JP3430944B2 - Impression evaluation method - Google Patents

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention displays a person's impression by an evaluation value objectively obtained based on the characteristics of a facial image so that the person who applies the makeup can easily select a desired makeup method. A method and apparatus.

[0002]

2. Description of the Related Art Conventionally, a makeup technician such as a makeup artist or the like must first determine the ratio of elements such as the entire face or eyes, nose and mouth of a model to be applied in order to realize a message or impression that he or she wants to convey. By observing, the features of the face are grasped, and the features are classified into various impression types. Then, based on the classification,
Various makeup techniques are applied while reflecting the individuality of each makeup technician, and individual makeup is performed so that the makeup as a whole gives the desired impression.

At a cosmetic store, a cosmetics technician such as a cosmetics adviser observes the face and body characteristics of a customer who wants to receive advice, grasps the impression caused by the physical appearance, and recognizes the customer's preference. And adjust the makeup method so that the customer's overall impression is achieved. Then, write down the makeup method on the paper,
I give advice by actually implementing it for some or all of the faces.

[0004]

However, according to this advice method by the cosmetic technician, the characteristics grasped by the cosmetic technician are not explicitly and objectively shown to the customer, and the characteristics of individual parts such as eyes, nose and mouth, etc. There is also no explicit or objective indication of the relationship between the makeup method applied there or the intended overall impression. Therefore, there is a problem that even if the makeup technician writes the makeup method on the paper, the image of the final finish held by the makeup technician is not sufficiently transmitted to the customer. Therefore, even when the makeup is actually applied, if the customer subsequently applies makeup himself, the desired overall impression often does not occur.

Further, the types of impressions classified by a makeup technician also differ depending on the makeup technician. Furthermore, most of the types of types that cosmetic technicians classify are subjective and abstract. Therefore, the classification of types by a cosmetician is artistically advantageous, but it is difficult to unify practically. In the end, the general public uses the classification of impression types of a cosmetician. It is not possible.

The present invention is intended to solve the above problems of the prior art, and a person who receives advice on makeup fully understands the makeup method proposed by the makeup technician,
In order to easily apply makeup to the face with the desired impression even when the makeup is applied by itself, the makeup technique is suitable for the person receiving the advice regardless of the characteristics or skill of the makeup technician. So that you can easily select
The impression that the person who receives makeup advice originally has
It is an object of the present invention to be able to display an evaluation value obtained objectively based on the characteristics of the face image of the person.

[0007]

In order to achieve the above object, the present invention obtains in advance a regression equation of the feature amount of face images of a plurality of persons and the evaluation value of the facial impression on a predetermined evaluation axis. ,
Then, an evaluation method of an impression is provided, which is characterized in that an evaluation value of the impression of the face of the subject is obtained based on the feature amount of the face image of the subject. In particular, in this impression evaluation method, the distance between predetermined measurement points in the face image and the evaluation value of the inquiry about the shape in the face image are used as the feature amount, and the gender is used as the evaluation axis of the face impression. A method of using an evaluation axis based on biological factors such as an impression element axis and an age impression element axis is provided.

Further, as a device suitable for carrying out this method, based on a face image input device, a regression equation of a feature amount of face images of a plurality of persons and an evaluation value of a facial impression on a predetermined evaluation axis, An impression evaluation device comprising a computer having a calculation function for outputting the evaluation value of the impression of the subject from the feature amount of the face image of the subject, and an information display device for displaying the evaluation value of the impression of the subject output from the computer. provide.

In the present invention, the feature amount of the face image includes the amount representing the feature of the shape and color of the face, for example, the measured value between predetermined parts in the face image, the whole face or the nose and nose. It is a numerical representation of the results of an inquiry regarding the shape of each element such as the mouth, etc., according to a predetermined standard.

According to the present invention, the regression equation based on the feature amount of the facial image and the evaluation value of the impression is used to objectively capture the facial features of the person who receives the makeup advice, and the impression that the person originally has. Can be shown to that person.

Therefore, a makeup technician who gives makeup advice naturally has a makeup method suitable for the person receiving the makeup advice to realize the desired facial impression. It is possible to explain the relationship between the facial impression, the makeup method, and the overall impression obtained by the makeup method.

On the other hand, a person who receives makeup advice can easily understand the explanation of the makeup technician, and even if he / she implements the makeup method presented by the makeup technician, he / she can easily obtain a desired impression. You will be able to put on makeup.

[0013]

BEST MODE FOR CARRYING OUT THE INVENTION The present invention will be described in detail below.

FIG. 1 is a block diagram of an embodiment of an apparatus 10 of the present invention for implementing the impression evaluation method of the present invention.

As shown in FIG. 1, the device 10 includes an image input device 1, an information input device 2, a computer 3 and an information display device 4.
(Monitor 4a, printer 4b).

Of these, the image input device 1 is a device such as a video camera or an image scanner that causes the computer 3 to capture a face image. The information input device 2 is a device for inputting numerical information and the like into a computer, and includes a keyboard and the like. The information display device 4 is a device that displays the output of the computer 3 and the like, and includes a monitor 4a and a printer 4b. In addition,
Each of the devices constituting the device 10 of the present invention does not necessarily have to be in the same room or the same facility. For example, the image input device 1, the information input device 2, and the information display device 4 are separated from the computer 3 in a cosmetics store. It may be installed at a store or the like, and these may be connected by a communication line to exchange information remotely.

The computer 3 calculates the evaluation value of the impression of the subject from the feature value of the face image of the subject based on the regression formula prepared in advance, which is the regression value of the feature amount of the face image and the evaluation value of the face impression on a predetermined evaluation axis. Use the one that has the calculation function to output.

In the present invention, for the regression equation, the feature amount of the face images of a plurality of persons and the evaluation value of the impression of the face on a predetermined evaluation axis are input in advance to the computer 3 for loading the face images of the subjects. The one obtained by the above may be used, or the one obtained by a separate computer may be used.

This regression equation can be obtained by processing the feature values of the facial images of a plurality of people and the evaluation value of the facial impression on a predetermined evaluation axis by a known statistical analysis program.

The regression equation is preferably obtained for each age or sex, and the regression equation for the age or sex to which the subject belongs is preferably used. As a result, the correlation between the evaluation value of the subject's impression obtained using the regression equation and the impression obtained from the cosmetician can be increased. In this case, it is preferable to classify the age groups according to the fact that cosmetics and makeup methods generally differ between young, middle-aged, and mature.
Dividing into teens to 20s, 30s to 40s, 50s to 60s, and teens to early 20s, late 20s to 3
It is preferable to divide into 0's, 40's, 50's to 60's.

In the present invention, it is preferable to use, for example, the distance between predetermined measurement points measured by the computer 3 based on the face image taken from the image input device 1 as the feature amount of the face image. As the distance between the measurement points, it is preferable to use the distance between the measurement points determined by statistical extraction. Here, determining the measurement points by statistical extraction means that stepwise multiple regression analysis, etc.
It means obtaining a highly efficient multiple regression equation with high correlation.
As a specific example of the distance between such measurement points, see FIG.
As shown in the front view of the face in (a), the face length Y0, the face width X0, the forehead length Y1, the nose length Y2, the cheek width, especially the cheek width at the mouth height ( It is also called the width of the chin) X4, the length of the upper chin Y3, the lateral width of the eye X1, the vertical width of the eye Y4, the eye-to-eye distance X2, and the lateral center of the eye from the contour line of the face at the eye height. Distance to the position (center of the inner and outer corners of the eye) (Eye position (horizontal)
X6 etc. can be mentioned.

As the feature amount of the face image, in addition to the measured values of the distance between the direct measurement points as described above, the ratio and area obtained by calculation from those measured values, for example, the vertical width Y5 of the mouth are used. The mouth roundness that can be calculated by dividing by the lateral width X3 and the jaw angle that is calculated using the jaw width X4 and the lower jaw length Y6 can be given. Also, the inclination of the straight line connecting the measurement points and the angle formed by the straight lines connecting the measurement points,
For example, the angle θ1 of the straight line L1 connecting the inner corner of the eye and the outer corner of the eye with respect to the horizontal line of the image and the width X5 of the nose are taken as the base, and the vertical bisector of this base is an intermediate line L4 between the upper line L2 of the eye and the lower line L3 of the eye. An apex angle θ2 (referred to as a nose muscle angle) of the isosceles triangle in the case of assuming an isosceles triangle having a vertex P1 at which it intersects with the vertex P1 can also be mentioned.

Further, the distance Z1 between the nose root P4 and the straight line L5 connecting the nose tip P2 and the eyebrow P3 in the side view of the face as shown in FIG. 2B can also be used as the feature amount.

The feature amount of these face images is calculated by the computer 3.
The device itself may be measured based on the image data input from the image input device 1, or may be externally input from the information input device 2 such as a keyboard.

Further, as the feature amount of the face image, inquiry data may be used in addition to the above-mentioned measured value or calculated value. In this case, it is preferable that the inquiry data be input as a numerical value according to a predetermined evaluation standard.

For example, as the inquiry data regarding the degree of rise or fall of the outer corner of the eye, a predetermined evaluation value is assigned to each of the three levels of whether the outer corner of the eye is higher than, lower than or equal to the inner corner of the eye. It is possible to use the obtained numerical value. In addition, as the interview data regarding the degree of passage of the nose, based on three levels, the height of the nose root is close to the height of the forehead, close to the height of the eyes, or between the two, A numerical value obtained by assigning a predetermined evaluation value to each can be used.

The feature amount of the face image is not limited to the above numerical values regarding the shape and size, but numerical values regarding color may be used. As the feature quantity related to this color, the calculator 3
Was output as RGB intensities (intensities of red, green, and blue components) at predetermined parts of the face image captured from the image input device 1, values obtained by a separate chromaticity meter, or digitized by predetermined criteria. It is possible to use the evaluation value of the inquiry.

Further, as the above-described feature amount of the face image, a regression equation calculated based on the feature amount of the face image and the evaluation value of the impression of the face on a predetermined evaluation axis, and further obtained by it. From the viewpoint that the impression is used for makeup advice, it is preferable to use one that includes information on specific facial parts such as eyes, nose and mouth.

On the other hand, in the present invention, the evaluation axis of the facial impression is not the subjective and abstract classification of impressions conventionally classified by a cosmetician, but an evaluation axis based on biological factors. It is preferable to use
It is preferable to use a gender impression element axis expressed as feminine and an age impression element axis expressed as childish-adult.

When the evaluation value of the facial impression is obtained by the regression equation with respect to the gender impression element axis or the age impression element axis, it is actually possible to select and use a feature amount that makes a large contribution to the evaluation axis. It is preferable because it enhances the correlation between the evaluation value obtained by the regression equation and the evaluation value obtained by the cosmetician, and facilitates the calculation of the regression equation and the application of the regression equation to each subject. From such a point, it is preferable to use the cheek width, the degree of ups and downs of the eyes, and the degree of passage of the nose as the feature amount in the regression equation of the gender impression element axis. It is preferable to use the size of the eyes, the roundness of the mouth, the length of the nose and the length of the upper jaw.

As long as the specific data corresponding to the feature amount of the gender impression element axis or the age impression element axis is an index of each feature value, the measured value, the calculated value based on the measured value, the interview evaluation It can be any of the values. Further, the measurement position, the criteria for the inquiry evaluation value, and the like can be set as appropriate. For example, regarding the feature amount of the gender impression element axis, the width X4 of the chin of FIG. 2A can be used as the width of the cheek, and the degree of ups and downs of the eye can be calculated by using the straight line L1 connecting the inner and outer corners of the eye. It is possible to use the angle θ1 with respect to the horizontal line of the image and the inquiry evaluation value regarding rising and falling of the outer corner of the eye. As the condition of the nose muscles, the angle of the nose muscles θ2 and FIG.
The straight line L5 connecting the nose tip P2 and the eyebrow P3 in (b) and the nose root P
The distance Z1 from 4 and the inquiry evaluation value can be used.
Further, regarding the feature amount of the age-targeted evaluation axis, the eye vertical width Y4 in FIG. 2A can be used as the eye size, and the mouth roundness is (mouth vertical width Y5). / (Horizontal width X3) can be used, and the length Y2 between the eyebrow and the tip of the nose can be used as the length of the nose.

Since the evaluation axis based on biological factors such as the gender impression element axis or the age impression element axis exists for all persons whose impression is evaluated, it is common to all persons who want to receive makeup advice. The person who exists and evaluates the person's impression has the basis of that impression. Therefore, it is possible to attribute the classification of impressions, which has been conventionally performed arbitrarily and subjectively by a cosmetic technologist or a general person, to the classification of evaluation axes based on this biological factor. For example, impressions such as sharpness, tightness, and behavior are attributed to those who have a strong masculine tendency in the gender impression element axis, while impressions such as soft, gentle, and gentle tend to have a feminine tendency in the gender impression element axis. Be attributed to you. In addition, the impression of adultness, elegance, calmness, etc. is attributed to the older one on the age impression element axis, and the impression of childish, cute, youthful, etc. is on the lower age axis. Be attributed to you.

As described above, since the evaluation axis based on biological factors exists at the root of all impressions, a person who receives makeup advice may apply makeup that gives an impression of, for example, "passionate" or "neat". Even if desired, it is important for the person who receives the makeup advice or the makeup technician who gives the makeup advice to the person to understand the impression that the person who receives the makeup advice originally has.

As a method of using the device 10 (see FIG. 1) in the present invention, first, a face image of a subject is input by the image input device 1, and the image information is transferred to the computer 3. Further, if necessary, the subject is inquired about the makeup finishing preference and the like, and the answer information is also input to the computer 3.

When the computer 3 receives the information on the face image of the subject from the image input device 1, the computer 3 outputs the impression classification result obtained according to the above regression equation as the impression evaluation result,
The information display device 4 displays it. In this case, computer 3
As the output mode of, the evaluation value corresponding to the feature amount of the face image may be simply output for each evaluation axis, but an appropriate evaluation value or predicted value is set, and the classification of impressions is compared with that. You may make it output a result. For example, it is possible to classify the average value of the evaluation values as a reference value so that when it is larger than that, it looks like an adult, and when it is less than that, it looks like a child.

Further, the output format may be a combination of a plurality of evaluation axes. For example, the classification result of impressions by the regression equation may be represented two-dimensionally with a gender impression element axis of masculine-feminine and an age impression element axis of childish-adult.

The evaluation of the subject's impression thus obtained may be displayed on the screen of the monitor 4a or output to the printer 4b.

The above image input device 1, information input device 2,
The device 10 of the present invention, which comprises the computer 3 and the information display device 4.
In addition, like a known makeup simulation device, the makeup image may be combined with makeup, hairstyle, and clothes on the face image of the subject, and a makeup simulation function that allows the final finish to be visually confirmed may be added. Good. Further, in this case, based on the impression that the face image of the subject inherently presents, a makeup technique including precautions may be introduced in a schematic diagram or an actual moving image so that the subject can select a makeup method.

[0039]

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

Example 1 (1) Acquisition of Regression Formulas Face photographs of 103 female panels in their twenties to thirties were taken, and at the same time, an impression type preferred by each panel was asked in the form of a questionnaire.

Further, the obtained facial photograph was evaluated for two impression element axes of "adultish-childish" and "feminine-masculine" with 20 makeup engineers as evaluators.
"Adult-like" is assigned 1, "Child-like" is assigned -1, and 0 is assigned between the two. Also, "feminine" is 1 and "masculine".
Was assigned to -1, and 0 was assigned to the middle of the two, and the evaluation results of 20 evaluators were averaged and standardized (Z conversion) to obtain an evaluation value of impression.

On the other hand, an image input device (Canon EOS-DCS3c
Digital cameras, information processing devices (Apple Power Maci)
ntosh 8500/150 computer), information input device (Apple Power Macintosh 8500/150 computer), information display device (Radius PressView 17SR display)
Using the evaluation device for impression of the present invention consisting of, the width of the face at eye height is 100
Forehead length, nose length, upper jaw length, lower jaw length, eye width, eye height, mouth width, mouth height, cheek width (face at mouth height) (Defined as the width of)) and the like. Among these, the angle of the jaw was calculated using the length of the lower jaw and the width of the cheek, and the roundness of the mouth was calculated using the vertical width of the mouth and the horizontal width of the mouth.
In addition, the scale evaluation of the degree of rise and fall of the outer corner of the eye and whether or not the nose was passed was performed.

Each of these values is normalized (Z conversion).
Then, using that value, a regression equation for predicting the evaluation value of the age impression factor for "adult-childish" and the evaluation value of the gender impression factor for "feminine-masculine" was obtained. . The formula obtained is shown below.

Age impression element evaluation value = (-0.083) x (length of forehead) + (0.476) x (length of nose) + (0.372) x (length of upper jaw) + (-0.124) x ( Chin angle) +
(-0.403) x (eye size (vertical)) + (-0.047) x (eye size
(Horizontal)) + (-0.249) × (roundness of mouth (= vertical / horizontal of mouth) Gender impression element evaluation value = (-0.294) × (cheek width) + (-0.124) ×
(Chin Angle) + (0.321) x (Extremity) + (0.328) x (Nose Muscle)

The correlation coefficient between the evaluation value obtained from these equations and the evaluation value directly obtained from the evaluator is 0.587, respectively.
It was 0.644. FIG. 3 and FIG. 4 show relationship diagrams between the evaluation values directly obtained from the respective evaluators of age and sex and the evaluation values (fitting values) based on the regression equation. This shows that the evaluation value based on the regression equation and the evaluation value directly obtained from the evaluator have a good correlation.

Further, with the evaluation value of 0 directly obtained from the evaluator as a reference point, the results of four classifications in which the age impression element axis and the sex impression element axis are respectively classified into two, and the evaluation value of 0 based on the regression equation are used as a reference. Similarly, when the results of the four classifications in which the respective two classifications are similarly performed are compared, all the results of the four classifications match (that is, the age impression evaluation value of “adult-childlike”,
For both classifications of “female-male” gender impression evaluation values, the evaluation value obtained directly from the evaluator and the evaluation value based on the regression equation reached 63.1%, and conversely both impressions About 7.8%, there was a discrepancy between the evaluation value obtained directly from the evaluator and the evaluation value based on the regression equation.
The rest are examples in which only one of the classifications matches. This also shows that the evaluation value based on the regression equation and the evaluation value directly obtained from the evaluator have a good correlation.

(2) Classification of Impressions Received from Face Shapes The subjects are 12 females in their twenties and thirties, and the features of the facial photographs of the subjects are measured using the same apparatus as in (1) above. , Listening to the interview items regarding the condition of the nose and the degree of rise or fall of the outer corners of the eyes. Then, the evaluation value of each age impression element and the evaluation value of the gender impression element were predicted based on the regression equation acquired in (1) above. Further, the obtained predicted value and the evaluation value when the person self-evaluates are classified into a total of 4 classifications in which the age impression element axis and the gender impression element axis are each classified into 2 as in the above (1). The degree of coincidence was judged according to the following criteria.

◎: When the classification by the person's self-evaluation and the classification of the evaluation obtained by the regression equation completely match ○: The classification by the person's self-evaluation and the evaluation classification obtained by the regression equation are completely Did not match, but the factors of the classification result obtained by the regression equation (for example, the relationship between the facial feature amount related to the size of the eyes and the evaluation value obtained by the regression equation)
If the person is satisfied by explaining to the person x: The classification by the person's self-evaluation does not match the classification of the evaluation obtained by the regression equation, and the factors of the classification result obtained by the regression equation are explained to the person Even if the person was not satisfied with the results, these results are shown in Table 1.

[0049]

[Table 1] Subject No. Regression classification Classification by the person Classification judgment 1 Childish-feminine Childish-feminine ◎ 2 Childish-feminine Childish-feminine 3 Adult-feminine Adult − Feminine ◎ 4 Childish − masculine Childish − feminine ○ 5 Childish − masculine Childish − masculine ◎ 6 Adult-feminine Adult − feminine ◎ 7 Adult − Male-Adult-Feminine ○ 8 Adult-Feminine Child-Feminine ○ 9 Adult-Feminine Adult-Feminine 10 Child-Men Child-Men Target ◎ 11 Childish-feminine Childish-feminine ◎ 12 Adult-feminine Adult-feminine ◎

As shown in Table 1, the classification of the evaluation values by the regression equation and the classification by the person match each other in 75% of the subjects, and the rest of the subjects also receive the explanation of the factors of the classification, so that the regression equation I was satisfied with the classification of evaluation values. From this, it can be seen that there is some hesitation in the classification of the person himself, and that the objective evaluation based on the regression equation is accepted by the general public as the evaluation by the cosmetician.

(3) Classification of Impressions Received from Faces and Explanation of Makeup Techniques According to the Impressions The subjects are all women in their twenties, image input device (Sony DCR-PC10 digital video camera), information processing device ( Using an impression evaluation device consisting of an Epson VA-626F computer), an information input device (Epson VA-626F computer), an information display device (Radius Press View 17SR display), and an information display device (Epson PM-700C printer), the above ( In the same manner as in 1), the feature amount of the subject's facial photograph was measured, and the interview items regarding the condition of the nose muscles and the degree of rising or falling of the outer corner of the eyes were heard. Then, the evaluation value of each age impression element and the evaluation value of the gender impression element were predicted based on the regression equation acquired in (1) above. Further, the obtained predicted value and the evaluation value when the person self-evaluates are classified into a total of 4 classifications in which the age impression element axis and the gender impression element axis are each classified into 2 as in the above (1). Was determined.

Further, the makeup technique according to the classification result was successively displayed on the display as a moving image for each facial part to give makeup advice. And
We asked them to evaluate this advice based on the following criteria, and we also asked them to freely give their reasons.

[0053] ◎: Very easy to understand advice ○: Advice is easy to understand ×: The advice is difficult to understand The results are shown in Table 2.

[0054]

[Table 2] Subject No. Classification by regression type Classification by the person Classification judgment By evaluation 1 Childish-feminine Adult-feminine ○ ◎ 2 Childish-feminine Childish-feminine ◎ ◎ 3 Childish-female Childish-feminine ◎ ◎ 4 Adult-masculine Childish-masculine ○ ◎ 5 Childish-masculine Childish-masculine ◎ ◎ 6 Childish-feminine Childish- Feminine ◎ ◎ 7 Adult-masculine Adult-feminine ○ ○ 8 Adult-feminine Adult-feminine ◎ 9 Child-feminine Child-feminine ◎ 10 Childish-masculine Childish-masculine ◎ ◎

As shown in Table 2, the classification of the evaluation values by the regression equation and the classification by the person are the same in 70% of the test subjects, and the rest of the test subjects are also explained by the regression formula by receiving the explanation of the factors of the classification. I was satisfied with the classification of evaluation values.

Furthermore, by showing a video of the makeup technique, most of the subjects evaluated that the makeup advice was easy to understand. Also, the reason is that by changing the makeup technique according to the impression classification, it becomes easier to understand how to adapt the basic makeup to yourself,
In many cases, this is because the balance of the entire makeup can be easily considered, which supports the effect of the present invention.

Example 2 (1) Face photographs of 100 female panels in their 40s to 50s were taken. The obtained facial photograph was evaluated by 20 makeup technologists and was classified into 2 categories: "adult-childish" (age impression element axis) and "feminine-masculine" (sex impression element axis). For each of the impression element axes, the evaluation values were assigned to the following 7 levels, and the evaluation values were averaged to obtain the evaluation value of the impression for each facial photograph.

Age impression element axis "Very mature" -3 "Adult-like" -2 "Slightly mature" -1 Midway between "adult-like" and "child-like" 0 "Slightly childish" 1 "Childish" 2 "Very childish" 3 Gender impression element axis "Very feminine" -3 "Feminine" -2 "Slightly feminine" -1 Between "feminine" and "masculine" 0 "Slightly masculine" 1 "Masculine" 2 "Very masculine" 3

On the other hand, the face part was measured using the same apparatus as in Example 1. In addition, scale evaluation was performed in the same manner as in Example 1 regarding the degree of rising and falling of the outer corners of the eyes and the presence or absence of passage of the nose muscles.

In this Example, unlike Example 1, normalization (Z conversion) was not performed for each of the measured value, calculated value and evaluated value, and the multiple regression equation was calculated using the obtained values as they were. . As a result, the following equation including the constant term was obtained.

Age impression element evaluation value = (0.066) × (eye size (vertical)) + (0.006) × (mouth roundness (= vertical / horizontal%)) + (-0.
157) × (length of nose) + (-0.061) × (length of upper jaw) +4.404 Gender impression element evaluation value = (-0.039) × (cheek width) + (0.509) × (external corner of the eye) + (0.349 ) × (nasal muscle) +3.133 The correlation coefficient between the evaluation value obtained from these expressions and the evaluation value directly obtained from the evaluator was 0.515 and 0.704, respectively.

(2) Instead of the above-mentioned scale evaluation, the horizontal line of the image of the straight line L1 connecting the inner corner of the eye and the outer corner of the eye is used as the characteristic amount as the index of the degree of rising and falling of the outer corner of the eye, and whether or not the nose muscle passes. Using the angle θ1 (°) and the nose ridge angle θ2 (°) (see FIG. 2 (a)) with respect to, a regression formula for obtaining the impression value of the gender was obtained, and the following formula was obtained.

Gender impression element evaluation value = (− 0.034) × (cheek width) +
(0.052) × (up and down of the outer corner of the eye (angle θ1)) + (-0.027)
× (nasal muscle angle θ2) +3.928 The correlation coefficient between the evaluation value obtained from this formula and the evaluation value directly obtained from the evaluator was 0.505.

(3) A measurement item is further added, and the angle of the chin and the position of the eye (horizontal) are used as the feature amount used in the regression equation.
(X6 in FIG. 2A) was added and a regression equation for obtaining the impression value of the gender impression element was obtained, and the following equation was obtained.

Gender impression element evaluation value = (-0.036) x (cheek width) +
(0.057) × (up and down of the outer corner of the eye (angle θ1)) + (-0.027)
× (nasal muscle angle θ2) + (-0.013) × (jaw angle) + (0.119) ×
(Eye position (horizontal)) + (-0.124) The correlation coefficient between the evaluation value obtained from this formula and the evaluation value obtained directly from the evaluator was 0.544, which was an improvement.

Example 3 (1) A photograph of the face of a panel of 203 females in their 20s to 50s was taken. 20 facial technologists were used as evaluators of the obtained facial photographs, and "adult-like" (age impression factor evaluation axis) and "feminine-masculine" (sex impression factor evaluation axis) For the two impression element axes of (1), the evaluation values were assigned to 7 grades in the same manner as in Example 2 and averaged to obtain the evaluation value of the impression for each facial photograph.

On the other hand, the face part was measured using the same apparatus as in Example 1. In addition, scale evaluation was performed in the same manner as in Example 1 regarding the degree of rising and falling of the outer corners of the eyes and the presence or absence of passage of the nose muscles.

In this example, as in Example 2, standardization (Z conversion) was not performed on each of the measured value, calculated value and evaluated value, and the obtained values were used as they were to calculate the multiple regression equation. Therefore, the following equation including the constant term is obtained.

Age impression factor evaluation value = (0.312) × (eye size (vertical)) + (0.034) × (mouth roundness (= vertical / horizontal%)) + (-0.
150) x (length of nose) + (-0.185) x (length of upper chin) + 4.458 Gender impression element evaluation value = (-0.049) x (cheek width) + (0.471) x (extremity) + (0.372 ) × (nasal muscle) +3.959 The correlation coefficient between the evaluation value obtained from these expressions and the evaluation value directly obtained from the evaluator was 0.625 and 0.662, respectively.

(2) Instead of the above-described scale evaluation, the angle is the same as in (2) of the second embodiment, as the characteristic amount used as the index of the degree of ups and downs of the outer corner of the eye, and whether the nose muscle is not passed. Using θ1 (°) and nasal muscle angle θ2 (°), a regression equation for obtaining the impression value of the sex impression factor was obtained, and the following equation was obtained.

Gender impression element evaluation value = (-0.053) x (cheek width) +
(0.037) × (up and down of the outer corner of the eye (angle θ1)) + (-0.024)
× (nasal muscle angle θ2) +5.413 The correlation coefficient between the evaluation value obtained from this formula and the evaluation value directly obtained from the evaluator was 0.494.

(3) Further, measurement items were added in the same manner as in (3) of Example 2, and a regression equation for obtaining the impression value of the sex impression element was obtained, and the following equation was obtained.

Gender impression element evaluation value = (− 0.050) × (cheek width) +
(0.049) × (up and down of the outer corner of the eye (angle θ1)) + (-0.022)
× (nasal muscle angle θ2) + (-0.006) × (chin angle) + (0.091) ×
(Eye position (horizontal)) + (3.519) The correlation coefficient between the evaluation value obtained from this equation and the evaluation value obtained directly from the evaluator was 0.511, indicating improvement.

[0074]

According to the present invention, it is possible to objectively evaluate the impression of a person who receives makeup advice based on the characteristics of the face image of the person. Therefore, based on this objectively evaluated impression, the person receiving the makeup advice fully understands the makeup method proposed by the makeup technician, and even if he / she applies the makeup himself, the face with the desired impression is obtained. Makes it easy to put on. Further, it becomes easy for the makeup technician to select a makeup method suitable for the person who receives the advice, without depending on the characteristics or skill.

[Brief description of drawings]

1 is a block diagram of an apparatus of the present invention.

FIG. 2 is an explanatory diagram of measurement points on a face image.

FIG. 3 is a diagram showing a relationship between an evaluation value based on a regression equation and an evaluation value directly obtained from an evaluator for “adult-childlike”.

FIG. 4 is a diagram showing a relationship between an evaluation value based on a regression equation and an evaluation value directly obtained from an evaluator for “feminine-male”.

[Explanation of symbols]

1 Image input device 2 Information input device 3 calculator 4 Information display device 4a monitor 4b printer 10 Impression evaluation device

─────────────────────────────────────────────────── ─── Continuation of the front page (72) Inventor Takahide Minami 2-1-3 Fumika, Sumida-ku, Tokyo Inside Kao Co. Ltd. (56) Reference JP-A-6-129834 (JP, A) JP-A-7 -302327 (JP, A) JP-A-8-1991956 (JP, A) JP-A-7-271964 (JP, A) JP-A-6-333023 (JP, A) Actual break flat 2-63159 (JP, U) ) (58) Fields investigated (Int.Cl. ⁷ , DB name) G06T 1/00 280-340 JISST file (JOIS)

Claims (2)

(57) [Claims] 1. A regression equation is calculated in advance between the facial image feature values of a plurality of people and the facial impression evaluation values on the gender impression element axis and the age impression element axis, and then based on the facial image feature values of the subjects. The evaluation value of the subject's facial impression by the regression equation
An evaluation method Luo seek impression, your gender impression element axis
Kicking regression evaluation value impression of face, as the feature amount, the width of the cheek, use <br/> street degree of eye rise and fall degree and nose, the evaluation value of the impression of the face in the age Impression element axis regression <br/> is, as the feature quantity, mesh size, mouth roundness, the evaluation method of the impression, characterized by using the length of the length and the upper jaw of the nose. 2. The method for evaluating an impression according to claim 1, wherein the regression equation is obtained for each of young, middle-aged or mature, or for each sex.