JP5897745B2 - Aging analysis method and aging analyzer - Google Patents

Description

  The present invention relates to an aging analysis technique.

  Aging care and prevention of aging are a matter of concern not only for women but also for men, and various products dedicated to it such as cosmetics and foodstuffs are in circulation. Many people care about their own aging situation, such as apparent age and age of each part. Patent Document 1 discloses that an aging pattern of a face is obtained from a two-dimensional face image, (1) change direction of a face shape, (2) depression of an upper eyelid, and (3) wrinkle condition of a corner of a mouth. And (4) a method of performing discrimination using five characteristics of the condition of the nasal lip and (5) the shape of the lower jaw as an index. Further, in Patent Document 2, a basis vector obtained by multivariate analysis of the three-dimensional shape information of the face surface is used to calculate a weighting factor of a basis vector related to the subject's face, and based on this weighting factor, the subject's face is calculated. A method for determining the degree of impression tendency of facial features has been proposed.

JP 2001-331791 A Japanese Patent No. 5231585

  As in the method proposed in Patent Document 1, many of the methods for determining the aging situation based on the shape change for each part of the head use the values obtained from the shape information for each part as they are. The tendency is judged. As a result, the aging situation of each person may not be analyzed properly. This is because each individual's original geometric features are directly reflected in the aging tendency. Furthermore, in these methods, the age impression (apparent age) of each person cannot be determined.

  This invention is made | formed in view of such a subject, and provides the technique which analyzes objectively a test subject's aging condition using a test subject's shape feature-value. Here, the head means the neck and the part above the neck in the human body.

  Each aspect of the present invention employs the following configurations in order to solve the above-described problems.

  The aging analysis method according to the first aspect obtains subject feature groups related to a plurality of geometric feature types correlated with age, and extracts a set of feature groups related to a population of a plurality of factors by factor analysis Using the plurality of aging common factors and the subject's feature amount group, a factor expression pattern indicating the expression status of the plurality of aging common factors in the subject is determined, and based on the subject's factor expression pattern, Obtaining aging information of the subject.

  An aging analyzer according to a second aspect is a factor analysis of an acquisition means for acquiring a subject feature group related to a plurality of geometric feature types correlated with age, and a set of feature groups related to a population of a plurality of persons Determination means for determining a factor expression pattern indicating an expression status of the plurality of aging common factors in the subject using the plurality of aging common factors extracted from Output means for outputting the aging information of the subject based on the expression pattern.

  As another aspect of the present invention, an aging care counseling method including the aging analysis method according to the first aspect, a program for causing at least one computer to execute the aging analysis method according to the first aspect, There may be a computer-readable storage medium storing such a program. This recording medium includes a non-transitory tangible medium.

  According to each said aspect, the technique which objectively analyzes a test subject's aging condition can be provided using a test subject's shape feature-value.

It is a figure which shows the aging analysis method in 1st Embodiment. It is a figure which shows notionally the hardware structural example of the aging analyzer in 1st Embodiment. It is a figure which shows notionally the process structural example of the aging analyzer in 1st Embodiment. It is a figure which shows an example (1st pattern determination method) of the apparent age analysis method (analysis method) in 2nd Embodiment. It is a figure which shows notionally the process structural example of the apparent age analyzer (analyzer) in 2nd Embodiment. It is a figure which shows the other process structural example of the apparent age analysis apparatus (analysis apparatus) in 2nd Embodiment. It is a figure which shows an example (2nd pattern determination method) of the apparent age analysis method (analysis method) in 3rd Embodiment. It is a figure which shows the apparent age analysis method (analysis method) in a modification. It is a figure which shows the process structural example of the apparent age analyzer (analyzer) in a modification. It is a figure which shows the example of a shape characteristic kind and an aging common factor. It is a figure explaining the shape characteristic kind illustrated by FIG. 10A. It is a figure which shows the correlation with the apparent age in each factor shown by FIG. 10A. It is a figure which shows the other example of a shape characteristic seed | species and an aging common factor. It is a figure which shows the example of the multiple regression type for obtaining a factor score from a feature-value group. It is a figure which shows the expression condition of the factor score of each test subject obtained from the multiple regression equation shown in FIG. It is a figure which shows the expression condition of the factor score of each test subject obtained by the factor analysis with respect to the set of the feature-value group of a population (497 persons) including a test subject, and an aging common factor. It is a figure which shows the determination method of the factor expression condition in a present Example. It is a figure which shows the classification | category condition of the individual sample for every age of apparent age. It is a figure which shows the example of classification | category of a factor expression pattern. It is a figure which shows the example of the aging information based on the classification | category of a factor expression pattern.

  Embodiments of the present invention will be described below. In addition, each embodiment given below is an illustration, respectively, and this invention is not limited to the structure of each following embodiment.

[First Embodiment]
FIG. 1 is a diagram illustrating an aging analysis method according to the first embodiment. As shown in FIG. 1, the aging analysis method according to the first embodiment acquires a feature amount group of a subject regarding a plurality of geometric feature types correlated with age (S11), and the feature amount group of the subject. And a plurality of aging common factors extracted by factor analysis of a set of feature amount groups related to a population of a plurality of people, and a factor expression indicating an expression status of the plurality of aging common factors in the subject The method includes determining a pattern (S13) and acquiring aging information of the subject based on the factor expression pattern of the subject (S15).

  The age is an apparent age or an actual age. Each feature amount included in the feature amount group acquired in (S11) is information regarding the shape of a part of the body having such a correlation with age. That is, the feature amount group acquired in (S11) is shape information of a plurality of parts of the body.

  The “feature amount” and “geometric feature type” are distinguished and used as follows. “Shape feature type” is information common to each individual that indicates the shape of a specific part of a human, and “feature amount” is a physical quantity that reflects the characteristics of each individual in the shape feature type. . For example, the “shape feature type” is the length under the nose, the relative size of the eye with respect to the cheek width, and the eye inclination (angle) in the vertical direction. “Amount” is a measurement of each individual such as 20 mm (millimeters), 0.32, 2 degrees. As long as such “geometric feature type” is information on a shape having a correlation with age, the body part having the shape is not limited. For example, the “geometric feature type” is a shape of a part of a face and a head (including a neck) other than the face. Further, the “geometric feature type” may be a shape other than the head, such as an abdomen, a back of the hand, and a leg. Each feature amount can be expressed in various units such as a length, an angle, a curvature of a curved surface or a curve, and a ratio.

  In (S13), a plurality of common factors obtained by performing factor analysis on a set of feature amount groups related to a population of a plurality of people are used. This common factor is expressed as an aging common factor. Then, the feature value group of each individual sample (each individual) included in the set of feature value groups corresponds to a plurality of geometric feature types similar to the feature value group of the subject obtained in (S11). In (S13), the factor expression pattern which shows the expression condition of the said several aging common factor is determined regarding a test subject using the test subject's feature-value group obtained by (S11). For example, the factor expression pattern can indicate the presence or absence of expression for each aging common factor. However, the expression method of the factor expression pattern is not limited. The factor expression pattern can indicate the degree of expression for each aging common factor, or can indicate only the aging common factor that is expressed or only the aging common factor that is not expressed.

  In (S15), the aging information of the subject is acquired based on the factor expression pattern of the subject determined in (S13). The acquired aging information can indicate the age impression (apparent age) of the subject, and the aging degree (hereinafter, the shape) such as what age the shape characteristic indicated by the feature group corresponds to in total. May also be described as age). Here, the acquisition of the aging information in (S15) can be executed in various modes. For example, aging information can be assigned in advance for each factor expression pattern. There may be as many factor expression patterns as there are combinations of common aging factors. In this case, aging information assigned to the factor expression pattern of the subject is acquired. Moreover, (S15) also includes making a test subject's aging information into the state which can be grasped | ascertained by the person. For example, (S15) may be realized in such a manner that a table representing aging information for each acquired factor expression pattern is provided together with the factor expression pattern of the subject determined in (S13). In this case, the person can grasp the aging information of the subject by comparing the factor expression pattern of the subject with the table. In addition, the specific aspect of acquisition of aging information in (S15) is mentioned later.

  The aging analysis method in the first embodiment can be executed by at least one computer such as an aging analysis apparatus described below. However, the above-described aging analysis method may include a step in which at least a part is performed by a person. For example, in (S13), the intermediate process may be executed by a computer, and only the final factor expression pattern may be determined by a person based on information calculated by the computer. Moreover, also in (S15), only acquisition (understanding) of the final test subject's aging information, such as a person comprehending the test subject's factor expression pattern and its table and ascertaining the test subject's aging information May be performed by a person. The aging analysis method according to the first embodiment uses a correlation between age and a person's shape characteristic species, and a correlation between a common factor between the shape characteristic kinds and age. Achieving a certain effect by repeatedly and continuously obtaining aging information of a subject easily and objectively without actually comparing it with other people's information or actually hearing a number of judges Is the method. Therefore, even if a process performed by a person is included, the aging analysis method according to the first embodiment can be said to be a creation of a technical idea using the laws of nature as a whole.

  FIG. 2 is a diagram conceptually illustrating a hardware configuration example of the aging analyzer 10 in the first embodiment. The aging analyzer 10 is a so-called computer, and includes, for example, a CPU (Central Processing Unit) 11, a memory 12, an input / output interface (I / F) 13, a communication unit 14, and the like that are connected to each other via a bus. The memory 12 is a RAM (Random Access Memory), a ROM (Read Only Memory), a hard disk, or the like. The communication unit 14 exchanges signals with other computers and devices. A portable recording medium or the like can be connected to the communication unit 14.

  The input / output I / F 13 can be connected to user interface devices such as the display device 15 and the input device 16. The display device 15 displays a screen corresponding to drawing data processed by a CPU 11 or a GPU (Graphics Processing Unit) (not shown) such as an LCD (Liquid Crystal Display) or a CRT (Cathode Ray Tube) display. Device. The input device 16 is a device that receives an input of a user operation such as a keyboard and a mouse. The display device 15 and the input device 16 may be integrated and realized as a touch panel. The hardware configuration of the aging analyzer 10 is not limited.

  FIG. 3 is a diagram conceptually illustrating a processing configuration example of the aging analyzer 10 in the first embodiment. As shown in FIG. 3, the aging analyzer 10 uses the acquisition unit 21 that acquires a subject feature amount group related to a plurality of geometric feature types correlated with age, and the subject feature amount group, A determination unit 22 for determining a factor expression pattern indicating an expression state of a plurality of common aging factors extracted by factor analysis on a set of feature amount groups of a population related to a plurality of geometric feature types; And an output processing unit 23 that outputs the aging information of the subject based on the expression pattern. Each of these processing units is realized, for example, by executing a program stored in the memory 12 by the CPU 11. Further, the program is installed from a portable recording medium such as a CD (Compact Disc) or a memory card or another computer on the network via the input / output I / F 13 or the communication unit 14 and is stored in the memory 12. It may be stored.

  The acquisition unit 21 executes (S11) described above. The acquisition unit 21 can also acquire a subject's feature amount group from information input by a user operating the input device 16 based on an input screen or the like, or from a portable recording medium, another computer, or the like. It can also be acquired via the communication unit 14 or the input / output I / F 13.

  Moreover, the acquisition part 21 can also produce | generate the test subject's feature-value group itself from the information regarding a test subject. For example, the aging analyzer 10 further includes at least one of a contact-type measurement unit and a non-contact type measurement unit (both not shown), and the acquisition unit 21 obtains the three-dimensional coordinates of a predetermined part of the subject obtained therefrom. It is possible to automatically calculate a feature amount group of the subject regarding the plurality of geometric feature types from the information. An example of the contact type measurement unit is a contact type three-dimensional digitizer. Examples of the non-contact measurement unit include a three-dimensional laser scanner and a distance image sensor. Further, the acquisition unit 21 can directly acquire a feature amount group from a two-dimensional image in which the subject is photographed, and can determine a predetermined amount of the subject by a well-known method from a plurality of two-dimensional images obtained by photographing the subject from different directions. The three-dimensional coordinate information of the part can be calculated, and the feature quantity group of the subject can be automatically calculated using this. In calculating the feature quantity group from the three-dimensional coordinate information, normalization based on a homologous model can also be used.

  In order to directly acquire the feature amount group from the two-dimensional image, a known method such as a method using the luminance distribution around each pixel or a method using the arrangement of face parts may be used. By using such a well-known method, it is possible to automatically calculate the feature amount group of the subject approximately from the two-dimensional image. Depending on the shape feature type, it may be difficult to detect feature points. In this case, the feature point can be predicted from the average value of each feature point automatically acquired from the population.

For example, in the following reference, in the measurement of feature points of maxillofacial skeleton shape, the coordinates of a point that cannot be directly measured by an unknown individual, the coordinates of measurable points around that point, and many known shape measurement values A method of estimating (interpolating) from the correspondence with the statistical average has been proposed. In this method, the coordinates (6 points) of the morphological feature points on the soft tissue are estimated from the coordinates of the surrounding measurable points (25 points) and the coordinate average of the sample population.
References: Yoshimitsu Aoki et al., Medical Imaging Technology, 22 (5), 250-258 (2004).
Thus, in this embodiment, the acquisition method of the test subject's feature-value group by the acquisition part 21 is not restrict | limited.

  The determination unit 22 executes (S13) described above.

  The output processing unit 23 executes (S15) described above and outputs the acquired aging information. However, the output mode of the aging information by the output processing unit 23 is not limited. The output processing unit 23 generates output data indicating aging information, and outputs the output data to the display device 15 and other output devices such as a printing device via the input / output I / F 13. The output processing unit 23 may transmit the output data to another device via the communication unit 14, or may record the output data on a portable recording medium.

[Operation and Effect of First Embodiment]
The present inventors select a plurality of geometric feature types having a correlation with age from all the geometric feature types possessed by a person, and select a plurality of common factors (additional) from the feature quantity group related to the selected geometric feature types. Common age factors) were found, and it was verified that each of these common factors correlated with age. Furthermore, the present inventors investigated the relationship between the expression pattern and the aging in each individual of the aging common factors extracted as described above, and found that the factor expression pattern represents an aging tendency.

  Thereby, according to 1st Embodiment, a factor expression pattern can be determined from a test subject's feature-value group, and a test subject's aging information can be obtained as a result. Here, since the factor expression pattern is determined not only from the feature amount of each specific part but also from the subject feature amount group regarding a plurality of geometric feature types, the aging information obtained from this factor expression pattern is Less susceptible to personal shape features. Furthermore, according to the first embodiment, information on the apparent age of the subject can also be obtained. Thus, according to the first embodiment, it is possible to objectively analyze the aging situation of the subject using the shape feature amount of the subject.

  In addition, the present inventors classify a plurality of geometric feature types for each aging common factor having a strong dependence on the basis of the factor loading of the aging common factor, so that each aging common factor is human. It was found that a specific variation of each part strongly affects the age, which is associated with each part. Therefore, according to the factor expression pattern, aging information for each part of the subject can be obtained.

  Hereinafter, the details of the first embodiment will be described. In the following, as a detailed embodiment, an apparent age analysis method (hereinafter sometimes abbreviated as an analysis method) and an apparent age analysis device (hereinafter also abbreviated as an analysis device) in the second embodiment are described. Illustrated. That is, in the following second embodiment, apparent age information is acquired as aging information. However, also in the following second embodiment, shape age information can be acquired as aging information as in the first embodiment described above. The following description will focus on the content different from the first embodiment described above, and the same content as in the first embodiment will be omitted as appropriate.

[Second Embodiment]
The analysis method in the second embodiment also includes the same steps as the aging analysis method in the first embodiment shown in FIG. In the second embodiment, further details of each step will be described. Hereinafter, the analysis method according to the second embodiment will be described with reference to FIGS. 1 and 4.

  The feature amount group acquired in (S11) is personal information of the subject regarding a plurality of geometric feature types correlated with the apparent age. Examples of such shape feature types include at least one of the length under the nose and the thinness of the lips, the size of the eyes, the inclination of the eyes, the inclination of the eyebrows, and the swelling of the lower jaw. Including at least two of them. These shape characteristic types are shape information of each part of the head that is extracted by the examination of the present inventors and shows a high correlation with the apparent age. However, in this embodiment, as long as there are a plurality of geometric feature types correlated with the apparent age, the body part representing the same is not limited.

  Here, the plurality of geometric feature types corresponding to the feature amount group of the subject and the feature amount group of each individual sample included in the population are selected from among a number of geometric feature types on the human head surface. , Sorted as follows. That is, the plurality of geometric feature types correlated with age are obtained for each of the geometric feature types obtained by multiple regression analysis using the feature quantity group related to the many geometric feature types as an explanatory variable group and age as a target variable. Sorted based on multiple correlation coefficients. For example, a shape feature type whose multiple correlation coefficient for each shape feature type is higher than the limit value of the significance level for the number of samples of the population of the multiple regression analysis is selected. In the present embodiment, an apparent age is used as the age.

  The plurality of aging common factors are extracted by performing factor analysis on the set of feature amount groups of the population related to the plurality of geometric feature types selected in this way. The number of factors used in the factor analysis can be determined according to, for example, the selected geometric feature type. Furthermore, for each extracted aging common factor, the correlation coefficient between the factor score (for example, average) of the aging common factor and age in the individual samples included in the factor analysis population is the number of samples in the population. It is desirable to be larger than the limit value of 1% significance level in However, the reference for the correlation coefficient height is not limited to this reference, and may be set as appropriate.

  The method for determining the factor expression pattern of the subject in (S13) includes a method using a factor score (hereinafter referred to as a first pattern determination method) and a method not using a factor score (hereinafter referred to as a second pattern determination method). ). The second pattern determination method will be described in the third embodiment.

  FIG. 4 is a diagram illustrating an example of the analysis method (first pattern determination method) in the second embodiment. When the first pattern determination method is used, the analysis method in the second embodiment is (S41), (S43), and (S45) instead of (S13) in the first embodiment, as shown in FIG. Including. That is, in 2nd Embodiment, the process (S13) which determines a test subject's factor expression pattern is detailed to (S41), (S43), and (S45).

(S41) calculates the test subject's factor score for each of the plurality of aging common factors using the test subject's feature value group acquired in (S11).
(S43) determines the expression status of the aging common factor in the subject by comparing the threshold indicating the predetermined position in the factor score distribution of the population with the factor score of the subject for each of the plurality of aging common factors.
(S45) determines a factor expression pattern indicating at least one aging common factor determined to be expressed as an expression state in (S43) for the subject.

  There are various methods for calculating the factor score of the subject in (S41). First, there is a method of estimating factor scores for each aging common factor of the subject from the feature amount group of the subject using multiple regression analysis. Specifically, a multiple regression analysis for each age-related common factor with the feature group related to the plurality of geometric feature types as the explanatory variable group and the factor score as the objective variable is performed on the set of feature groups related to the population. A factor score for each aging common factor of the subject can be calculated by applying the feature amount group of the subject to each multiple regression equation obtained by applying. This method is referred to as a first factor score calculation method.

  In the factor analysis for extracting the aging common factor, the factor loading for each aging common factor is calculated for each shape feature type corresponding to each feature amount. Further, factor scores for each aging common factor are calculated for each individual sample of the population used as a factor analysis sample. As a result of factor analysis, the factor loading amount for each aging common factor and the factor score for each aging common factor for each individual sample of the population are stored in the computer together with the set of feature amount groups in the population. Stored.

  In the first factor score calculation method, multiple regression for each aging common factor is performed in advance by multiple regression analysis using the factor score for each aging common factor and the feature amount group for each individual sample of the population. An equation is obtained, and this multiple regression equation is further stored in a computer such as an analyzer 30 described later. Each multiple regression equation is an equation for explaining the factor score of each individual aging common factor from the individual feature amount group. The first factor score calculation method calculates a subject's factor score for each of a plurality of common aging factors by substituting the subject's feature amount group for each multiple regression equation stored in the computer. . As described above, the first factor score calculation method uses the multiple regression equation obtained in advance by multiple regression analysis for the set of feature quantity groups of the population not including the subject, so that the factor score of the subject can be obtained at high speed. Can be calculated.

  Further, after calculating the factor score of the subject in this way using the first factor score calculation method, the feature amount group of the subject is added to the set of feature amount groups of the original population, and a new feature amount group obtained Apply new multiple regression analysis for each aging common factor to the set, and replace (update) each original multiple regression equation obtained from the original population with each obtained multiple regression equation It may be. In this way, the number of populations increases each time the number of subjects increases, and the accuracy of the multiple regression equation stored in the computer can be improved.

  As another method for calculating a factor score of a subject, there is a method in which a subject is added to the population and factor analysis is applied to the new population. According to this method, the factor loadings for each aging common factor are calculated in consideration of the subject's feature amount group, and the aging common factor is calculated for each individual subject included in the population. Each factor score can be calculated. In this method, a set of feature amount groups (matrix A) regarding a new population including subjects, a correlation coefficient group (matrix B) between a plurality of geometric feature types, and factors newly obtained by factor analysis By using the load group (matrix C) and multiplying the matrix A and the inverse matrix of the matrix A and the matrix B, the factor score of each aging common factor related to the subjects included in the population can be calculated. According to this method, the factor score of the subject can be calculated.

  Threshold values indicating predetermined positions in the population factor score distribution used in (S43) are determined for each aging common factor. One threshold may be set for each aging common factor, or a plurality of thresholds may be set. For example, an average value of the factor score distribution may be used as the threshold value, or a value obtained by adding or subtracting a certain value to the average value may be used. (S43) can determine the presence or absence of expression for each aging common factor as the expression status. For example, if a subject's factor score for a common aging factor is greater than the threshold for the common aging factor, the subject determines that the common aging factor is expressed in the subject and is below that threshold. It can be determined that the aging common factor is not expressed in the subject. Moreover, (S43) may each determine the expression score which shows the degree of expression about each aging common factor. For example, when the factor score of a certain common aging factor of the subject is larger than the first threshold value for the common aging factor, the expression score of the common aging factor in the subject is +1, and the factor score is the second When the expression score is -1 when the value is smaller than the threshold, and the factor score is equal to or lower than the first threshold and equal to or higher than the second threshold, the expression score can be determined to be 0.

  In (S45), the factor expression pattern of the subject is determined in accordance with the expression status determined in (S43). In 2nd Embodiment, a factor expression pattern is expressed in the aspect which shows the at least 1 aging common factor determined to be expressing. In this case, the total number of factor expression patterns is equal to the number of all combinations of common aging factors (a value obtained by multiplying 2 by the number of common aging factors). As described above, the expression mode of the factor expression pattern is not limited.

  In the second embodiment, (S15) acquires the apparent age information of the subject based on the factor expression pattern of the subject determined in (S45). Here, the apparent age information can be predetermined for each factor expression pattern. For example, the apparent age is evaluated for each individual sample of the population by sensory evaluation by an expert. Then, in the same manner as described above, a factor expression pattern is obtained in advance for each individual sample of the population, the population is classified according to the factor expression pattern, and an average apparent age is calculated for each factor expression pattern. Keep it. In this way, the average apparent age can be prepared in advance for each factor expression pattern. In this case, (S15) acquires the average apparent age of the subject's factor expression pattern as apparent age information. However, the apparent age information acquisition method is not limited to such an example.

  As another example, all factor expression patterns are classified in advance into a plurality of groups according to the commonality of aging information (apparent age information), and aging information (apparent age information) is assigned to each group. You can also keep it. For example, all factor expression patterns can be classified into groups based on at least one of the actual or apparent age, factor score, and factor expression number of each individual sample with the same factor expression pattern. .

  According to the study by the present inventors, it has been found that the apparent age tends to be higher as the number of common aging factors is lower and the apparent age is lower. Has been. Therefore, for example, if all the factor expression patterns are classified into a plurality of groups according to the number of expression of the aging common factor, an apparent age group can be assigned to each group. Specifically, a factor expression pattern in which only one aging common factor is expressed is classified into the lowest age group among the plurality of groups, and all aging common factors are expressed. An expression pattern and a factor expression pattern in which one less common aging factor is expressed than all the common aging factors can be classified into the highest age group among a plurality of groups.

  When the factor expression patterns are pre-classified in this way, (S15) selects a group corresponding to the subject's factor expression pattern from the plurality of groups, and is assigned to the selected group. Apparent age information can be acquired. However, the classification of the factor expression pattern is not limited only to the classification based on the expression number of the aging common factor. For example, as described above, each group obtained by classification based on the expression number of the aging common factor can be classified into child groups by cluster analysis based on the factor score of each individual sample belonging to the group. According to this further classification, the factor expression pattern groups classified into the same age group can be further grouped according to the tendency of factor scores, and not only the apparent age group but also the tendency of factor scores. Apparent age information can be obtained.

〔Device configuration〕
Next, the analyzer according to the second embodiment will be described focusing on the contents different from the first embodiment. The same contents as those in the first embodiment are omitted as appropriate. The analysis apparatus in the second embodiment has the same hardware configuration as the aging analysis apparatus 10 in the first embodiment shown in FIG. 2, and the above-described hardware configuration and the processing configuration described below are used. The analysis method in the second embodiment is executed. The above-mentioned content regarding the analysis method in the second embodiment is followed by the analysis apparatus in the second embodiment.

  FIG. 5 is a diagram conceptually illustrating a processing configuration example of the analysis apparatus 30 according to the second embodiment. As illustrated in FIG. 5, the analysis device 30 includes an acquisition unit 31, a determination unit 32, an output processing unit 33, and the like. The determination unit 32 includes a calculation unit 34. The acquisition unit 31, the determination unit 32, and the output processing unit 33 correspond to the acquisition unit 21, the determination unit 22, and the output processing unit 23, respectively, in the first embodiment. Each of these processing units is realized by executing a program stored in the memory 12 by the CPU 11 in the same manner as each processing unit in the first embodiment.

  The acquisition unit 31 executes (S11). Furthermore, the acquisition unit 31 is included in a set of feature amount groups related to the population, a correlation coefficient group between a plurality of geometric feature types, a factor load amount group obtained by factor analysis, and the population as necessary. You may make it acquire the factor score group etc. of each individual sample. The analysis device 30 may hold the information itself, or may acquire it from a portable recording medium, another computer, or the like.

  As described above, the determination unit 32 includes the calculation unit 34 that calculates the subject's factor score for each of the plurality of age-related common factors using the subject's feature amount group acquired by the acquisition unit 31, respectively. The factor expression pattern of the subject is determined based on the factor score for each aging common factor of the subject calculated by the calculation unit 34. That is, the determination unit 32 executes (S41), (S43), and (S45). Among these, (S41) is executed by the calculation unit 34.

  As described above, the calculation unit 34 can calculate the factor score of the subject by various methods. For example, the calculation unit 34 can use the first factor score calculation method described above. In this case, the calculation unit 34 performs a multiple regression analysis for each aging common factor with the feature quantity group related to the plurality of geometric feature types as an explanatory variable group and the factor score as an objective variable to a set of feature quantity groups related to the population. By applying the subject's feature quantity group to each of the multiple regression equations obtained by applying to each of the multiple regression equations, the factor score for each subject's aging common factor can be calculated.

  The calculation unit 34 can also use a multiple regression equation of each aging common factor stored in advance, or can use the multiple regression equation acquired from another device. Further, the calculation unit 34 applies the multiple regression analysis for each aging common factor to the set of feature amount groups related to the population and the feature amount group of the subject, It is also possible to replace the original multiple regression equations obtained from the original population.

  The calculation unit 34 can also calculate a factor score for each aging common factor of the subject by adding the subject to the original population and newly performing factor analysis on the new population. In this factor analysis, the calculation unit 34 updates the factor load amount group, and further calculates a factor score for each aging common factor for each individual sample (including subjects) of the new population. In this case, the calculation unit 34 calculates the age-related common factors related to subjects included in the population by multiplying the above matrix based on the set of feature amount groups, the correlation coefficient group, and the factor load amount group acquired by the acquiring unit. A factor score for each can also be calculated.

  The determination unit 32 determines the aging common factor expressed in the subject by comparing the threshold indicating the predetermined position in the factor score distribution of the population and the factor score of the subject for each of the plurality of aging common factors. A factor expression pattern indicative of at least one age-related common factor determined to be expressed is determined for the subject. The determination part 32 may hold | maintain the said threshold value for every aging common factor previously, and may acquire it from another computer. A specific factor expression pattern determination method is as described above.

  The output processing unit 33 executes (S15), and further outputs the acquired apparent age information of the subject as aging information. For example, the output processing unit 33 outputs aging information indicating the name given to each of the plurality of aging common factors and the factor expression pattern of the subject. The name of each aging common factor can be given, for example, so as to represent a characteristic tendency common to geometric feature species having a large factor loading of each aging common factor. As a result, according to the output aging information, it is possible to express the subject's characteristic tendency regarding the apparent age. Furthermore, the output processing unit 33 can also include the average apparent age for each factor expression pattern prepared in advance as described above in the apparent age information. The output mode of the output processing unit 33 is the same as that of the output processing unit 23.

  Moreover, the output process part 33 can also output the aging information attached | subjected to the group selected according to the test subject's factor expression pattern. In this case, as described above, all the factor expression patterns are classified into a plurality of groups based on at least one of the age, factor score, and factor expression number of each individual sample having the same factor expression pattern, Each group is preliminarily attached with aging tendency information regarding the apparent age according to the classification method. The output processing unit 33 outputs apparent age information about the subject by acquiring group expression information of factor expression patterns and apparent age information (aging information) attached to each group. The factor expression pattern grouping information and apparent age information attached to each group may be held in advance by the selection unit 35 described later, or may be acquired from another computer.

When the factor expression patterns are classified in this way, the analysis device 30 further includes a selection unit 35 in addition to the configuration shown in FIG. 5 as shown in FIG.
FIG. 6 is a diagram illustrating another processing configuration example of the analyzer 30 according to the second embodiment.
The selection unit 35 corresponds to a factor expression pattern of a subject from among a plurality of groups in which all factor expression patterns corresponding to all combinations of expression states of each aging common factor are classified according to the commonality of aging. Select the group to be used. The output processing unit 33 outputs apparent age information attached to the group selected by the selection unit 35 as aging information of the subject.

[Operation and Effect of Second Embodiment]
As described above, in the second embodiment, the subject's factor score for each of the plurality of common aging factors is calculated using the subject's feature quantity group, and the subject's factor score is calculated based on the calculated factor score. An expression pattern is determined. Here, each factor score of the subject represents the degree of expression of the corresponding common age-related factor in the subject, and each age-common factor has a high correlation with the apparent age, so the subject's score obtained from the factor score The factor expression pattern is considered to represent the occurrence of some factor that affects the apparent age in the subject. Therefore, according to the second embodiment in which the apparent age information of the subject is obtained based on such a factor expression pattern, the apparent age of the subject can be analyzed with high accuracy.

  In the first factor score calculation method, which is one of the methods for calculating the subject's factor score, the subject's factor score can be obtained at high speed by substituting the subject's feature quantity group into the previously obtained multiple regression equation. Can be obtained. Then, by comparing this factor score of the subject with a threshold indicating a predetermined position in the factor score distribution of the population, the expression status of the subject's aging common factor is determined, and this expression status determines the factor expression pattern of the subject. It is determined. In this way, by determining the expression status of the subject's aging common factor in consideration of the factor score distribution of the population, the objectivity of the subject's factor expression pattern can be enhanced, and from that factor expression pattern, The objectivity and credibility of the obtained aging information (apparent age information) can be improved.

  In the second embodiment, each factor expression pattern is classified into a plurality of groups based on at least one of the age, factor score, and factor expression number of each individual sample having the same factor expression pattern. A group corresponding to the factor expression pattern is selected. And the apparent age information attached | subjected to the selected group is acquired as a test subject's information. In this way, it is possible to acquire the apparent age group, factor expression tendency, and the like attached to each group as the apparent age information as the apparent age information of the subject.

[Third Embodiment]
In the second embodiment described above, the factor expression pattern of the subject is determined based on the factor score for each aging common factor of the subject (first pattern determination method). In 3rd Embodiment, a test subject's factor expression pattern is determined based on a test subject's feature-value group, without using a test subject's factor score (2nd pattern determination method). Hereinafter, the third embodiment will be described focusing on the contents different from the above-described embodiments, and the same contents will be appropriately omitted.

  The analysis method in the third embodiment also includes the same steps as the aging analysis method in the first embodiment shown in FIG. In the third embodiment, further details of each step will be described. Hereinafter, the analysis method in the third embodiment will be described with reference to FIG.

  FIG. 7 is a diagram illustrating an example of the analysis method (second pattern determination method) in the third embodiment. When the second pattern determination method is used, as shown in FIG. 7, the analysis method in the third embodiment replaces (S13) in the first embodiment with (S71), (S73), and (S75). Including. That is, in 3rd Embodiment, the process (S13) which determines a test subject's factor expression pattern is detailed to (S71), (S73), and (S75).

  (S71) acquires a representative value of the feature amount group for each of the plurality of factor expression patterns. In the third embodiment, for each individual sample of the population, the factor expression pattern is determined according to the factor score for each aging common factor of the individual sample, and for each of all the factor expression patterns The representative value of the feature amount group of each individual sample having the same factor expression pattern is determined. The representative value of each factor expression pattern is determined from the feature value group of each individual sample having the same factor expression pattern. Data indicating the representative position in the feature value space (vector having the same number of elements as the feature value group) For example, it is data indicating the center of gravity of the feature quantity group. However, the representative value is not limited to the center of gravity.

  (S73) calculates the distance between the feature amount group of the subject acquired in (S11) and the representative value of each factor expression pattern acquired in (S71). As described above, the representative value is a vector having the same number of elements as the feature amount group. In (S73), the Euclid distance is calculated.

  (S75) determines the factor expression pattern of the subject based on the distance between the representative value of each factor expression pattern calculated in (S73) and the feature amount group of the subject. Specifically, the factor expression pattern indicating the minimum distance is determined as the factor expression pattern of the subject.

〔Device configuration〕
Next, the analyzer according to the third embodiment will be described focusing on the contents different from the first embodiment. The same contents as those in the first embodiment are omitted as appropriate. The analysis apparatus in the third embodiment has the same hardware configuration as the aging analysis apparatus 10 in the first embodiment shown in FIG. 2, and the above-described hardware configuration and the following processing configuration are used to The analysis method in the third embodiment is executed. The above-mentioned content regarding the analysis method in the third embodiment is also followed by the analysis apparatus in the third embodiment.

  The analyzer 30 in the third embodiment has the same processing configuration as that of the second embodiment shown in FIG. 5 or FIG. However, each of the following processing units executes processing different from that of the second embodiment.

  The acquisition unit 31 further executes (S71) described above in addition to (S11). The acquisition unit 31 may itself hold a representative value of the feature quantity group of each factor expression pattern, or may acquire it from a portable recording medium, another computer, or the like.

The calculation unit 34 executes (S73) described above.
The determination unit 32 executes (S75) described above.

[Operations and effects in the third embodiment]
As described above, in the third embodiment, the distance between the subject feature amount group and the representative value of the feature amount group of each factor expression pattern calculated in advance is calculated, and based on this distance, the subject factor is calculated. An expression pattern is determined. Therefore, according to the third embodiment, the factor expression pattern of the subject can be determined at high speed only by the distance calculation without calculating the factor score of the subject.

[Modification]
The factor analysis itself based on the above-mentioned aging common factor may be executed by the aging analyzer 10 in the first embodiment and the analyzer 30 in the second and third embodiments, or another computer. May be executed. In addition, the factor analysis population may be updated by sequentially capturing subjects to be analyzed by the aging analyzer 10 and the analyzer 30. In this case, as shown in FIG. 8, (S17), (S18), and (S19) are newly executed. FIG. 8 shows an example in which a new processing step is added to the analysis method in the second embodiment shown in FIG. 4, but the new processing step is an addition in the first embodiment shown in FIG. You may add to the age analysis method or the analysis method in 3rd Embodiment shown by FIG.

  FIG. 8 is a diagram illustrating an analysis method in the modification. The analysis method acquires a feature group of a new individual sample that is not included in the population regarding a plurality of geometric feature types, and collects a feature group of the new individual sample and a feature group of the original population. The factor analysis using the number of the aging common factors is executed (S17), and the factor load amount group newly obtained by the factor analysis of (S17) is used to add the new population to the original population. A factor score for each aging common factor is newly calculated for each individual sample of the new population to which the individual sample is added (S18), and the newly obtained factor loading group and each individual sample of the new population The method further includes updating the original information obtained from the original population with the factor score for (S19). In FIG. 8, the feature amount group of the subject is used as a feature amount group of a new individual sample (S11). However, the new individual sample may be other than the subject.

  FIG. 9 is a diagram illustrating a processing configuration example of the analysis apparatus 30 according to the modification. In this case, the analysis apparatus 30 further includes an analysis processing unit 37 that performs factor analysis using the number of common aging factors for a plurality of individual sample feature amount groups regarding the plurality of geometric feature types. The acquisition unit 31 further acquires a feature amount group of a new individual sample that is not included in the original population regarding the plurality of geometric feature types. The analysis processing unit 37 performs factor analysis on the feature group of the new individual sample and the set of feature groups of the original population, and uses the factor load group newly obtained by the factor analysis. Then, a new factor score for each age-related common factor for each individual sample of the new population with the new individual sample added to the original population is newly calculated. The original information obtained from the original population is updated with the factor scores for each individual sample of the population. The factor load amount group and the factor score group obtained from the original population and the new population may be held by the analysis device 30 or may be held by another computer. Moreover, the acquisition part 31 can also use a test subject's feature-value group as a feature-value group of a new individual sample.

  According to this modification, a feature group of a new person (including a subject) not included in the original population is added to the set of feature groups of the original population, and factor analysis is performed again. Since the original information is updated in the newly obtained factor load amount group and factor score group, learning progresses and analysis accuracy can be improved each time a new human feature amount group is obtained.

  Examples will be given below to describe the above-described embodiments in more detail. The present invention is not limited in any way by the following examples.

  FIG. 10A is a diagram illustrating examples of shape feature types and aging common factors. FIG. 10A shows an example in which five common factors are extracted by factor analysis on the feature quantities of 13 shape feature species in a population of 497 Japanese women in their 20s to 60s. In FIG. 10A, the five common factors are denoted by factor 1 to factor 5. Thirteen geometric feature types are indicated by numerals 1 to 13. The subscript y in FIG. 10A means a component in the vertical direction (direction perpendicular to the ground). For example, “(eyebrows-eyebrow head) y” means a vertical drop (vertical distance) from the eyebrows to the eyebrows.

FIG. 10B is a diagram illustrating the shape feature types illustrated in FIG. 10A.
The vertical width y of the eye indicates the length of the longest portion in the vertical direction of the eye, and the area of the eye is the exposed area of the mucous membrane of the eye. The cheek width is different from the cheek arch width, in the position of the cheekbone below the outer eye angle on the intersection of the plane perpendicular to the ground passing through the left and right outer eye angles of the person in front view and the surface of the subject's face. Width.
The degree of suspension is an angle formed by a straight line connecting the inner eye angle (the head of the eye) and the outer eye angle (the eye corner) and a straight line extending horizontally from the inner eye angle to the ground. The apparent hang degree is an angle formed by a straight line connecting the apparent butt point and the inner eye angle (eye) and a straight line extending horizontally from the inner eye angle to the ground. The apparent corner point means an apparent corner point, for example, the outer eye corner end of the upper eyelid that hangs down from the outer eye angle.
(Eyebrow-eyebrows) y indicates the vertical drop (vertical distance) from the eyebrows to Mt. Brow, and (eyebrows-eyebrows) y indicates the vertical drop (vertical distance) from the eyebrows to the eyebrows. Show.
The fish degree indicates the inclination in the front-rear direction of a straight line connecting the inner eye angle and the outer eye angle. The fish degree can be said to be the degree of retraction of the corner of the eye.
“A tragus point−mandibular angle point / mandibular angle point−chin point” is a value obtained by dividing the distance from the tragus point to the mandibular angle point by the distance from the mandibular angle point to the chin point.
The lower jaw corner point width is the width between the re-projections near the left and right lower jaw corner points.
(Lip upper end [left and right]-lower end) y / crack width indicates the aspect ratio of the lips. Specifically, the vertical drop from the upper end of the lip to the lower end of the lip (vertical distance) Indicates the divided value. (Lower nose-upper lip [left / right] / lower nasal-chin) y indicates the length below the nose, specifically the vertical drop from the lower nose to the upper lip. The value is divided by the vertical drop from the lower point to the lower jaw.
Although not shown in FIG. 10B, the left cheek angle is an angle formed by a straight line connecting the left tragus point and the left cheek width point and a straight line connecting the left cheek width point and the left nose wing point. The left cheek minimum curvature passes through the cheekbone position below the outer eye angle on the intersection of the plane perpendicular to the ground and the surface of the subject's face passing through the left and right outer eye angles of the person in front view from the nose. The minimum curvature of the curve. In other words, the left cheek portion minimum curvature is the minimum curvature of the cheek portion of the cross section of the head and face that passes through the left cheek width point, the nose lower point, and the right cheek width point.

  The present inventors extracted 13 such geometric feature species as follows. The present inventors analyzed more than 4000 data points (feature points) in the homologous model based on the analysis of the amount of change of the feature points correlating with the apparent age and the knowledge of interpreting the relationship between the feature points of the modeling experts. Using this, a relationship (geometric feature type) between 60 feature points, which is presumed to be related to the apparent age, was derived. In this process, the shape feature types are excluded so that there are not a plurality of highly correlated shape feature types. Then, the inventors statistically analyze the relationship between the feature amount of 60 geometric feature types and the apparent age, and has a high correlation with the apparent age (the multiple correlation coefficient is greater than 0.200). Thirteen geometric feature types as described above were extracted. As the statistical analysis, a multiple regression analysis was used in which feature amount groups related to 60 shape feature types were explanatory variable groups and apparent age was an objective variable.

  FIG. 10C is a diagram showing a correlation with the apparent age in each factor shown in FIG. 10A. Further, the inventors calculated the correlation coefficient between the factor score and the apparent age in each individual sample included in the population for each common factor, and the correlation coefficient of each common factor is shown in FIG. 10C. As shown, it was confirmed that each was larger than the limit value of the 1% significance level in the number of samples of the population. In this way, common aging factors related to the apparent age, such as the example of FIG. 10A, were found.

  On the other hand, comparing the factor loadings for each geometric feature type for each common factor, as shown in FIG. 10A, classify the 13 geometric feature types by the aging common factors that are strongly influenced. Can do. Specifically, the shape feature types 1 to 3 (hanging degree, apparent degree and fish degree) are highly dependent on factor 1, and shape feature types 4 and 5 are highly dependent on factor 2. Geometric feature types 6 and 7 are highly dependent on factor 3, geometric feature types 8 and 9 are highly dependent on factor 4, and geometric feature types 10 and 11 are highly dependent on factor 5. From this relationship, factor 1 is a factor corresponding to the inclination of the eye, factor 2 is a factor corresponding to the inclination of the eyebrows, factor 3 is a factor corresponding to the size of the eye, and factor 4 is The factor 5 corresponds to the swelling of the lower jaw, and the factor 5 is considered to be a factor corresponding to the length under the nose and the thinness of the lips.

  However, the number of aging common factors and the shape characteristic species are not limited to those shown in FIG. 10A. For example, a group of feature quantities related to a plurality of geometric feature types correlated with apparent age includes (a) a plurality of feature quantities indicating the length of the nose and lip thinness, and (a) the eye relative to the cheek width. (C) a plurality of feature quantities indicating the inclination of the eyes in the vertical direction and the front-rear direction, (d) a plurality of feature quantities indicating the drooping state of the eyebrows with respect to the cheek width, and (e It is sufficient to include a plurality of feature amounts indicating the size of the lower jaw angle width with respect to the cheek width and the positional relationship between the jaw tip, the ear, and the lower jaw angle.

  FIG. 11 is a diagram illustrating another example of the shape feature species and the aging common factor. In the example of FIG. 11, an example in which five aging common factors are extracted with 12 shape feature types different from FIG. 10A is shown. Also in the example of FIG. 11, it was confirmed that the correlation coefficient with the apparent age of each aging common factor becomes larger than the limit value of the 1% significance level in the number of samples of the population.

  Next, a specific example of a technique (see the second embodiment) for calculating a factor score for each subject's common aging factor by applying the subject's feature group to the multiple regression equation for each common aging factor. explain. In this specific example, an example of obtaining a multiple regression equation and an evaluation of a subject's factor score obtained by the multiple regression equation are described. Here, the geometric feature species and aging common factors shown in FIG. 10A are used.

  In this specific example, first, 472 people who were randomly selected as subjects from the above-mentioned population of 497 Japanese women in their 20s to 60s were randomly selected as subjects. Factor analysis was performed on a set of feature groups in the temporary population. Thereby, the factor loadings for each aging common factor for each of the 13 geometric feature types are calculated, respectively, and the factor scores for each aging common factor for each individual sample of the temporary population are respectively calculated. Calculated. Next, the multiple regression equation for each aging common factor shown in FIG. 12 is obtained by multiple regression analysis using the factor score for each aging common factor and the feature amount group for each individual sample of the temporary population. Calculated.

  FIG. 12 is a diagram illustrating an example of a multiple regression equation for obtaining a factor score from a feature amount group. FIG. 12 shows partial regression coefficients for each geometric feature type for each aging common factor (factor 1 to factor 5).

  FIG. 13 is a diagram showing the factor score of each subject obtained from the multiple regression equation shown in FIG. 12 and the expression status of the aging common factor. FIG. 13 shows factor scores for each aging common factor of 25 subjects. In addition, the expression status of each aging common factor is represented by expression scores of -1, 0, and +1, and factor scores with an expression score of -1 or +1 are shaded. In the leftmost column of FIG. 13, the age of the subject is shown as the age (20's etc.), the number of the subject is shown in the second column from the left, and the expression is shown in the rightmost column. The number of the aging common factor whose score is -1 or +1 is shown.

  FIG. 14 is a diagram showing the factor score of each subject and the expression status of common aging factors obtained by factor analysis on a set of feature quantity groups of a population (497 persons) including subjects. FIG. 14 shows factor scores obtained by factor analysis for a population including subjects, so that the factor scores of each subject shown in FIG. 14 and the expression status of aging common factors should be originally obtained (correct answer). It is. Therefore, when the expression status of each aging common factor is compared between FIG. 13 and FIG. 14, only the expression status of factor 1 of the subject (309) is different between FIG. 13 and FIG. The expression score of the factor 1 of the subject (309) estimated in FIG. 13 is +1, whereas the correct expression score shown in FIG. However, the expression status of common aging factors for the other 24 subjects is consistent, and as a result, it is verified that the factor score estimated by the method of FIG. 13 shows a high accuracy of 24/25. It was done.

  Furthermore, the present inventors conducted a factor analysis with a common factor number of 4 for a set of feature quantity groups of nine geometric feature types in a population of 248 people, and for each individual sample of the population. Factor expression pattern was determined. On the other hand, the present inventors classify each individual sample by the factor expression pattern (16), calculate the center of gravity in the feature space (9 dimensions) for each factor expression pattern, and for each individual sample, the factor expression pattern The distance from the center of gravity for each was calculated, and the factor expression pattern with the smallest distance was determined as the factor expression pattern of each individual sample. As a result, it was confirmed that the factor expression pattern determined using the distance coincided with the factor expression pattern determined by factor analysis at a rate of 81.1%.

  FIG. 15 is a diagram showing a method for determining the factor expression status in this example. In the present embodiment, the threshold value for obtaining the factor expression status from the factor score is a first threshold value obtained by adding 0.25σ (σ is a standard deviation) to the average factor score of the aging common factor ( + 0.25σ) and a second threshold (−0.25σ) obtained by subtracting 0.25σ from the average was used. Since the standard deviation σ is calculated for each aging common factor, the first threshold value and the second threshold value are provided for each aging common factor. In this example, when the factor score of a subject's common aging factor is greater than the first threshold for that aging common factor, the expression score of that aging common factor in the subject is +1, When the factor score was smaller than the second threshold, the expression score was -1, and when the factor score was the first threshold or less and the second threshold or more, the expression score was determined to be 0.

  Furthermore, the present inventors conducted a factor analysis on the above-mentioned population (497 persons) based on the expression score of the aging common factor in order to verify the correlation between the expression status of the aging common factor and the apparent age. Each individual sample obtained by the above was classified into the following four types (S group, N group, X group, Y group). The S group is a type in which the +1 expression score is 2 or more more than the −1 expression score, the N group has the +1 expression score of 1 or less, the −1 expression score of 1 or less, and the +1 The difference between the number of expression scores of -1 and the number of expression scores of -1 is 1 or less, the X group has 2 or more expression scores of +1, 2 or more expression scores of -1, and The difference between the number of +1 expression scores and the number of -1 expression scores is one or less, and the Y group is a type in which the expression score of -1 is two or more more than the expression score of +1.

  The present inventors have summarized the classification status of individual samples for each age of apparent age, and as shown in FIG. 16, demonstrated that there is a correlation between the appearance status of the aging common factor and the apparent age. FIG. 16 is a diagram showing the classification status of individual samples for each age of apparent age. According to FIG. 16, as the apparent age is higher, the ratio of the individual samples belonging to the S group is increased, and as the apparent age is lower, the ratio of the individual samples belonging to the Y group is increased. Thus, it was proved that there was a correlation between the appearance of common aging factors and the apparent age, and the more the common aging factors were expressed, the higher the apparent age.

  FIG. 17 is a diagram illustrating a classification example of factor expression patterns. Using the factor expression pattern of each individual sample determined in the example of FIG. 10A, the inventors calculate the average apparent age of individual samples having the factor expression pattern for each factor expression pattern. By using age, factor expression patterns were classified into three groups (denoted as region 1, region 2, and region 3). In FIG. 17, the number of the common aging factor determined that the circled number is expressed is shown, and a square surrounding at least one circled number indicates one factor expression pattern, and is placed below the square. The numbers in parentheses indicate the average age, and the numbers to the right of the numbers in parentheses indicate the number of individual samples belonging to the factor expression pattern.

  As shown in FIG. 17, it can be seen that each group corresponds to an apparent age group of 30s, 40s, and 50s. Furthermore, the factor expression pattern in which only one aging common factor is expressed is classified into the lowest age group among a plurality of groups, and all aging common factors (5) are expressed. The factor expression pattern in which the number of common aging factors, which is one less than all the common aging factors (4) is expressed, is classified into the highest age group among several groups. Yes.

  The present inventors further analyze the factor expression patterns belonging to each group represented as region 2 and region 3 by performing cluster analysis using the factor score of each individual sample, thereby further dividing the factor expression patterns into further groups. Classified. FIG. 17 shows that the factor expression patterns surrounded by circles belong to the same group.

  FIG. 18 is a diagram illustrating an example of aging information based on the classification of factor expression patterns. For each group generated by the classification of the factor expression patterns as described above, a name, a factor expression pattern, the number of aging common factors expressed, an average value of apparent ages, and the number of persons are assigned. By presenting such a table to the subject together with the factor expression pattern determined for the subject, aging status information can be provided to the subject. For example, for a subject who expresses only one common aging factor, the apparent age should be indicated on the site (eyes, eyebrows, lower jaw, lower nose, lips, etc.) corresponding to the expressed common aging factor. It is possible to present an analysis result that there is an increasing factor. In this way, if a site with a factor that increases the apparent age can be identified, a cosmetic treatment corresponding to the identified site can be proposed. Further, the average apparent age assigned to the corresponding group can be presented as the apparent age of the subject.

  Thus, the method and apparatus in the above-described embodiments can be used for aging care counseling. In this case, if information on the corresponding cosmetic treatment is associated with each factor expression pattern or each group in which the factor expression pattern is classified in advance, the cosmetic treatment information can be provided together with the aging information of the subject. .

  In addition, in the some flowchart used by the above-mentioned description, although several process (process) is described in order, the execution order of the process performed by each embodiment is not restrict | limited to the order of the description. In each embodiment, the order of the illustrated steps can be changed within a range that does not hinder the contents. Moreover, each above-mentioned embodiment and each modification can be combined in the range with which the content does not conflict.

10 Aging analyzer 11 CPU
12 memory 15 display device 16 input device 21 acquisition unit 22 determination unit 23 output processing unit 30 apparent age analysis device (analysis device)
31 Acquisition Unit 32 Determination Unit 33 Output Processing Unit 34 Calculation Unit 35 Selection Unit 37 Analysis Processing Unit

Claims (17)

Acquire subject feature groups for multiple geometric feature types that correlate with age,
Expression of the plurality of common aging factors in the subject using a plurality of common aging factors extracted by factor analysis of a set of feature amount groups related to a population of a plurality of persons, and the feature amount group of the subject Determine the factor expression pattern that indicates the situation,
Based on the factor expression pattern of the subject, to obtain the aging information of the subject,
Aging analysis method including the above. Determination of the factor expression pattern is
Using the feature quantity group of the subject, the factor score of the subject for each of the plurality of aging common factors is respectively calculated.
Including
Based on the factor score for each aging common factor of the subject, determine the factor expression pattern of the subject,
The aging analysis method according to claim 1. Determination of the factor expression pattern is
By comparing the threshold indicating a predetermined position in the factor score distribution of the population and the factor scores of the individual samples included in the subject or the population for the plurality of aging common factors, respectively, the subject or the individual samples Determine the expression status of aging common factors in
Determining the factor expression pattern for at least one aging common factor determined to be expressed as the expression status for the subject or the individual sample;
The aging analysis method according to claim 2. The calculation of the subject's factor score is:
Applying a multiple regression analysis for each of the age-related common factors having a feature group related to the plurality of geometric feature types as an explanatory variable group and a factor score as a target variable to the set of the feature groups related to the population Applying the feature group of the subject to each of the multiple regression equations respectively obtained in
The aging analysis method according to claim 2 or 3. The population does not include the subject;
The aging analysis method is:
Applying multiple regression analysis for each aging common factor to the set of feature groups related to the population and the feature groups of the subject,
Replace each original multiple regression equation obtained from the population with each multiple regression equation obtained by multiple regression analysis for each aging common factor,
The aging analysis method according to claim 4, further comprising: The population includes the subjects;
The calculation of the factor score of the subject is performed using the set of the feature amount groups related to the population, the correlation coefficient group between the plurality of geometric feature types, and the factor load amount group obtained by the factor analysis. Calculating a factor score for each aging common factor for each individual sample of the population,
The aging analysis method according to claim 2 or 3. Each individual sample of the population has a factor expression pattern determined according to the factor score for each aging common factor of the individual sample,
Determination of the factor expression pattern is
For each of a plurality of factor expression patterns determined with respect to the population, respectively, to obtain a representative value of a feature amount group of individual samples having the same factor expression pattern,
Calculating the distance between the feature quantity group of the subject and the representative value of each factor expression pattern,
Including
Determining the factor expression pattern of the subject based on the distance;
The aging analysis method according to claim 1.   The correlation coefficient between the factor score and the age in each individual sample included in the population with respect to each aging common factor is larger than the limit value of the 1% significance level in the number of samples of the population, respectively. The aging analysis method according to any one of 1 to 7. The plurality of geometric feature species correlated with age is at least one of the length of the nose and the thinness of the lips, the size of the eyes, the tilt of the eyes, the tilt of the eyebrows, and the swelling of the lower jaw Including two,
The aging analysis method according to any one of claims 1 to 8. The age is an apparent age,
The number of aging common factors is five,
The feature amount group relating to the plurality of geometric feature types correlated with the apparent age indicates a plurality of feature amounts indicating a length of the nose and a thinness of the lips, and a relative size of the eyes with respect to the cheek width. A plurality of feature amounts, a plurality of feature amounts indicating the inclination of the eyes in the vertical direction and the front-rear direction, a plurality of feature amounts indicating the hanging state of the eyebrows relative to the cheek width, and the size of the lower jaw angle width and the chin tip Includes multiple features that indicate the positional relationship between the ear and mandibular angle,
The aging analysis method according to any one of claims 1 to 9. The plurality of geometric feature types correlated with age is an explanatory variable group of feature quantity groups related to the large number of geometric feature types on the human head surface, and age is an objective variable. Is selected based on the multiple correlation coefficient for each geometric feature type obtained by multiple regression analysis.
The aging analysis method according to any one of claims 1 to 10. An aging care counseling method using the aging analysis method according to any one of claims 1 to 11 ,
For each factor expression pattern or for each group in which the factor expression pattern is classified, in advance, associate the corresponding cosmetic treatment information,
Providing information on the corresponding cosmetic treatment based on the factor expression pattern determined for the subject by the aging analysis method;
Aging care counseling methods including that . An acquisition means for acquiring a subject's feature amount group related to a plurality of geometric feature types correlated with age;
Expression of the plurality of common aging factors in the subject using a plurality of common aging factors extracted by factor analysis of a set of feature amount groups related to a population of a plurality of persons, and the feature amount group of the subject A determination means for determining a factor expression pattern indicating the situation;
Based on the factor expression pattern of the subject, output means for outputting the aging information of the subject,
An aging analyzer comprising: The output means outputs the aging information indicating the name given to each of the plurality of aging common factors and the factor expression pattern of the subject.
The aging analyzer according to claim 13. The output means outputs the aging information attached to a group selected according to the factor expression pattern of the subject.
The aging analyzer according to claim 13 or 14. Analysis processing means for performing factor analysis using the number of the common factors for aging with respect to the feature amount group of a plurality of individual samples regarding the plurality of geometric feature types,
The acquisition means further acquires a feature amount group of a new individual sample that is not included in the population regarding the plurality of geometric feature types,
The analysis processing means includes
Performing the factor analysis on the set of feature values of the new individual sample and the set of feature values of the population;
Using the factor loading group newly obtained by the factor analysis, the factor score for each age-related common factor for each individual sample of the new population obtained by adding the new individual sample to the population is newly set. Calculate
Update the original information obtained from the population by factor scores for each individual sample of the newly obtained factor loading group and the new population,
The aging analyzer according to any one of claims 13 to 15.   A program for causing at least one computer to execute the aging analysis method according to any one of claims 1 to 11.

Images