be added to make it easier to visualize the color of the pigment.

[0063] Any type of camera may be used to measure the internally reflected light from the subject's skin, including a consumer mirrorless camera, a single-lens reflex camera, or an industrial camera, or a camera that is sensitive to a specific wavelength.

[0064] For example, an internal reflection image can be obtained by attaching polarizing plates to the front of the lighting fixture and the front of the camera and orthogonally ...

[0065] Alternatively, to simulate the appearance when a pigment concentration is given or when the pigment concentration changes, the non-shading vector when the pigment concentration is set can be calculated based on the relationship between the non-shading vector and physiological pigment concentration described above, and then pixel values ​​can be calculated from the non-shading vector to create a simulation image. The simulation image can also be based on pigment concentrations estimated from internal reflection images, or the set value of the pigment concentration can be calculated by multiplying the estimated pigment concentration by a constant. Note that the non-shading vector used here is preferably a linear combination of only the pigment vectors, M m + H h, because this minimizes the influence of the pigment on the shading vector.

[0066] To improve visibility, a simulation of only the skin region can be performed in combination with a conventional method for extracting the skin region from the image. The simulation image created in this way is simulated based on the actual pigment concentration, allowing for accurate simulation of the appearance corresponding to the set rate and amount of change in pigment concentration. Furthermore, although the image obtained here excludes surface reflection, a glossy, natural-looking image can be created by adding a previously obtained surface reflection image. Surface reflection images can be obtained using, for example, a polarizing plate [N. Tsumura, et al., "Image-based skin color and texture analysis / synthesis by extracting hemoglobin and melanin information in the skin," ACM Trans. Graph. 22, 770 (2003)].

[0067] (Physiological pigment density estimation system and skin simulation image generation system) The estimation system for implementing the above-described method for estimating physiological pigment concentration of skin of the present invention includes: A function for storing the relationship between the non-shadow vector and the dye concentration obtained by setting constraints on the dye concentration; A function of calculating a non-shade vector by excluding a shade component from a color space vector of a pixel value of an internally reflected light image of the skin when the pixel value is input; and A function of calculating the physiological pigment concentration or the amount of change thereof in the skin using the non-shadow vector of the skin based on the constraint conditions and the relationship. The apparatus includes a computing device having:

[0068] Also, a system for forming a simulated image of skin by providing a physiological pigment concentration of the skin includes a computing device and a display, the computing device comprising: A function for storing the relationship between the non-shadow vector and the dye concentration obtained by setting constraints on the dye concentration; a function for calculating a non-shadow vector of an internally reflected light image of the skin from a physiological pigment concentration of the skin when the physiological pigment concentration of the skin is input; It has the function of calculating a simulated image of the skin based on the calculated non-shadow vector value and outputting it to a display.

[0069] In both the pigment concentration estimation system and the skin simulation image formation system, the calculation device preferably has the function of calculating the non-shading vector when the skin contains pigment by simulating it using a Monte Carlo simulation or the like, and obtaining the relationship between the non-shading vector and the physiological pigment concentration.

[0070] A specific example of such a computing device is a personal computer equipped with image analysis software such as ImageJ and numerical calculation software such as MATLAB (registered trademark), which can separate pigment component vectors and shading vectors. Also preferred is a computer equipped with a look-up table function. There are no particular limitations on the display used in the system of the present invention. [Example]

[0071] The present invention will be described in detail below with reference to specific examples. [Example 1 (obtaining non-shadow vectors)] For example, let Cdh = 0 (StO2 = 100%) be the constraint. In this case, the non-shading vector should take into account the melanin concentration and oxygenated hemoglobin concentration. If the linear combination of h and m in (Equation 1) (M m + H h) is used as the non-shading vector, m and h can be calculated as follows:

[0072] First, consider the changes in vectors r, g, and b when the melanin concentration and oxygenated hemoglobin concentration change near their average values. Using Coh = 0.3 as the average oxygenated hemoglobin concentration, we used the Monte Carlo simulation described above to find vectors r = -logR, g = -logG, and b = -logB when the oxygenated hemoglobin concentration Coh = 0.3 and the melanin concentration Cm = 1, ..., 10, and obtain the relationships shown in Figures 6A, 6B, and 6C.

[0073] The relationship between these figures can be approximated by a straight line, and the melanin vector can be calculated from the slope of the line. That is, when the slopes of the lines in Figures 6A, 6B, and 6C are a, b, and c, the melanin vector d = (a 2 +b 2 +c 2 ) and the melanin vector (a / d, b / d, c / d) can be obtained. This melanin vector m has a length of 1 and the r:g:b ratio is a:b:c. The r:g:b ratio actually calculated from Figures 6A, 6B, and 6C is 0.090:0.12:0.16, and as a result, the melanin vector m is (0.40, 0.54, 0.74).

[0074] Similarly, using Cm = 5 as the average melanin concentration, the Monte Carlo simulation was used to calculate vector r = -logR, vector g = -logG, and vector b = -logB for oxygenated hemoglobin concentrations Coh = 0.05, 0.1, ..., 1.5, yielding the relationships shown in Figures 7A, 7B, and 7C. From these figures, the hemoglobin vector h is (0.35, 0.69, 0.64). Because the hemoglobin vector calculated in this way corresponds to the oxygenated hemoglobin concentration, it can be called the oxygenated hemoglobin vector.

[0075] Using the obtained non-shading vectors m and h, the relationship between the pigment concentration and the non-shading vector is obtained as shown in [Specific Example 2], and then this relationship is used to obtain the melanin concentration Cm and the oxygenated hemoglobin concentration Coh.

[0076] Note that M calculated using (Equation 1) is actually strongly related to the melanin concentration, and H is also strongly related to the oxygenated hemoglobin concentration; as these concentrations increase, the fluctuations in the coefficient M or H relative to concentration fluctuations become smaller. Therefore, the melanin concentration Cm can be calculated by correcting M, and the oxygenated hemoglobin concentration Coh can be calculated by correcting H.

[0077] [Example 2 (obtaining the relationship between non-shadow vectors and dye concentrations, and estimating dye concentrations based on the relationship)] Following the above-mentioned [Specific Example 1], assuming that the deoxygenated hemoglobin concentration Cdh = 0, the oxygenated hemoglobin concentration Coh is constant at 0.3, and the melanin concentration Cm is changed from 0 to 1, ..., 10, the relationship between the pigment component vector coefficient M and the melanin concentration Cm is calculated, and Figure 8A is obtained.

[0078] The plot of FIG. 8A can be approximated by the following quadratic equation: Cm=1.6M 2 +2.3M-0.57

[0079] This quadratic equation can incorporate the tendency that as the melanin concentration Cm increases, the variation of the coefficient M relative to the variation of the concentration Cm decreases.

[0080] Therefore, by calculating the coefficient MHL of the pigment component vector in oblique coordinates from the pixel values ​​RGB of the internally reflected light of the skin of a subject, etc., the pigment concentration Cm can be estimated from the coefficient M according to the relationship in Figure 8A. In other words, the physiological melanin concentration can be estimated. Furthermore, by calculating for each pixel, a melanin concentration image of the skin can be formed from the melanin concentration Cm.

[0081] Similarly, assuming that the deoxygenated hemoglobin concentration Cdh = 0, the melanin concentration Cm = 5, and the oxygenated hemoglobin concentration Coh are varied from 0.05 to 0.1, ..., 2, the relationship between the pigment component vector coefficient H and the oxygenated hemoglobin concentration Coh is calculated, and Figure 8B is obtained.

[0082] The plot of FIG. 8B can be approximated by the following quadratic equation: Coh=1.8H 2 +0.46H+0.093

[0083] This quadratic equation can capture the tendency that as the oxygenated hemoglobin concentration Coh increases, the variation of the coefficient H relative to the variation of the concentration Coh decreases.

[0084] Therefore, by calculating the coefficient MHL of the pigment component vector of the oblique coordinates from the pixel values ​​RGB of the internally reflected light of the skin of the subject, the pigment concentration Coh can be estimated from the coefficient H according to the relationship in Figure 8B. That is, the physiological oxygenated hemoglobin concentration Coh of the skin can be estimated. Furthermore, by performing calculations for each pixel, an oxygenated hemoglobin concentration image of the skin can be formed from the oxygenated hemoglobin concentration Coh.

[0085] On the other hand, from the relationship in Fig. 8A, the coefficient M can be determined when the melanin concentration Cm is given, and from the relationship in Fig. 8B, the coefficient H of the oxygenated hemoglobin vector can be determined when the oxygenated hemoglobin concentration Coh is given. Therefore, the non-shading vector, which is the linear sum of the physiological melanin concentration Cm and oxygenated hemoglobin concentration Coh of the skin, can be determined for each pixel, and a simulation image of the skin can be formed by combining this with the original shading vector.

[0086] In the above method for estimating the melanin concentration Cm, the formula for M for Cm is approximately not affected by the value of H, but to further improve accuracy, the coefficients of the quadratic formula may be expressed as a function of H. Similarly, to further improve accuracy, the coefficients of the quadratic formula for H for Coh may be expressed as a function of M.

[0087] [Example 3 (correction)] When obtaining the relationship between the non-shading vector and the pigment concentration under the constraint that the deoxygenated hemoglobin concentration Cdh is constant or the oxygen saturation StO2 is constant, the effect of the deoxygenated hemoglobin concentration Cdh on the pigment component vector coefficients M and H can be corrected using a function, and then the melanin concentration Cm can be approximated by a function of the pigment component vector coefficient M, and the oxygenated hemoglobin concentration Coh can be approximated by a function of the pigment component vector coefficient H.

[0088] For example, for the lips, the melanin concentration Cm = 5 and oxygen saturation StO2 = 50% are set as constraints, and the oxygenated hemoglobin concentration Coh is varied from 0.2, 0.4, ..., 5.0. The coefficients (M, H) of the non-shading components of the pigment component vector when Cdh = 0 and the coefficients (M', H') of the non-shading vector when Cdh = Coh are calculated using the Monte Carlo simulation described above, and H is plotted against this H', resulting in Figure 9A. This shows that H changes as Cdh changes from Cdh = Coh to Cdh = 0.

[0089] The curve in FIG. 9A can be approximated by the following quadratic equation: H=0.17H' 2 +0.87H'-0.041

[0090] Therefore, when the coefficient H' of the oxygenated hemoglobin vector is calculated from the pixel values ​​RGB of the orthogonal coordinates of the internally reflected light in the lip area of ​​the subject, the coefficient H when Cdh = 0 can be obtained from the coefficient H' when Cdh = Coh using the approximation formula of the curve in Figure 9A. Coh can be estimated from H using the relational formula between the coefficient H when Cdh = 0 and the oxygenated hemoglobin concentration Coh calculated in Specific Example 2. Furthermore, since Cdh = Coh is assumed, Cdh can also be calculated automatically.

[0091] Similarly, plotting (M-M') against H' gives Figure 9B, which shows that M changes as Cdh changes from Cdh=Coh to Cdh=0. The curve in FIG. 9B can be approximated by the following quadratic equation: M=M'+0.16H'2 +0.22H'-0.044

[0092] Therefore, if the coefficients M' and H' of the melanin vector and oxygenated hemoglobin vector are calculated from the pixel values ​​RGB of the orthogonal coordinates of the internally reflected light from the skin of the subject, etc., the coefficient M can be obtained from the coefficients M' and H' using the approximation formula for the curve in Figure 9B, and the pigment concentration Cm can be estimated from M using the relational formula between the coefficient M and the melanin concentration Cm when Cdh = 0 calculated in Specific Example 2.

[0093] On the other hand, when the oxygenated hemoglobin concentration Coh and the melanin concentration Cm are given, M and H can be calculated for each pixel when Cdh = 0, and M' and H' can also be calculated when Cdh = Coh, so a simulated image of the skin under the condition of StO2 = 50% can also be created.

[0094] [Example 4 (obtaining the relationship between non-shadow vectors and dye concentrations using a lookup table)] The relationship between the non-shading vector and the dye density can also be obtained using a lookup table (LUT). For example, an LUT can be created as follows. The following explanation is for the case where the constraint Cdh = 0 is set, but it can also be handled in the same way when other constraints are set.

[0095] Cm is assumed to take values ​​from 0 to 10 in increments of 0.1, and Coh is assumed to take values ​​from 0 to 5 in increments of 0.02. For each Cm, Coh pair, the reflectance spectrum is calculated using the Monte Carlo simulation described above, and RGB is calculated taking into account the spectral characteristics of the camera and light source. From this, the pair of melanin vector coefficients and oxygenated hemoglobin vector coefficients or uv coordinates are calculated.

[0096] When the uv coordinates are obtained, an LUT is obtained in which Cm and Coh set to predetermined values ​​in a grid-like lookup table are used as input values ​​and the uv values ​​are used as output values.

[0097] On the other hand, by solving the inverse problem (optimization problem) of finding Cm and Coh from uv, it is possible to obtain an LUT that uses uv values ​​set in a grid pattern as input values ​​and Cm and Coh as output values.

[0098] The optimization problem can be solved, for example, using MATLAB's built-in function fminsearch. In fminsearch, the sum of squares of the error (u0 - u) when u and v are obtained when certain values ​​are assumed for Cm and Coh for the values ​​of u0 and v0 for which Cm and Coh are to be predicted. 2 +(v0- v) 2 By setting the convergence condition so that is minimized, it is possible to find Cm and Coh when this condition is satisfied, that is, Cm and Coh when u and v are almost equal to u0 and v0.

[0099] Once the LUT is calculated, Cm and Coh can be calculated immediately for any pair of uv and uv values ​​using general interpolation methods such as linear interpolation within the range. Specifically, the LUT was created by taking u0 from -0.8 to -0.2 in increments of 0.0025 and v0 from -0.2 to 0.1 in increments of 0.00125.

[0100] The same can be done when using a pair of coefficients of the melanin vector and the oxygenated hemoglobin vector.

[0101] [Comparison of specific effects of the present invention with conventional examples] (Image of reactive lip hyperemia) Figure 10(a) shows an internal reflection image of lips showing reactive hyperemia caused by pressing and releasing a cotton swab against the lips, captured using a VISIA facial imaging device. Figure 10(b) shows an image of the change in deoxyhemoglobin (DHE) concentration before and after reactive hyperemia, as measured by the method of the present invention. More specifically, the image in Figure 10(b) was created by converting the pixel values ​​RGB of the internal reflection image in (a) into color space vectors (r, g, b). The coefficient M of the melanin vector before reactive hyperemia was defined as M1, and the coefficient M of the melanin vector after reactive hyperemia was defined as M2. The change M2-M1 was then correlated with the change in deoxyhemoglobin concentration. Because the time between reactive hyperemia and reactive hyperemia was extremely short, the actual melanin concentration barely changed. Therefore, the change M2-M1 can be attributed to a change in deoxyhemoglobin concentration, not a change in melanin concentration.

[0102] In image (b), the area surrounded by the dashed line is dark, indicating that the deoxyhemoglobin concentration Cdh has decreased.

[0103] On the other hand, in a similarly created image (not shown) of the change in oxygenated hemoglobin, the same areas become brighter. Therefore, it is clear that the present invention can distinguish between the change in deoxygenated hemoglobin concentration and the change in oxygenated hemoglobin concentration.

[0104] Furthermore, according to the present invention, it is also possible to obtain the decrease in deoxyhemoglobin concentration as a physiological value from the image (b).

[0105] In contrast, the method described in Patent Document 1 can obtain a value H related to hemoglobin concentration, but cannot link it to a specific concentration, and in particular cannot estimate the deoxygenated hemoglobin concentration. In addition, it cannot obtain information about the oxygen saturation of hemoglobin.

[0106] Furthermore, according to the method described in Non-Patent Document 1, the image of the change in deoxyhemoglobin is affected by shading, and therefore the image of the change in deoxyhemoglobin of the lip reactive congestion is affected by the three-dimensional shape of the lips, including fine wrinkles, and therefore cannot capture subtle changes such as reactive congestion.

[0107] (Relationship between physiological pigment concentration and pigment concentration estimated from images) In a survey, facial images of 154 people aged 20 to 50 were taken, and (r, g, b) was calculated from the pixel values ​​RGB of the cheek area of ​​the images, and then MHL was calculated.

[0108] Meanwhile, M and H obtained by Monte Carlo simulation were corrected using the procedure described in the above-mentioned specific example 3, and the relationship between the coefficient M and melanin concentration and the relationship between the coefficient H and oxygenated hemoglobin concentration were obtained using the corrected M and H. Then, using these relationships, the melanin concentration and oxygenated hemoglobin concentration were estimated by the method of the present invention from M and H obtained from the pixel values ​​of each of the 154 people in the actual survey.

[0109] In this case, the reflectance spectrum of the same area where the melanin concentration and oxygenated hemoglobin concentration were estimated was obtained using a spectrophotometer (CM-2600d, manufactured by Konica Minolta Japan Inc.), and the melanin concentration and oxygenated hemoglobin concentration in the skin were estimated from the reflectance spectrum using the Monte Carlo simulation described above, and these were used as the physiological melanin concentration and physiological oxygenated hemoglobin concentration.

[0110] The relationship between physiological melanin concentration and the estimated melanin concentration from the images obtained by the method of the present invention is shown in FIG. 11A, and the relationship between physiological oxyhemoglobin concentration and the estimated oxyhemoglobin concentration from the images is shown in FIG. 11B.

[0111] As a comparative example, melanin concentration and oxygenated hemoglobin concentration on the cheek were estimated from the RGB pixel values ​​of the skin obtained in the above-mentioned field survey using the method described in Patent Document 1 (however, the derivation of the pigment vector follows Example 1). In this case, since most of the hemoglobin concentration can be considered to be oxygenated hemoglobin as shown in FIG. 5B, the hemoglobin concentration was considered to be oxygenated hemoglobin. The relationship between the physiological melanin concentration and the melanin concentration estimated using the method described in Patent Document 1 (however, the derivation of the pigment vector follows Example 1) is shown in FIG. 12A, and the relationship between the physiological oxygenated hemoglobin concentration and the oxygenated hemoglobin concentration estimated using the method described in Patent Document 1 (however, the derivation of the pigment vector follows Example 1) is shown in FIG. 12B.

[0112] According to the method of the present invention, as shown in Fig. 11A, the relationship between the estimated melanin concentration calculated from the image and the physiological melanin concentration can be approximated by a straight line passing through the origin, and this approximated line and the plot generally coincide. Furthermore, as shown in Fig. 11B, the relationship between the estimated oxygenated hemoglobin concentration calculated from the image and the physiological oxygenated hemoglobin concentration can also be approximated by a straight line passing through the origin, and this approximated line and the plot generally coincide. Therefore, it can be said that the physiological pigment concentration and the pigment concentration estimated from the image correspond accurately.

[0113] In contrast, according to the method described in Patent Document 1, when the relationship between the melanin image value M calculated from an image and the physiological melanin concentration was approximated by a straight line passing through the origin, as shown in FIG. 12A, the approximated line and the plot generally coincided. On the other hand, as shown in FIG. 12B, the relationship between the oxygenated hemoglobin image value H calculated from an image and the physiological oxygenated hemoglobin concentration did not converge to the origin. This means that even if the physiological oxygenated hemoglobin concentration is zero, the estimated oxygenated hemoglobin concentration calculated from the image does not become zero, and it is estimated from the image that a certain amount of oxygenated hemoglobin is present. Thus, conventional methods cannot accurately link the physiological pigment concentration and the pigment concentration value calculated from the image.

Claims

1. A method for estimating physiological pigment concentration of biological tissue including skin and mucous membrane (hereinafter simply referred to as skin) from pixel values ​​of an internal reflection light image of the skin, comprising: A constraint condition is set in advance for the dye concentration to obtain the relationship between the non-shadow vector and the dye concentration, Obtain pixel values ​​of the internal reflection light image of the skin; A non-shade vector is obtained by removing the shade component from the color space vector of the pixel value; A method for estimating physiological pigment concentration of skin, which estimates the physiological pigment concentration or a change in the concentration of the skin using a non-shading vector of the skin based on the constraint condition and the relationship.

2. 2. The method for estimating skin pigment concentration according to claim 1, wherein the relationship between the non-shading vector and the physiological pigment concentration is obtained by simulating the non-shading vector when the skin contains pigment at a predetermined concentration.

3. 3. The method for estimating skin pigment concentration according to claim 1, wherein components of a plane perpendicular to the shading vector in the color space vector are defined as non-shading vectors in order to eliminate the influence of the shading vector.

4. 3. The method for estimating skin pigment density according to claim 1, wherein the relationship between the non-shadow vector and the physiological pigment density is obtained by using a look-up table.

5. 3. The method for estimating skin pigment concentration according to claim 1, wherein the relationship between the non-shadow vector and the physiological pigment concentration is obtained using these approximate expressions.

6. 3. The method for estimating skin pigment concentration according to claim 1, wherein oxygenated hemoglobin and deoxygenated hemoglobin are distinguished in obtaining the relationship between the non-shadow vector and the physiological pigment concentration.

7. 7. The method for estimating skin pigment concentration according to claim 6, wherein the constraint is that the oxygenated hemoglobin concentration and the deoxygenated hemoglobin concentration must be a predetermined ratio.

8. 3. The method for estimating skin pigment concentration according to claim 1, wherein the constraint condition is that the melanin concentration is constant, and the amount of change in pigment concentration is estimated as the amount of change in oxygenated hemoglobin concentration and the amount of change in deoxygenated hemoglobin concentration.

9. 1. A method for generating a simulated image of skin based on physiological pigment concentrations of the skin, comprising: A constraint condition is set in advance for the dye concentration to obtain the relationship between the non-shadow vector and the physiological dye concentration. obtaining a physiological pigment concentration of the skin; estimating a non-shading vector of an internally reflected light image of the skin from the physiological pigment concentration of the skin based on the constraint and the relationship; A method for forming a simulated image of skin, which forms a simulated image of skin based on the values ​​of the non-shading vectors.

10. 10. The method for forming a simulated image of skin according to claim 9, wherein the relationship between the non-shadow vector and the physiological pigment concentration is obtained by simulating the non-shadow vector when the skin contains a pigment at a predetermined concentration.

11. 11. The method for forming a simulated skin image according to claim 9, wherein in the simulation of the non-shade vectors, a linear combination of only pigment vectors is simulated as the non-shade vectors to eliminate the influence of the shading vectors.

12. 11. The method for forming a simulation image of skin according to claim 9, wherein the non-shade vector is a linear combination of pigment vectors that lie in a plane perpendicular to the shading vector in the color space vector.

13. 11. The method for generating a simulation image of skin according to claim 9 or 10, wherein the relationship between the non-shadow vector and the physiological pigment concentration is obtained by using a look-up table.

14. 11. The method for forming a simulation image of skin according to claim 9 or 10, wherein the relationship between the non-shadow vector and the physiological pigment concentration is obtained using these approximate expressions.

15. 11. The method for forming a simulation image of skin according to claim 9 or 10, wherein the relationship between the non-shadow vector and the physiological pigment concentration is obtained by distinguishing between oxygenated hemoglobin and deoxygenated hemoglobin.

16. A system for estimating a physiological pigment concentration in skin from pixel values ​​of an internally reflected light image of the skin, the system comprising: A function for storing the relationship between the non-shadow vector and the dye concentration obtained by setting constraints on the dye concentration; a function of calculating a non-shade vector by excluding a shade component from a color space vector of a pixel value of an internally reflected light image of the skin when the pixel value is input; A function of calculating the physiological pigment concentration or the amount of change thereof in the skin using the non-shadow vector of the skin based on the constraint conditions and the relationship. A physiological dye concentration estimation system with

17. The physiological pigment concentration estimation system of claim 16, wherein the computing device has a function of calculating the relationship between the non-shadow vector and the physiological pigment concentration by simulating the non-shadow vector when the skin contains a predetermined concentration of pigment.

18. A system for generating a simulated image of skin from physiological pigment concentrations of the skin, the system comprising a computing device and a display, the computing device comprising: A function for storing the relationship between the non-shadow vector and the dye concentration obtained by setting constraints on the dye concentration; a function for calculating a non-shadow vector of an internally reflected light image of the skin from a physiological pigment concentration of the skin when the physiological pigment concentration of the skin is input; A function that calculates a simulated image of the skin based on the calculated non-shadow vector value and outputs it to the display. A system for forming a simulation image of skin having the above structure.

19. 19. The system for forming a simulation image of skin according to claim 18, wherein the computing device has a function of calculating the relationship between the non-shadow vector and the physiological pigment concentration by simulating the non-shadow vector when the skin contains pigment at a predetermined concentration.

Images