JP7412767B2 - Skin color unevenness evaluation device, skin color unevenness evaluation method, and skin color unevenness evaluation program - Google Patents

Description

The present disclosure relates to a skin color unevenness evaluation technique for evaluating skin color unevenness.

Uneven skin tone affects the appearance of one's age and can also be an important indicator when evaluating the effectiveness of cosmetics. Therefore, a technique for analyzing skin color unevenness has been proposed (for example, Patent Document 1).

Japanese Patent Application Publication No. 2011-240086

However, the above-mentioned techniques require the use of dedicated measuring equipment and measurement in a predetermined measurement environment, making analysis and evaluation of skin color unevenness complicated. Furthermore, the above technique has a problem in that the correlation between the skin color unevenness evaluation based on the sensory evaluation and the skin color unevenness evaluation using the measurement results is low, and the evaluation accuracy of the skin color unevenness is low.

Therefore, there is a need for a skin color unevenness evaluation that does not depend on the measurement environment, can use general-purpose equipment, and can improve the accuracy of skin color unevenness evaluation.

In order to solve at least part of the above problems, the present disclosure adopts the following various aspects.

A first aspect provides a skin color unevenness evaluation device. The skin color unevenness evaluation device according to the first aspect includes a skin image acquisition unit that acquires a skin image of an evaluation subject, and a skin image acquisition unit that performs spatial frequency analysis on the acquired skin image, and calculates an intensity value for each frequency. A skin image analysis unit that acquires percentiles, a color unevenness evaluation formula prepared for each percentile value using the intensity value of each frequency as a variable, and a predetermined percentile among the intensity values for each frequency acquired above. and a determination unit that determines a color unevenness evaluation value using the intensity value of the value.

According to the color unevenness evaluation device according to the first aspect, the intensity value of each frequency is used as a variable, and a color unevenness evaluation formula prepared for each percentile value and a predetermined intensity value for each acquired frequency are used. Since the color unevenness evaluation value is determined using the intensity value of the determined percentile value, it is possible to use general-purpose equipment without depending on the measurement environment, improving the accuracy of skin color unevenness evaluation. be able to.

In the color unevenness evaluation device according to the first aspect, the color unevenness evaluation formula includes a color unevenness quantitative evaluation value based on sensory evaluation of the sample skin image, and an intensity value at each frequency obtained by spatial frequency analysis of the sample skin image. , is derived for each percentile value of the intensity value. In this case, it is possible to obtain a color unevenness evaluation formula according to the percentile value of the intensity value.

In the color unevenness evaluation device according to the first aspect, the predetermined percentile value is a percentile value equal to or higher than the 90th percentile value. In this case, it is possible to accurately evaluate color unevenness by reflecting skin characteristics.

In the color unevenness evaluation device according to the first aspect, the skin image analysis section separates the acquired skin image into images of each RGB component and performs spatial frequency analysis. In this case, it becomes possible to obtain intensity values for each of the RGB components, and the accuracy of the color unevenness evaluation value can be further improved.

In the color unevenness evaluation device according to the first aspect, each of the frequencies is a frequency in a range of 1 to 39 Hz. In this case, highly accurate color unevenness evaluation results can be provided while reducing the calculation load for color unevenness evaluation.

In the color unevenness evaluation device according to the first aspect, the skin image acquisition unit acquires a captured skin image from a general-purpose imaging device. In this case, it is possible to easily obtain an evaluation of skin color unevenness.

A second aspect provides a method for evaluating skin color unevenness. The evaluation method for color unevenness according to the second aspect includes acquiring a skin image of a person to be evaluated, performing spatial frequency analysis on the acquired skin image, acquiring a percentile of the intensity value for each frequency, and Using the intensity value of the frequency as a variable, the color unevenness evaluation formula prepared for each percentile value and the intensity value of the predetermined percentile value among the intensity values for each frequency obtained above are used to calculate color unevenness. determining an evaluation value.

According to the color unevenness evaluation method according to the second aspect, the intensity value of each frequency is used as a variable, and a color unevenness evaluation formula prepared for each percentile value and a predetermined intensity value for each frequency obtained are used. Since the color unevenness evaluation value is determined using the intensity value of the determined percentile value, it is possible to use general-purpose equipment without depending on the measurement environment, improving the accuracy of skin color unevenness evaluation. be able to.

A third aspect provides a skin color unevenness evaluation program. The color unevenness evaluation program according to the third aspect has a function of acquiring a skin image of an evaluation subject, and a function of performing spatial frequency analysis on the acquired skin image and acquiring a percentile of intensity values for each frequency. Then, using the intensity value of each frequency as a variable, the color unevenness evaluation formula prepared for each percentile value, and the intensity value of a predetermined percentile value among the intensity values for each frequency obtained above. , and a function of determining a color unevenness evaluation value are realized by a computer.

According to the color unevenness evaluation program according to the third aspect, the intensity value of each frequency is used as a variable, and the color unevenness evaluation formula prepared for each percentile value and the intensity value for each acquired frequency are determined in advance. Since the color unevenness evaluation value is determined using the intensity value of the percentile value obtained, it is possible to use general-purpose equipment without depending on the measurement environment, improving the accuracy of skin color unevenness evaluation. be able to.

The color unevenness evaluation program according to the third aspect may be stored in a computer readable medium such as a CD, DVD, Blu-ray disc, or semiconductor memory.

1 is an explanatory diagram showing an example of a skin color unevenness evaluation device and a skin color unevenness evaluation system according to a first embodiment; FIG. 1 is an explanatory diagram showing an example of a skin color unevenness evaluation device and a skin color unevenness evaluation system according to a first embodiment; FIG. FIG. 1 is an explanatory diagram showing the skin color unevenness evaluation system according to the first embodiment by means of functional units. FIG. 1 is a block diagram showing the functional internal configuration of the skin color unevenness evaluation device according to the first embodiment. 5 is a flowchart showing each processing step of a skin color unevenness evaluation process executed by the skin color unevenness evaluation apparatus according to the first embodiment. FIG. 3 is an explanatory diagram showing an example of spatial frequency analysis on acquired skin image data. FIG. 7 is an explanatory diagram showing an example of a procedure for obtaining percentiles for component strength of each frequency. FIG. 3 is an explanatory diagram showing the correlation between the component intensity and percentile at one frequency of one color component. An explanatory diagram showing an example of quantitative evaluation. A graph showing part of the verification results between frequency and correlation function. A graph showing the correlation between the calculated evaluation value and the sensory evaluation value at the 50th percentile value. A graph showing the coefficient of determination R ² between the calculated evaluation value and the sensory evaluation value at percentile values from 10 to 100. 13 is a table showing the correspondence between percentile values and correlation coefficients R in FIG. 12. An explanatory diagram showing skin image samples used to verify the correlation between percentile values and human perception. An explanatory diagram showing skin image samples used to verify the correlation between percentile values and human perception. FIG. 16 is an explanatory diagram showing calculated evaluation values using the skin color unevenness evaluation formula of the 50th percentile value for the skin image samples shown in FIGS. 14 and 15. FIG. FIG. 16 is an explanatory diagram showing calculated evaluation values using the skin color unevenness evaluation formula of the 99th percentile value for the skin image samples shown in FIGS. 14 and 15. FIG. FIG. 16 is an explanatory diagram showing the results of sensory evaluation for each skin image sample shown in FIGS. 14 and 15. FIG. 7 is a graph showing the correlation between the color unevenness evaluation value when a dedicated device is used and the color unevenness evaluation value when a smartphone is used in the second embodiment. FIG. 20 is an explanatory diagram showing a part of the skin image sample used for verifying the correlation in FIG. 19;

In response to the problem that there is still a discrepancy between the evaluation of skin color unevenness using conventional measuring instruments and the skin color unevenness that humans perceive, the present inventors have developed a method to improve skin tone unevenness that humans perceive. focused on the relationship between perception and texture, which are characteristic of the skin, and carefully considered how this would be reflected in the evaluation of skin tone unevenness using measuring instruments. As a result, they discovered that by focusing on the intensity of specific frequency components, skin tone unevenness can be evaluated objectively and with high precision. We have found that the accuracy of evaluating unevenness can be improved. DESCRIPTION OF THE PREFERRED EMBODIMENTS Hereinafter, a skin color unevenness evaluation device, a skin color unevenness evaluation method, and a skin color unevenness evaluation program according to the present disclosure will be described based on embodiments with reference to the drawings.

First embodiment:
1 and 2 show an example of a skin color unevenness evaluation device and a skin color unevenness evaluation system according to a first embodiment. The skin color unevenness evaluation device 100 may be realized as a tablet terminal or a smartphone (hereinafter simply referred to as a "smartphone") as shown in FIG. 1, or may be realized as a desktop type as shown in FIG. It may also be implemented as a notebook computer. When the smartphone 100 functions as a skin tone unevenness evaluation device, it includes an imaging section 20 and a display section 21, and constitutes a skin tone unevenness evaluation system 10 by itself. In the smartphone 100, the imaging unit 20 captures a skin image to be evaluated, and the display unit 21 displays the skin color unevenness evaluation result. On the other hand, when the computer 100 functions as a skin tone unevenness evaluation device, the imaging device 30 and the display device 31 are separately used, and the computer terminal 100, the imaging device 30, and the display device 31 are used to create the skin color unevenness evaluation system 10. is configured. The imaging device 30 captures a skin image to be evaluated. As the imaging device 30, a VISIA (manufactured by Canfield Scientific), a general digital camera, or a smartphone, which is known as a skin diagnostic device and is capable of imaging in a certain imaging environment, may be used. The display device 31 displays the skin color unevenness evaluation result, for example, a numerical evaluation value or a skin age generated using the evaluation value. The computer 100 and the imaging device 30 are connected by wire or wirelessly, and imaging data can be transmitted from the imaging device 30 to the computer 100. Furthermore, since a tablet terminal or a smartphone can be used as a display device, the skin color unevenness evaluation result may be transmitted from the computer 100 to the tablet terminal or smartphone by wireless communication. In this case, the skin color unevenness evaluation results can be sent to the user's smartphone. The wireless connection is realized, for example, by wireless communication based on a short-range wireless communication standard such as Wifi (registered trademark) or Bluetooth (registered trademark). Note that the computer 100 may be a server computer that functions as a server connected to a plurality of terminals via a wired or wireless network, and in this case, skin images to be evaluated are received from a plurality of various client terminals. The data can be uploaded, a skin color unevenness evaluation value can be calculated in the server computer, and the calculated skin color unevenness evaluation value can be transmitted from the server computer to various client terminals.

FIG. 3 is an explanatory diagram showing the skin color unevenness evaluation system 10 according to the first embodiment in terms of functional parts. The skin color unevenness evaluation device 100 includes a skin image acquisition section F1, a skin image analysis section F2, and a color unevenness evaluation value determination section F3. The skin image acquisition section F1 acquires skin image data from the imaging section 20 or the imaging device 30. The skin image analysis unit F2 includes a spatial frequency analysis unit F21 that performs spatial frequency analysis on the image data acquired by the skin image acquisition unit F1, and a percentile that acquires a percentile of intensity values for each frequency using the frequency analysis results. The acquisition unit F22 is also provided. The color unevenness evaluation value determination unit F3 determines the intensity of a predetermined percentile value among the intensity values for each frequency acquired by the skin image analysis unit F2, that is, the percentile acquisition unit F22, and the intensity value prepared for each percentile value. The color unevenness evaluation value is determined using a color unevenness evaluation formula in which the intensity value of each frequency is used as a variable. The color unevenness evaluation value determined by the color unevenness evaluation value determining section F3 is output to the display section 21 or the display device 31.

FIG. 4 is a block diagram showing the functional internal configuration of the skin color unevenness evaluation device 100 according to the first embodiment. The skin tone unevenness evaluation device 100 includes a central processing unit (CPU) 101, a memory 102, an input/output interface 103, and a clock generator (not shown). The CPU 101, memory 102, input/output interface 103, and clock generator are connected via an internal bus 104 so that they can communicate in both directions. The memory 102 includes a memory that non-volatilely and read-only stores a skin tone unevenness evaluation program Pr1 for evaluating skin tone unevenness, such as a ROM, and a memory that can be read and written by the CPU 101, such as a RAM. . The read/write area of the memory 102 includes an evaluation image storage area 102a for storing image data captured for evaluation.

The CPU 101, that is, the color unevenness evaluation device 100, develops the skin color unevenness evaluation program Pr1 stored in the memory 102 into a readable/writable memory and executes it, thereby executing the skin image analysis unit F2 and color unevenness evaluation value determination. It functions as section F3. Note that the CPU 101 may be a single CPU, may be multiple CPUs that execute each program, or may be a multitasking type or multithreading type CPU that can simultaneously execute multiple programs. Also good. Note that the skin image analysis section F2 and the color unevenness evaluation value determination section F3 may be realized in hardware as a logic circuit.

The input/output interface 103 functions as the skin image acquisition unit F1, and is a physical and logical interface used to connect the uneven skin tone evaluation device 100 with various other devices, such as a USB terminal. , a LAN terminal, or a serial bus terminal. In the first embodiment, a display unit 21 having, for example, a liquid crystal panel, a display device 31, an imaging unit 20 having an image sensor, and an imaging device 30 are connected to the input/output interface 103 via signal lines, respectively. . Note that, instead of or together with the display device 31, a printing device may be provided that outputs, that is, prints the color unevenness evaluation value on a printing medium, for example, paper.

The skin color unevenness evaluation process executed by the skin color unevenness evaluation device 100 will be described. FIG. 5 is a flowchart showing each processing step of the skin color unevenness evaluation process executed by the skin color unevenness evaluation apparatus 100 according to the first embodiment. More specifically, the CPU 101 executes the skin tone unevenness evaluation program Pr1.

The CPU 101 acquires skin image data to be evaluated via the input/output interface 103 (step S100). The acquired skin image data is, for example, 950 x 950 pixels (approximately 3.5 cm x 3.5 cm) captured by the imaging device 30 under diffused illumination, which is the optimal environmental light that can suppress reflections and shadows on the skin surface. This is RGB data of a skin image with a size of 5 cm). The skin image data is acquired by imaging the evaluation subject's cheeks and arms using the imaging device 30, for example. Uneven skin tone is caused by the unevenness of the skin surface tone due to locally excessive melanin pigments and hemoglobin pigments, which determine skin color. In particular, it is known that redness components have a large effect on the evaluation of uneven skin tone. Cheeks are known as a region of human skin that is prone to uneven skin tone such as age spots, freckles, and dullness. In addition, skin color unevenness may include various types of skin color that can be identified by color values and pigment components such as melanin components and hemoglobin components, such as acne, acne scars, pimples, burn scars, and moles.

The CPU 101 analyzes the acquired skin image data (step S102). The specific processing procedure will be explained with reference to FIGS. 6 to 8. As shown in FIG. 6, the CPU 101 first separates the RGB skin image data into R component, G component, and B component color component image data, calculates the average pixel value for each component image data, and calculates the average pixel value of each component image data. The difference between each pixel value constituting the data and the average pixel value is calculated, and spatial frequency processing is performed to obtain a power spectrum image showing the component strength of the spatial frequency. The spatial frequency processing is performed by a two-dimensional discrete Fourier transform, and in this embodiment, it is performed by a two-dimensional discrete Fourier transform using a seven-term Blackman-Harris window as a window function. As is known, the power spectrum image defines the component intensity using the frequency r and the angle θ as indices. As shown in FIG. 7, the CPU 101 performs polar coordinate transformation on the obtained power spectrum image to obtain a polar coordinate transformed image in which the column direction indicates frequency (Hz) and the row direction indicates angle (rad). The CPU 101 further calculates the percentile of the component intensity values in the polar coordinate transformed image for each frequency, that is, sorts the component intensity values in order of intensity to obtain a percentile image in which the row direction indicates the percentile value (0 to 100). Note that instead of the polar coordinate transformed image and the percentile image, a table in which the component intensity is determined by angle and frequency, or a table in which the component intensity is determined by percentile value and frequency may be obtained. As a result, the strength of one component is determined by the frequency and the percentile value. For example, the percentile of the component intensity of one frequency of one color component has the characteristics shown in FIG. In FIG. 8, the vertical axis shows the logarithmically transformed intensity, and the horizontal axis shows the percentile.

The CPU 101 determines a color unevenness evaluation value using the following color unevenness evaluation formula prepared in advance (step S104). Specifically, the CPU 101 calculates R, G, Using each component intensity value of B, a color unevenness evaluation value is determined using the following color unevenness evaluation formula prepared in advance. The specific percentile value is preferably at least the 90th percentile, more preferably at least the 95th percentile, and most preferably at the 99th percentile. The color unevenness evaluation value is expressed by a value between -1.000 and +1.000, and as it goes towards +1.000, there is no color unevenness, that is, a high evaluation, and as it goes towards -1.000, it indicates a high evaluation. There is color unevenness, that is, it shows a low evaluation.
Color unevenness evaluation value = (a1 x Ir1) +... + (a39 x Ir39) + (g1 x Ig1) +... + (g39 x Ig39) + (b1 x Ib1) +... + (b39 x Ib39) + C (constant)
Here, a1 to a39, g1 to g39, and b1 to b39 are coefficients at specific percentile values predetermined in the procedure described later, and Ir1 to Ir39, Ig1 to Ig39, and Ib1 to Ib39 are the coefficients of the images obtained by imaging. Each component intensity value at a specific percentile value calculated for

The CPU 101 outputs the color unevenness evaluation value calculated or determined using the color unevenness evaluation formula to the display unit 21 or the display device 31 (step S106), and stores this processing routine.

Procedure for determining color unevenness evaluation formula First, 30 randomly selected skin images were prepared as sample skin images, and a sensory test was conducted by 38 evaluators. Specifically, for all pairs of skin images, i.e., (30×29)/2=435 pairs, one point is assigned to the skin image (one) with inconspicuous uneven skin tone. Evaluation was performed by assigning -1 point to the noticeable skin image (the other side). Using the obtained evaluation results, the color unevenness of the skin image was quantitatively evaluated by Scheffe's paired comparison method (Nakaya's modified method) to obtain a color unevenness quantitative evaluation value based on sensory evaluation, that is, a sensory evaluation value. FIG. 9 is an explanatory diagram showing an example of quantitative evaluation. In FIG. 9, the numbers on the upper side of the horizontal axis indicate the numbers of the skin image samples, and the numbers on the lower side of the horizontal axis indicate the color unevenness quantitative evaluation values. Note that as the Scheffe paired comparison method, methods other than Nakaya's modified method, such as Ura's modified method, may be used, or as the paired comparison method, for example, Thurston's paired comparison method may be used. Also good.

The RGB skin image data of the sample skin image used for the quantitative evaluation is separated into the R component, G component, and B component as described above, and the average pixel value is calculated for each component image data. The difference between each pixel value constituting the component image data and the average pixel value was calculated, and spatial frequency processing was performed to obtain a power spectrum image showing the component strength of the spatial frequency. The obtained power spectrum image was subjected to polar coordinate transformation, and further, the component intensities were sorted in order of intensity for each frequency from 1 to 512 Hz, and a table of component intensities with percentile values and frequencies as indexes was obtained.

Next, the color unevenness quantitative evaluation value (-1.000 to +1.000) obtained by the sensory evaluation, that is, the sensory evaluation value, is used as the objective variable, and the component intensity of each frequency at each percentile value is used as the explanatory variable. Regression coefficients and constants were determined by regression (partial least squares regression), and a color unevenness evaluation formula for each percentile value was derived. Note that the color unevenness evaluation formula may be determined by another regression method that approximates a straight line. As described above, the color unevenness evaluation formula obtained as a result is a first-order polynomial including the component intensity and constant of coefficient times frequency for each RGB component. The obtained color unevenness evaluation formula for each percentile value is stored in the memory 102 in a nonvolatile and read-only manner. Note that, as described below, the memory 102 may store only the color unevenness evaluation formula corresponding to a specific percentile value.

Verification of the obtained color unevenness evaluation formula The upper limit of the frequency to be used was verified. Generally, skin, that is, a skin image, has almost no high frequency components, and for example, a frequency of 40 Hz is at a level corresponding to fluctuations in pores. Furthermore, it is known that the effects of shadows and wrinkles become noticeable in the frequency range of approximately 1 to 30 Hz. Furthermore, when there are many target frequencies, the calculation load increases. Therefore, among the frequencies from 1 to 512 Hz, frequencies that are a significant upper limit from the accuracy of the calculated evaluation value, that is, the correlation with the sensory evaluation value, were verified as the color unevenness evaluation formula. FIG. 10 shows a part of the graph showing the verification results, ie, the correlation coefficient R for the frequency range of 20 to 60 Hz. As is clear from FIG. 10, the increase in the correlation becomes noticeably slower after 35 Hz, where the correlation coefficient R = 0.927, and reaches its first peak at 39 Hz, where the correlation coefficient R = 0.9284. , even after that, the correlation coefficient R fluctuates between 0.927 and 0.929. Therefore, even if the component intensity corresponding to a frequency of 35 Hz or higher, or even 39 Hz or higher, is used, the correlation coefficient does not change significantly, and the sensory It is understood that there is a high correlation with the evaluation value, with a correlation coefficient R=0.927 or higher. In addition, by using a frequency of 35 Hz or less or 39 Hz or less, the calculation load can be reduced and calculation evaluation values can be obtained quickly, compared to using frequencies up to 512 Hz, or even up to 100 Hz. , equivalent and good correlations can be obtained. Based on this result, in step S104 of FIG. 5, a color unevenness evaluation formula with an upper limit of 39 Hz is used.

Next, we verified the percentile values used. That is, the correlation between the calculated evaluation value and the sensory evaluation value was verified for predetermined percentile values. For example, FIG. 11 is a graph showing the correlation between the calculated evaluation value and the sensory evaluation value at the 50th percentile value. The correlation coefficient between the calculated evaluation value and the sensory evaluation value at the 50th percentile value obtained using a known calculation formula was R=0.9108. Similarly, for the 10th, 20th, 30th, 40th, 60th, 70th, 80th, 90th, 95th, 99th, and 100th percentile values, the correlation coefficient between the sensory evaluation value and the calculated evaluation value in the range of 10 to 100 percentile values. R was evaluated. FIG. 12 is a graph showing the coefficient of determination R ² between the calculated evaluation value and the sensory evaluation value in the 10th to 100th percentile values. FIG. 13 is a table showing the correspondence between percentile values and correlation coefficients R in FIG. 12. As shown in FIG. 13, the correlation coefficient R takes a value of 0.9188 at the 10th percentile value, a minimum value of 0.9082 at the 40th percentile value, and 0.9108 at the 50th percentile value or higher. Further, a correlation coefficient R of 0.9214 is taken for the 90th percentile value, a correlation coefficient R of 0.9283 is taken for the 99th percentile value, and a correlation coefficient R of 0.9282 is taken for the 100th percentile value. Therefore, it can be said that the 90th percentile or higher percentile value has a high correlation with the sensory evaluation value, and in particular, the 99th percentile value has the highest correlation. On the other hand, for percentile values below the 40th percentile value, a correlation coefficient R of about 0.90 is taken, but as shown below, it does not match the sensory evaluation in determining skin redness.

The correlation between percentile values and human perception will be explained with reference to FIGS. 14 to 18. 14 and 15 show skin image samples taken in 2015 and 2018 of the same skin area, which were used to verify the correlation between percentile values and human perception. In FIG. 16, evaluation values calculated using the skin color unevenness evaluation formula of the 50th percentile value are plotted for the skin image samples shown in FIGS. 14 and 15. In FIG. 17, evaluation values calculated using the skin color unevenness evaluation formula of the 99th percentile value are plotted for the skin image samples shown in FIGS. 14 and 15. 16 and 17, the vertical axis shows the calculated evaluation value of color unevenness, the horizontal axis shows the number of the skin image sample, the calculated evaluation value in 2015 is shown by △, and the calculated evaluation value in 2018 is shown by ○. ing. FIG. 18 shows skin image samples that were determined to have less color unevenness among the skin image samples of 2015 and 2018, as a result of the sensory evaluation of each skin image sample shown in FIGS. 14 and 15. . In addition, in the sensory evaluation shown in FIG. 18, the color unevenness of the red component, which gives a strong impression to people in skin perception, was mainly evaluated.

Referring to FIG. 16 and FIG. 17, the calculated evaluation values at the 50th percentile value are large for skin image samples 1, 4, 6, and 8 in 2015, and for skin image samples 2, 3, and 8. Regarding No. 5 and No. 7, the calculated evaluation value of the skin image sample in 2018 is large. On the other hand, regarding the calculated evaluation value at the 99th percentile value, the calculated evaluation value for the skin image sample in 2015 is large for skin image samples 1 and 4 to 6, and the calculated evaluation value for the skin image sample in 2018 is large for skin image samples 2, 3, 7, and 8. The calculated evaluation value of the image sample is large. Looking at the results of the sensory evaluation in FIG. 18, skin image samples 1 and 4 to 6 have a high evaluation of color unevenness in 2015 skin image samples, and skin image samples 2, 3, 7, and 8 have high evaluations in color unevenness in 2018. The color unevenness of skin image samples in 2017 was highly rated. From this result, it is understood that the calculated evaluation value at the 50th percentile value does not match the sensory evaluation result regarding the redness component that gives a strong impression to people in terms of skin perception. This means that depending on the component intensity of the 50th percentile value, which indicates the component intensity below the average in the image as a statistical value, it is not possible to give the image characteristics and impressions that humans perceive from the image, and in particular, This is because it is not possible to express the reddish component that gives an impression. Therefore, a calculated evaluation value calculated using a percentile value below the 50th percentile value deviates from human perception and is inappropriate for evaluating uneven skin tone.

As described above, according to the skin color unevenness evaluation device 100 according to the present embodiment, the color unevenness evaluation formula prepared for each percentile value and the component intensity value of each frequency are used as variables. Since the color unevenness evaluation value is determined using the intensity value of the predetermined percentile value among the intensity values, it is possible to use general-purpose equipment without depending on the measurement environment, and it is possible to evaluate skin color unevenness. accuracy can be improved. That is, by using a 90th percentile value or more, more preferably a 95th percentile value or more, and still more preferably a 99th percentile value as a predetermined percentile value, a correlation with a correlation coefficient R = 0.92 or more can be used to improve the effect on the skin. A calculated skin evaluation value that matches human perception, that is, a color unevenness evaluation value can be obtained. In other words, if percentiles of component strengths are not taken into consideration, only average features can be extracted from skin image samples, and features unique to the skin cannot be extracted with high accuracy, resulting in a discrepancy between human perception of skin. Therefore, it was not possible to accurately evaluate skin color unevenness. On the other hand, in this embodiment, strong component intensities included in skin image samples are selectively used, so features unique to the skin can be extracted with high accuracy, and the calculated evaluation value can be matched with the human perception of the skin. It is now possible to match, and uneven skin tone can be evaluated with high accuracy.

In addition, by using a frequency of 1 to 39 Hz as the frequency of the component intensity used in the color unevenness evaluation formula, the calculation load of the color unevenness evaluation value can be reduced and the calculation load can be reduced quickly and without reducing the accuracy of the color unevenness evaluation value. , it is possible to calculate the color unevenness evaluation value without depending on the calculation capacity of the skin color unevenness evaluation device 100, that is, the type of device.

Second embodiment:
In the first embodiment, a dedicated device was used as the uneven skin tone evaluation device 100, but in the second embodiment, a smartphone/tablet was used as the uneven skin tone evaluation device 100. We verified the difference between using dedicated equipment. Figure 19 is a graph showing the correlation between the color unevenness evaluation value when using a dedicated device and the color unevenness evaluation value when using a smartphone. FIG. 2 is an explanatory diagram showing a part of an image sample. Each number in the plot in FIG. 19 is a sample number.

Verification was performed on 11 types of skin image samples, including the three types of skin image samples (sample numbers 3, 4, and 5) shown in FIG. As can be seen from the skin image sample in FIG. 20, when a smartphone is used as the skin color unevenness evaluation device 100, more specifically as a general-purpose imaging device, the environmental light, that is, the shooting light does not become diffused light. , many shadows can be seen. On the other hand, as shown in Figure 19, it can be seen that there is a high correlation between the evaluation values of images captured using a dedicated device and the evaluation values of images captured using a smartphone. The coefficient of determination R ² of the regression equation showing a straight line is 0.9027, and the correlation coefficient R is 0.9501. It was confirmed that there is an extremely high correlation with the values.

As described above, it has been confirmed that according to this embodiment, even when a smartphone is used as the skin color unevenness evaluation device 100, it is possible to obtain the same calculated evaluation value as when using a dedicated device. did it. That is, even if a shadow is reflected in the captured image, a calculated evaluation value that has a high correlation with the sensory evaluation value can be obtained without being affected by the shadow. Therefore, it is possible to evaluate skin tone unevenness with high accuracy using a predetermined imaging environment, that is, an image easily captured by a smartphone without using diffused light illumination to prevent shadows. . In addition, when a smartphone is used as the skin tone unevenness evaluation device 100, the skin tone unevenness evaluation program Pr1 including the skin tone unevenness evaluation formula used in this embodiment can be downloaded from the application providing server to the smartphone. , it is possible to easily evaluate skin color unevenness. When a smartphone is used as the skin tone unevenness evaluation device 100, the user can easily perform the skin tone unevenness evaluation because the imaging and the display of the color unevenness evaluation results can be performed by one device. You can try. Furthermore, it is possible to easily and quickly present an evaluation of skin color unevenness to a prospective purchaser at a permanent or temporary beauty salon or beauty corner without the need for special equipment.

Other embodiments:
(1) In each of the above embodiments, the obtained skin color unevenness evaluation value may be expressed numerically or graphically. When displayed as a graph, it is possible to help the user intuitively understand the evaluation of skin color unevenness, while when displayed as a numerical value, it is possible to more specifically and directly evaluate the skin color. Can show unevenness evaluation. Furthermore, the captured skin image may be displayed simultaneously or selectively on the display unit 21 or the display device 31 together with the evaluation result of color unevenness. In this case, the user can understand the skin color unevenness evaluation along with the skin condition.

(2) In each of the above embodiments, the acquired skin image data is separated into RGB component images, but color components other than RGB components, such as CMY components, Lab components, YIQ components, YUV components, and XYZ components It may be separated into Furthermore, instead of RGB skin image data, CMY skin image data may be used as the skin image data.

Although the present disclosure has been described above based on various embodiments, the embodiments described above are for facilitating understanding of the present disclosure, and do not limit the present disclosure. This disclosure may be modified and improved without departing from its spirit or the scope of the claims, and this disclosure includes equivalents thereof.

DESCRIPTION OF SYMBOLS 10... Skin color unevenness evaluation system, 100... Skin color unevenness evaluation device, 101... Central processing unit (CPU), 102... Memory, 102a... Evaluation image storage area, 20... Imaging unit, 21... Display unit, 30 ...imaging device, 31...display device, F1...skin image acquisition section, F2...skin image analysis section, F21...spatial frequency analysis section, F22...percentile acquisition section, F3...color unevenness evaluation value determination section, Pr1...skin color Mura evaluation program.

Claims (8)

A skin color unevenness evaluation device,
a skin image acquisition unit that acquires a skin image of the evaluation subject;
a skin image analysis unit that performs spatial frequency analysis on the acquired skin image and obtains a percentile of intensity values for each frequency;
Using the intensity value of each frequency as a variable, the color unevenness evaluation formula prepared for each percentile value, and the intensity value of a predetermined percentile value among the intensity values for each frequency obtained above, the color a determining unit that determines an unevenness evaluation value;
A color unevenness evaluation device comprising: The color unevenness evaluation device according to claim 1,
The color unevenness evaluation formula is derived for each percentile value of the intensity value using the color unevenness quantitative evaluation value based on the sensory evaluation of the sample skin image and the intensity value at each frequency obtained by spatial frequency analysis of the sample skin image. Color unevenness evaluation device. The color unevenness evaluation device according to claim 1 or 2,
The color unevenness evaluation device, wherein the predetermined percentile value is a percentile value greater than or equal to a 90th percentile value. The color unevenness evaluation device according to any one of claims 1 to 3,
The skin image analysis unit is a color unevenness evaluation device that separates the acquired skin image into images of each RGB component and performs spatial frequency analysis. The color unevenness evaluation device according to any one of claims 1 to 4,
A color unevenness evaluation device, wherein each of the frequencies is a frequency in a range of 1 to 39 Hz. The color unevenness evaluation device according to any one of claims 1 to 5,
The skin image acquisition unit is a color unevenness evaluation device that acquires a captured skin image from a general-purpose imaging device. A method for evaluating uneven skin tone,
Obtain a skin image of the person to be evaluated,
Performing spatial frequency analysis on the acquired skin image and obtaining percentiles of intensity values for each frequency;
Using the intensity value of each frequency as a variable, the color unevenness evaluation formula prepared for each percentile value, and the intensity value of a predetermined percentile value among the intensity values for each frequency obtained above, the color determining an unevenness evaluation value;
A color unevenness evaluation method comprising: It is a skin tone unevenness evaluation program,
A function to acquire skin images of the evaluation subject,
a function of performing spatial frequency analysis on the acquired skin image and obtaining percentiles of intensity values for each frequency;
Using the intensity value of each frequency as a variable, the color unevenness evaluation formula prepared for each percentile value, and the intensity value of a predetermined percentile value among the intensity values for each frequency obtained above, the color A function to determine the unevenness evaluation value,
This is a color unevenness evaluation program that is realized by a computer.

Images