KR20160094204A - Generation method of standard 3-dimensional avatar by human body type - Google Patents

Abstract

The present invention relates to a method for generating a standard 3D avatar by human body shape. The method for generating a standard 3D avatar by human body shape according to the present invention comprises the steps of: average size analysis; image plane; 3D human body modeling; and 3D research avatar evaluation and supplementation. Through the steps, an average body shape and four group body shapes by body shape for adult women are designed to be utilized in a 3D virtual fitting program. A result from size suitability analysis for 3D avatar human body model groups by body shape for adult women established in the present invention shows a 0.5 error rate, thereby proving that the 3D avatar human body model groups can be utilized as a virtual fitting model in the ready-made clothes industry.

Description

{GENERATION METHOD OF STANDARD 3-DIMENSIONAL AVATAR BY HUMAN BODY TYPE}

More particularly, the present invention relates to a method of generating a 3D representative avatar for a human body by dividing a human body type into four groups, analyzing an average size of each body type, The present invention relates to a method of generating a 3D representative avatar for a human body modeling a human body in 3D by designing a two-dimensional image of the front face.

Today, the Virtual Fitting System is emerging as a new and efficient system for garment production and sales. In addition, the utilization of avatars, which are virtual models, is increasing, and demands of 3D avatars similar to actual body types are increasing.

The reproduction of the actual human shape of the avatar in the virtual fitting is very important factor. Unlike other avatars, avatars are required for seamless virtual fittings. However, most of the avatars used in virtual fitting programs are closer to the ideal body shape than the actual body shape.

Research on human body modeling for avatar production has been conducted in various fields. Currently, 3D avatars are used in virtual simulation of various apparel industries such as virtual apparel, internet shopping, and apparel industry, games, automobiles, and furniture. This suggests that avatars need integrated research in various industries and academies that need virtual human models.

The virtual model avatar can be classified into the CG direct production method, the dimension input method, and the morphing method of the 3D data shape. The virtual model production method of the dimension input method allows the user to easily create the avatar. However, It is not possible to represent all body types by realistic numerical input.

Since the virtual fitting program creates a virtual model as a dimension input method in the representative avatar, it is necessary to study the human dimensional conformity of the representative avatar. In addition, since the human figure is not limited to one figure but divided into several figure types, there is a problem in the present method of producing a virtual model by presenting only one representative avatar.

SUMMARY OF THE INVENTION The present invention has been made to overcome the above-described problems of the prior art, and it is an object of the present invention to classify human body types into four groups and to realize a 3D standard avatar representing four body types.

In order to achieve the above object, the present invention proceeds from a process leading to an average dimensional analysis, an image plane, a 3D human body modeling, and evaluation and supplementation of a 3D research avatar. Through this design, 4 are designed to be used in 3D virtual fitting program. As a result of the dimensional fit analysis of the 3D avatar human body model group according to the body shape of the adult women constructed in this study, it was verified that the error rate is ± 0.5 and that the 3D human body model seen in the ready-to-wear industry can be utilized as a virtual fitting model .

According to the 3D representative avatar generation method of the human body type according to the present invention having the above-described structure, the 3D standard avatar type that represents each body type is realized by classifying human body types into four groups.

BRIEF DESCRIPTION OF THE DRAWINGS FIG. 1 is a diagram showing a virtual fitting program and representative avatars of the present invention. FIG.
FIG. 2 is a diagram showing the size measurement items of the avatar according to the present invention. FIG.
3 is a flowchart showing the flow of the 3D human body modeling step of the present invention.
4 is a diagram showing an image plane design item of the present invention.
5 is a diagram illustrating a sizing system process of a virtual fitting program of the present invention.
6 is a diagram showing a comparative analysis of the number of avatar sizing items of the present invention.
FIG. 7 is a diagram showing comparison and analysis of landmarks and similar landmarks according to the virtual fitting program of the present invention. FIG.
FIG. 8 is a diagram showing a comparison between the number of landmark items of the DC Suite and the technical standard center 2010 of the Ministry of Commerce, Industry and Energy of the present invention.
9 is a view showing a size comparison analysis of avatar and adult female average body type according to the automatic size conversion system of the present invention.
FIG. 10 is a view showing the average body shape of 20 Korean adult women per virtual fitting program of the present invention. FIG.
11 is a view showing an analysis of the average body size of Korean adult women according to the avatar type of the present invention.
FIG. 12 is a diagram illustrating an analysis of planar dimensions through avatar size conversion according to adult female body type according to the present invention.
13 is a diagram showing an avatar shape according to an adult female body type according to the virtual fitting program of the present invention.
14 is a diagram showing factors and factor loadings for the research item of the present invention.
15 is a diagram showing factors and factor loadings for the research item of the present invention.
16 is a diagram showing the distribution of the study subjects according to the number of clusters of the present invention.
FIG. 17 is a diagram showing a human body silhouette and degree of polymerization according to body type of a Korean adult female of the present invention. FIG.
18 is a diagram showing a design process of a research avatar according to the present invention
19 is a diagram showing reference points and base line settings of the Image Plane of the present invention.
20 is a view showing the human standard posture and divided images of the present invention.
21 is a diagram showing an outline setting of an image plane according to a research process of the present invention.
22 is a diagram showing an avatar shape according to adult female body types for setting an outline of an image plane of the present invention.
23 is a diagram showing the execution of the Image Plane of the present invention.
24 is a view showing the trunk modeling of the present invention.
25 is a diagram showing limb modeling of the present invention.
26 is a view showing the trimming of the human body detail of the present invention.
Fig. 27 is a view showing a human curved surface application shape of the present invention. Fig.
28 is a diagram showing the completion of the left figure of the human body modeling of the present invention.
29 is a diagram showing the completion of human body modeling of the present invention.
30 is a diagram showing a key scale enlargement of the present invention.
31 is a diagram showing the setting of a measurement reference point of the present invention.
32 is a view showing the circumferential dimension measurement of the present invention.
33 is a view showing the length measurement of the present invention.
34 is a view showing a modification of the dimensions of the present invention.
FIG. 35 is a view showing the verified avatars of the present invention. FIG.
36 is a diagram showing an avatar according to the average and body type of Korean adult women of the present invention.
FIG. 37 is a diagram showing comparative analysis of body types of the commercial avatar and the research avatar of the present invention. FIG.
FIG. 38 is a view showing the dimensional fitness analysis of the research avatar according to the Korean adult female body type of the present invention. FIG.
39 is a diagram showing a research avatar import in the virtual fitting program of the present invention.

The present invention may have various modifications and various embodiments, and specific embodiments are illustrated in the drawings and described in detail in the detailed description. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Hereinafter, the present invention will be described in detail with reference to the accompanying drawings.

However, the present invention is not limited to this, and a 3D representative avatar including a man, an infant, an adult, and the like can be created.

Currently, a variety of 3D virtual fitting programs are commercialized in the apparel industry and education field. It can be classified into programs that can identify the costume coordination and wearing patterns in online shopping malls, and programs that can produce sample clothes and 3D avatar clothes in the apparel industry. Among them, three sophisticated programs were selected for related research and industry. The order of presentation of the subjects was alphabetical and the real names were used in consideration of the open programs. The screen of the program used in the research and the representative avatar are shown in Fig.

Through interviews with the training staff of program producers and searching through each program 's tutorial book and homepage, the avatars sizing system related functions and avatar characteristics of the program were investigated. The computer used to run the program is Microsoft Windows 7 Home Premium, Intel Core i7 650 CPU, 500GB hard disk and 16GB memory. In order to respond to the global sales network and global education, we set the program language in English. The survey period is from June 2012 to October 2012.

By running the three selected virtual fitting programs (CLO 3D, DC Suite, and 3D Runway), we compare and analyze avatar sizing items and principles in the avatar sizing system by avatar sizing manual. For the programs that can check the landmark Landmark items were analyzed.

In order to analyze the human body suitability of Korean avatar size, we compare and analyze the avatar size generated by the dimension direct input method according to the size of the representative avatar of each program and the sizing system. The order of research is summarized as follows.

Enter the average size of the 20 Korean adult women presented by the Ministry of Commerce, Industry and Energy (2004) in the basic entry system of the program, and check that the size of the detailed item set automatically matches the size. Next, in order to verify the size difference of the program automatic size conversion system, the result of conversion of the detailed dimension according to the basic dimension input per program was analyzed. The analytical items are shown in Table 1 as 13 items and common calculation items (key / head vertical length, waist height / key, hip-waist circumference difference) and 16 avatar size items set in common for each program.

In order to objectively verify the degree of implementation and program specificity of the Korean adult female standard body group of the virtual fitting avatar sizing system, four standard body types of 20 female Korean adult body types were produced and analyzed through the avatar sizing system .

For this purpose, basic and detailed items applicable to the avatar sizing program were entered and implemented as four body type avatars, and the measured dimensions of each program avatar were compared and analyzed. Using the 2D measurement system of AutoCAD, the measurement method was 5 items in height, 4 items in width, 4 items in thickness, 14 items in avatar left, right and side. Is shown in FIG. The measurement method was based on the human body measurement method of the Korea Technology Standard Institute (2010).

This study was carried out using the 3D shape measurement measured in the 6th Korean Human Dimension Survey Project. The study subjects were 410 women in their 20s and 30s. Among them, four case cases with extreme values found in each analysis item among cases were extracted through the plot box search and excluded to improve the accuracy of the study. The research ages were 205 (50.4%) in the 20s (20 to 29 years) and 201 (49.6%) in the 30s (30 to 39 years) The study was conducted together.

After analyzing the characteristics of Korean women 's body shape and classifying their body type, the following body type analysis was performed to present 3D standard avatar human body model group according to each body type. The 3D measurement data was analyzed using IBM SPSS Statistic 21.0, and the process was as follows.

First, analyze the average size of Korean women in their 20s and 30s by performing descriptive statistics according to age groups.

  Second, body type classification is performed for the avatar figure of Korean women. To do this, the relationship between the measurement items is identified and a factor analysis is conducted to summarize the information that the measurement itself has. Next, we perform cluster analysis with each extracted factor as independent variable as a result of factor analysis, and implement the type by body type.

Third, the image type plan for the 3D standard avatar human body modeling by using the virtual fitting program of the average body shape of each adult body type and the body shape of each body type was verified by ANOVA Build data.

 Among 73 items of 3-D anthropometric measurement items and 21 direct measurement items for database search conducted by the 6th Korean Human Dimension Survey (Ministry of Commerce, Industry and Energy, 2011), 3D human body shape Two items of measurement, 50 items of 3D measurement and 55 items of 3 items were selected and used in the study as shown in Table 3.

Through the body type analysis of Korean women, the average size analysis by body type was carried out. Based on this, AutoCAD 2010 program is used to design Top, Front, Side (right) Image Plane. Based on the next designed Image Plane, 3D human body modeling was performed in Maya 2013. The flow of the 3D human body modeling step is shown in FIG.

The image plane design corresponding to the plan view was performed in the order of base line setting using plane size, outline setting using avatars of standard posture and commercial virtual fitting program. The detailed procedure is as follows.

  (13), length (4), width (7), and slope (1) items of the 3D measurement items of the Agency for Technology and Standards (2010) (2) 33 design items and 9 reference point items were used as shown in FIG.

Next, we set the posture of the human body based on the normal posture and the human body proposal presented in Im, Dong-Taek (2006), and use the avatar sizing system implemented by using DC Suite's avatar sizing system, Respectively.

We used Maya 2013 version for 3D human body modeling design and evaluation.

In order to verify the size of the 3D human body modeling shape by the body shape of the Korean adult female, it is necessary to determine the size of the adult female with reference to the dimensions of the girth of KS K 0051 among the circumference dimension and length dimension items excluded from the image plane creation 4 items for circumference, 5 items for length. The details are shown in Table 4, and the measurement method was based on the measurement method of the 6th Korean Human Body Survey. At this time, measurements were made using the Maya 2013 3D numerical measurement tool.

Table 5 shows the comparison and analysis results of sizing related functions and avatar characteristics of the selected 3D virtual fitting program. Two programs were developed in Korea and one in Israel. All three programs were used in apparel companies that are engaged in export and import business mainly related to the trade business. One program is used not only in the apparel industry but also in the production of computer 3D companies, games, and movies.

As a result of analyzing the human body dimension setting and the functional difference of the program, there was a dimension conversion function which can change the size of the avatar in all three programs, but the landmark setting function for setting the dimension between each part was operated in only one program . In the DC Suite, researchers were able to set the desired landmarks directly on the avatar body. Although CLO 3D did not explain the names of landmarks or measurement points, it has a joint point view function among the avatar functions by a function similar to a landmark, and the avatar sizing reference point is indicated based on the joints of the body. 3D Runway does not have a function to recognize the landmark. When the avatar is set to the new dimension or the dimension name is clicked, the avatar body is displayed with the dimension along with the dimension, and the measurement point can be estimated on the screen at each dimension setting .

All three programs can create and use avatar for male and female. CLO 3D provides avatars for East, West, South and West, and 3D Runway for East and West, male and female avatars and children's avatars. Unlike the two programs, the DC Suite has the ability to change the avatar's detailed facial features such as facial expressions, hair styles, makeups, and accessories

As a result of examining the characteristics of each program, the basic structure is the same as that of manual configuration, method, and detailed functions, but virtual sewing is performed using clothing patterns and virtual description is realized using avatar. The development of various functions, realistic and more realistic avatars and clothes were under development.

The sizing system of the avatar is a function that transforms into a size other than the basic body model provided by the program. All of the programs can directly input the dimension for each part or change the size by using a bar-shaped menu called a controller. As a result of comparing and analyzing the avatar sizing path, command method, and sizing items of the virtual fitting program, the sizing system was accessed through 4 to 5 paths according to the program as shown in FIG. Different terms for accessing the same menu for each program are different. In order to spread and commercialize the program, it is necessary to standardize the terms through exchanges between companies and programs.

In terms of the path to the sizing controller screen and the terminology, CLO 3D preceded the process of loading the avatar size. To do this, go to the Window-Avatar Size Loader, load the pre-programmed avatar size, and enter the Window-Avatar Size Editor. Then, you can open the avatar size editor through the path where the avatar sizing editor appears. DC Suite and 3D Runway have no avatar size loader process, and avatar sizing conversion is possible through path of Window-Current body / Generated-Body Science / Generation and View / 3D Windows / Model / Model Properties respectively.

For avatar size design, all three programs basically enter basic dimensions first, then input detailed dimensions. CLO 3D enters the key dimensions and the waist circumference after the selection process of the avatar body type, and the remaining dimensions are autonomously adjusted. At this time, the avatar body types that can be selected are divided into four body types (tall and skinny, tall and fat, tall, skinny, and tall and fat). DC Suite has DC Model avatar and Generation avatars separated, so you can control each dimension in Generation avatars that can be avatar sizing. Also, when the Body Generation Controller is executed, there is a difference in that the appearance and color of the avatar are converted. Next, 3D Runway will show the controller divided into 6 panels (Adveance, Heights, Pose, Shape, Basic) in the window named Morph. At this time, the Basic panel is used for changing the basic dimensions and is used to specify most of the body sizes. The Adveance panel is used to fine-tune the body. The Pose and Shape panels can be used to change the position, posture, and detail of the model.

As a result of analyzing the avatar size conversion manual of each program, if the dimensions are inputted into the basic items of the avatar set in all of the three companies, the remaining details are configured to be automatically adjusted according to human proportions. As a result of analyzing the avatar sizing items, the avatar sizing items of the 3D Runway were the highest with 39 items, followed by the CLO 3D (28 items) and the DC Suite (22 items). In addition, after inputting the basic dimensions, detailed dimensions can be input individually or the controller can be adjusted. Of these, the basic items are 2, 4, and 8 items, and the rest are the detailed items. 3D Runway has 20 Adityance panels, 9 Heights panels, 6 Pose panels, 12 Shape panels, and 8 Basic panels in the Actual Avatar Body Dimension window. As a result, there were 55 total of all six panels. At this time, only 39 items were included in the study, except for the 16 items which were considered to be irrelevant to the size of the body because the pose and motion included sizing items. In addition, the ratio of the basic items to the total number of items is 18.2% (total number of items: 22, number of basic items: 4), and CLO 3D is 7.1% The number of items: 2), indicating that the accuracy of the automatic setting of the detailed items after input of the basic items requires a follow-up study to verify the human body dimension setting.

Table 6 shows the result of analyzing avatar sizing items for each program by basic items and detail items, and dividing the detailed items into height, length, circumference, and other items.

In all three programs, the adjustable part of the avatar body size, the name of each part, the number of adjustable basic items and the detailed items were different, but the same items set in common were 8 items in total, , Waist circumference, hip circumference, knee circumference, calf circumference, ankle circumference, and knee height. Among them, the key and waist circumference were found to be the basic items in all three programs. In detail, the number of perimeter items was the highest in all programs. Among them, CLO 3D has 17 most circumferential items among 28 items.

Since the avatar of the virtual fitting program has to reproduce the shape of the human body in a realistic manner, various items that the items of the sizing system can reflect the human body should be evenly distributed. However, as a result of analyzing the size conversion manual, it is pointed out that the measurement items do not show various distributions according to the measurement areas, and the use of different measurement items or terms for each program is a problem. In order to design an avatar that can reflect the human body type of Koreans, research on avatar sizing item setting that can clearly reflect the difference of avatar human body type or size by combining with existing human body size setting study, Research will be necessary.

One program showed the landmark directly in the program and showed the measurement form of the 3D avatar. In the other two programs, as shown in FIG. 7, a method of deriving the landmark from the viewpoint of the joint point and the measurement site is suggested. In the DC Suite, the location and name of the landmark can be confirmed through the path of Window / Body-Current Body / Generated-Body Science / Landmarks, and furthermore, a new reference point can be created or deleted. When you click the name of the landmark, the spot turns red and you can see the location and name.

Fig. 7 shows the comparison result of the program-specific landmarks with reference to the comparison of the 3 dimensional full automatic measurement reference point photograph data of the Ministry of Commerce, Industry and Energy (2010) and the avatar sizing reference point of Lee Minjung (2012). In the DC Suite, you can check the name and location of the landmark, but in the other two programs, you can check Show X-ray joints in CLO 3D, PDS / View / 3D / Model / Windows / Since it is possible to make analogy only through the analysis of the position of the measurement item through the Displa check, the exact position and name of the landmark could not be confirmed.

The avatar of the virtual fitting program has the main purpose of representing the size of the human body shape accurately through the sizing system, and it is the main object of the virtual dressing of the suit suitable for the fit. However, the landmark which is the standard of human body measurement, Can be pointed out as a problem.

In the Ministry of Commerce, Industry and Energy (2010), the human body measurement standard point for the human body measurement standard is 70 items, and only one measurement point is set based on the right side for the same item on the left and right sides. In addition to this, 15 items are divided into left and right, and the measuring point is set. The total measuring point is 85 items. DC Suite presents 47 items, and except for the 11 intermediate measurement points in the human body, the 36 items with the left and right are divided into 83 categories. (2004), the number of measurement points of the Ministry of Commerce, Industry and Energy (2004) measurement points for all parts except for legs and feet, as shown in Figure 8, Respectively.

The Ministry of Commerce, Industry and Energy's technical standardization institute established Korean human standard information composed of 185,000 human body dimensions, 20,000 kinds of dynamic dimensions and 120,000 kinds of 3D human body shape data, various processing data derived from them, And it is spreading to industry. Through these activities, we will develop and support unique Korean industrial products through the supply of industries such as clothing, shoes, furniture, kitchen appliances, home appliances and automobiles. In addition, And the product considering the coverage ratio (Ministry of Commerce, Industry and Energy, Korea Standard Technology Institute, 2010). Therefore, it is natural that a variety of measurement points, that is, landmarks, are used in comparison with a virtual avatar for costume production. In order to facilitate the use of the site and consumers in the beginning of the virtual fitting program, it is necessary to first analyze the landmarks of the avatar that are necessary for making clothes and describe them directly, It needs to be clarified. Next, it is suggested that the system of landmark proposed by the Ministry of Commerce, Industry and Energy's technical standards institute should be applied in order to apply the virtual fitting avatar utilization to various industries other than clothes and clothing concept.

The three virtual fitting programs are automatic size conversion systems in which other dimensions are automatically set when basic dimensions are input. The basic dimensions vary from 2 to 8, depending on the manufacturer. The results of analyzing the difference between the average size of 20 Korean adults and the rest of the dimensions are shown in Figure 9. The CLO 3D avatar with the smallest number of items in the base dimension has a chest circumference of 4.14 cm ), Hip circumference (2.19 cm), and knee height (2.66 cm), and the remaining 11 items, arms and leg circumference related items were small, and the avatar shape was relatively glaring and the legs were thin And it was found. In the case of DC Suite avatars, it was found that the knee and waist height were large and the femoral circumference was 14.24 cm in the leg related peripheries, which was the smallest among the program avatars, and the calf and knee items And the tendency to express the lower body as a slim ideal body shape. For the 3D Runway avatars, the appearance of the basic avatar was expressed as a white female, but the remaining items showed the smallest difference from the average size of the 20s.

In the case of DC Suite, the average circumference of the twenties was the average size of Korean women and 11.42 cm, indicating the problem. Analyzing the conversion dimension system of the size automatic conversion system, it is considered that the three programs are expressed inconsistently with different body images despite the input of the same basic dimensions, and the accuracy should be improved in the application of the avatar dimension of the virtual fitting program .

The Ministry of Commerce, Industry, and Energy (2010) presented the average and detailed size and shape of Korean women in their twenties on the basis of the 5th 3D image of Korean women. Based on this, the average size of 20 Korean adult women in each program was implemented as a 'virtual model of dimension input method' using the sizing process of each program as shown in FIGS. 10 and 11. In comparison with the naked eye, the avatar implemented through the virtual fitting program was expressed in tall and slender form compared with the average body shape of Korean adult women presented by the Korea Technology Standards Agency (2010).

For accurate numerical analysis, an image file is measured on AutoCAD using a plane measurement system, and measurements of height 5 items, width 4 items, and thickness 4 items are performed in the left, right, The measured dimensions were compared and analyzed as shown in FIG.

As a result of comparison and analysis of measured sizes of avatars by program, it was confirmed that the average body shape of 20s can not be expressed in the same or similar manner for each virtual fitting program. In the height items, the shoulders and hip items were higher in the three avatars. Among them, the shoulder items are large up to 7.8cm in the DC Suite, and even though the same key is set, the avatar's height is expressed more than the actual body type. In the width and thickness items related to the remaining circumference, the thickness items showed less avatars. The width items were generally the same or larger, but with the exception of the shoulder item, we could see that the avatar was small from 3.87cm up to 7.69cm. The shoulder-related factors were found to be a significant factor in distinguishing the type of body shape of Korean women in the Ministry of Commerce, Industry and Energy's (2010) standard. The expression power of the shoulder width was decreased, It can be pointed out as a problem. In the thickness item, it was found that the avatar is expressed thinly from a minimum of 2.7 cm to a maximum of 4.12 cm in the waist thickness compared to other items.

Analysis of the size difference of the avatar according to the virtual fitting program shows that CLO 3D has a smaller shoulder than the other programs and is expressed in a taller and lighter expression. In the DC Suite, 6 items of 14 items And showed the closest trend to the size of women. Based on these results, we created an outline based on the DC Suite avatars for areas and curves that can not be displayed in the image plane process of the 3D human body modeling process in the next research stage.

The Ministry of Commerce, Industry and Energy (KSDA) (2010) divides the 20 Korean adult female bodies into four types of body shapes, small reverse osmosis, large triangle, reverse osmosis, and square, and presents the average size of 119 items for each body type . Based on these data, the size of all basic and detailed items applicable to the sizing program of each avatar is written, and 4 Korean female representative body type avatars are implemented for each program. Then, the height of 5 avatars in the left, right, Item, width 4 items, and thickness 4 items were measured, and the measured dimensions of the avatars by the program were compared and analyzed. As a result, it was confirmed that each program can not express the four body types uniformly or similarly.

The measured width and thickness of avatars in the program were compared with those of Korean female figure. The results showed that the three avatars except for inverted triangles had higher avatars in chest, head in waist, hip in waist and hip in Table 7 and Fig. 12 as shown in Fig. As a result of examination by items, it was found that the degree of agreement with the 20th Korean adult female body presented by the Ministry of Commerce, Industry and Energy (KSAT) (2010) was the lowest among 4 items of height, width and thickness. The thickness of the chest was larger than that of the Korean female body (2.69 cm ~ 6.75 cm), the waist thickness (-3.75 ~ -6.99 cm), the hip thickness (-10.23 cm ~ -11.25 mm) cm < / RTI > The measurement of the plane which determines the perimeter of the three-dimensional perimeter is the thickness and the width. When the same circumference is entered, width and thickness are inversely proportional to each other, It can be judged that it is difficult to judge the accuracy of the peripheral input value.

As a result of examining the difference in implementation by program, the 3D Runway avatar measurement was represented by 5 items of reverse osmosis, 7 items of large triangle, 6 items of reverse osmosis, 6 items of rectangle, We confirmed that the best implementation of the program was successful. However, when compared with the naked eye, the degree of agreement with the 20th Korean adult female body presented by the Korea Institute of Technology and Technology (2010) showed the smallest difference, and there was a difference between the input size and the actual measurement item in the height entry item. This is because there is concern about the possibility of errors due to the measurement on the plane image, and it is understood that there is a limitation in comparing and judging the accuracy of size-related programs among companies.

As shown in Fig. 13, it was confirmed that the image types of the human body are not similar to each other even though the same human body type is implemented on a program-by-program basis. Especially, 3D Runway avatar without an avatar is larger than other avatar, so it can be seen that there is a problem to apply to the body shape of Korean adult female. The result of this study is that it is necessary to consider the error range according to the avatar posture because it is impossible to switch the avatar with the same pose in all programs because there is a limitation in adjusting the arm and leg opening angle when setting the human body measurement attitude.

The results of analyzing body type for adult human female 3D human body model group are as follows.

For this study, we divided the body type among 156 items of 3 dimensional human scan dimension of the 6th Korean human survey project. The average body shape of the 20 ~ And Table 9. As a result of analyzing the body shape by age group, the height items related to the key and the length items were larger in the 20s except for the front center length. The subjects in the 30s were larger in circumference, thickness, width, weight, and BMI, which were related to obesity. Among them, differences in waist circumference and abdominal circumference were 3.9cm and 4.01cm, respectively, Of the total. It can be explained that the subcutaneous fat is deposited in the abdomen as the female body shape is folded to 30, which is consistent with research on the female body type of the Agency for Technology and Standards (2004) of the Ministry of Commerce, Industry and Energy.

As such, there is a difference in body shape between 20s and 30s. However, due to the down-aging phenomenon in the fashion marketing market, existing brands targeting the early 20s and mid-30s tend to expand to the early 30s (Lee, In addition, avatars of sub-ages are difficult to apply to apparel companies and virtual fitting programs. In addition, the sizing system of the virtual fitting program is capable of sizing the avatar more closely to each age-specific body shape through sizing control after setting a representative body shape. Therefore, this study proposes avatars of young adult Korean women who integrate body shapes of 20s and 30s, which can be instantly utilized in the industry as a virtual model, and the average values and outline images of each size are shown in Tables 8 and 9 same.

For the purpose of designing the 3D avatar human body model group which represents Korean adult women, to identify the body constituent elements and to utilize them as the basic data of the cluster analysis for body type classification, the 3 - dimensional human body measurement The factor analysis was performed on 55 items.

As a result of the screening test, the number of factors was determined to be 7, which is the level at which the eigenvalue is 1 or more and the total dispersion is not greatly changed. Table 10 shows the items included in the factors and factors named in the study. The variance that can be explained by 7 factors is 81.28%, and the factors of 1 and 2 are 37.43% and 23.01%, respectively. In order to clarify the factor loadings of each variable, we tried the orthogonal rotation by the Varimax method and presented the loadings of each factor in Fig. 14 and Fig. The first factor is high item for 29 items such as weight, BMI, circumference, thickness, and width items, indicating the degree of obesity of the body. Among them, BMI (.949), body weight (.949), waist circumference (.913), breasts (.910), abdomen (.888), and hip circumference (. 882) showed high load values, indicating that they represent obesity.

The second factor belongs to height, and 13 items belong to it. All items showed a high load of .850 or more. Among them, the item with the highest load was the waist height (.970), and the longer the lower body was, the larger the overall height was.

The third factor was highly loaded on the two items, drop and left, indicating the female body flexion, which was .863 and .754, respectively.

The fourth factor was the shoulder length, the shoulder width, and the length between the shoulders and the back of the armpit, which were related to the shoulder skeleton. The total score was 3.703, .675, and .625, respectively. The eigen value of the four factors is 2.457, and the explanatory power of the four factors corresponds to 4.468% of the total, and the total variance is 70.533%, so that only 1, 2, 3, and 4 factors have 70% or more explanatory power over the total variance.

The fifth factor was the head vertical length, head circumference, head index, and head width. The two-sided index (key / head vertical length) shows a negative load (-.706), which is inversely related to the rest of the head factor. The eigenvalue of the five factors was 2.375, and the amount of variance was 4.318% and the total variance was 78.115%.

In the sixth factor, the factor of load of the shoulder was found to be .863, .800, and the fourth factor was not included as the same shoulder item or the same factor. It is analyzed as a special factor that determines the body type according to the angle of the shoulder rather than the factor related to the width. The eigenvalue of the sixth factor was 1.796, the variance was 3.265%, and the total variance was 78.115%.

The seventh factor can be analyzed as the upper center length of the anterior center length and the back length. The load of the factor was shown as the front center length (.812), the back length (.681), and the load of the front center length was found to be high. It can be analyzed that the front center length represents the upper half length, Since the vertical upper body length may also be influenced by the volume of the breast in the upper body of a woman, these factors must be considered differently from the male body type when analyzing the body shape. The eigenvalues of the seven factors are 1.741, the variance is 3.165%, and the total variance is 81.280%.

In order to classify body types into several homogeneous groups based on the similarity in human body data of 3 - dimensional adult women, 7 factors calculated from factor analysis were used as independent variables and 20 ~ 30 Korean adult women A total of 406 people were analyzed by cluster analysis.

Because there is no device for verifying the statistical significance of the final results compared to other analysis methods, the number of clusters is determined by analyzing one to five clusters in each cluster. It is desirable to select a cluster that is suitable for the classification of the cluster with a large difference between the clusters (Eun-Kyung Jeon, 1992).

Table 11 shows the results of testing the difference of factor scores between groups according to the number of groups. In this study, there was a significant difference in all types. However, in determining the body type of the human body, the two types are not enough as the result of analysis on the theoretical background, and it is preferable to classify them into three or more will be.

The results of analyzing the distribution of research subjects according to the number of clusters are shown in FIG. 16 as a method for determining the number of clusters. Among the three types except for the two types, there were 1 group (183, 45.6%), 2 groups (118, 29.4%) and 3 groups (100, 24.9%). Four groups were divided into 1 group (83 persons, 20.7%), 2 groups (183 persons, 34.4%), 3 groups (119 persons, 29.7%) and 4 groups (61 persons, 15.2%). In the 5 type classification group, there were 1 group (105 persons, 26.2%), 2 groups (42 persons, 10.5%), 3 groups (96 persons, 23.9%), 4 groups (90 persons, 22.4% Respectively. In conclusion, the larger the group, the smaller group will appear and the more detailed the group will be. However, as the size of the human body becomes farther away from the actual average group, the distribution ratio becomes low. Based on this, it is judged that classification of three or four body types, in which a small number of groups and a large number of groups are appropriately harmonized, The third processor is used to determine the number of groupings. Table 12 and Table 13 show the factor scores for each cluster.

In the results of the preceding factor analysis, the cumulative variance of factors 1 and 2 is more than 60%. In this way, when we typify the cluster, we try to name the type of human body easily by volume and height factors.

First, when divided into three clusters, it can be named Slim & Short, Slim & Tall, and Heavy & Little Tall. Next, divided into four clusters, they could be named as Full & Short, Slim & Short, Full & Tall, and Body 4 (Slim & Tall).

In this study, it is necessary to have a name that can easily distinguish the avatar body type between the consumer and the industry since the main purpose of the Korean representative avatar is to be instantly used in the 3D virtual fitting program. It is difficult for consumers to perceive the form in which the body shape of Korean adult women of the Ministry of Commerce, Industry and Energy (2004) is differentiated by each part of the human body. Therefore, it is aimed to draw the final research results with four clusters that clearly understand the clusters and can be easily understood.

In order to examine the characteristics of body type, the mean and ANOVA tests of the four items were conducted. The results are presented in Table 14 and Table 15, respectively. As a result of Duncan's test, a significant difference was found between the items in all 55 items, and the difference between the types was clear. . Particularly, among the factors that distinguish the body type, the volume items are larger and smaller than the average body size, and it can be understood that the body type of Full and Slim is classified into Tall and Short according to the size of the item being large and small. The characteristics of body type are as follows.

The first body type (Full & Short) shows the largest value among all four types in all volume items except for the hip width, and can be said to be the most obese type. In addition, height items were smaller than other types except for breasts height, and body shape was the least, thus reflecting the characteristics of full and short body, overall obese and tall. In the second body type (Slim & Short), the volume items showed the least amount of hip circumference except the knee circumference, the height items had a medium degree, and the shoulder skeleton factors showed the least. Therefore, it can be seen that the overall body shape is Slim and short compared to the average body shape and the body shape 3 and 4. On the contrary, body type 3 and body 4 showed a larger height than the average body type and 1 and 2 body type, respectively. However, in the volume factor, the third body type is B, The lowest value of C, and the third body type was named Full & Tall, and the four body type was named Slim & Tall. In addition, body flexion and shoulder skeleton were the largest in the third body type, and intermediate body type in the fourth body type. The result of presenting the image of each body type is as shown in FIG. 17, and the body shape characteristic can be seen as the outer silhouette.

In this study, we propose a process for human body modeling design which has excellent dimensional fit for the generation of representative avatar for female adult female body. The design process is shown in Fig. 18, and the detail order is as follows.

First, we analyzed the average size of the standard human body type by analyzing the body shape of Korean adult women. Based on this, AutoCAD 2010 program was used to design 3D, 2D, and 3D models for Top, Front and Side (right). At this time, the curves and human body parts which can not be expressed in exact size were imaged by the raster image of the DC Suite representative avatar, and the image plane design was carried out based on this. 3D human body modeling was performed in Maya 2013 using the next designed Top, Front, Side (right) 2D image plane. The size conformity evaluation of the 3D human body modeling is based on the Maya 2013 numerical analysis system, and after the evaluation of the 3D perimeter measurement, the modeling of each part is corrected for the large error range item, and the final modeling result is presented.

This 3D human body modeling process uses the AutoCAD design program with a small systematic error rate and implements the process of 3D modeling after preliminary designing of 2D human body plane surface suitable for the average body shape, so that error in human body modeling can be minimized . In addition, it has the advantage of reflecting ideal posture of adult woman by pre-designing three image planes of front, top and side. When modeling with Maya 2013, it is out of the existing skeleton model, and mapping is done according to the outline and dimensions, thereby minimizing the file capacity and improving the usability.

The 3D avatar design process will be described step by step as follows.

Step 1: Average dimension analysis

Using IBM SPSS Statistics Ver.21, we derive the body type of Korean women and analyze the average size of each body type. In this case, the average size includes a two-dimensional item such as width, thickness, length, and angle, and a circumference and a length item for modeling evaluation. Accurate statistical analysis for each body type must precede to reduce errors after modeling.

Step 2: Create an image plane

Using the AutoCAD 2012 program, design the image plane of Front, Top, and Side based on the plane size. The image plane step is divided into a base line setting step and an outline setting step. First, in the base line setting step, a basic line of width, height, and thickness of the 2D size is set, and a reference point essential for avatar production is set as shown in Fig.

 When setting the outline of the image plane, the standard posture is divided into posture and division as shown in FIG. 20 with reference to the Leonardo da Vinci type human body segmentation of Impeok Tack (2006). Next, using the DC Suite avatar sizing system to create the avatar shape through the avatar implementation of each body shape, the front and rear division based on the human body curve and the center line of the side line is performed. In the 2D image plane design in AutoCAD 2012, Using the image function, the final outline was set as shown in FIG. The shape of the Korean adult female figure of the utilized DC Suite is shown in FIG.

Step 3: Conduct 3D human body modeling

Using Maya 2013 program, external mapping is done based on image plane. The detailed procedure is as follows.

a. Run Image plane

Use the Maya 2013 program to model the outline image of human body pre-made with AutoCAD 2012, as shown in Figure 23.

b. Body modeling

After making a square cube, we use the edge segmentation tool such as insert edge loop tool, split polygon tool, and extrude along the outline of the image plane to set the body part of the human body. As shown in Fig. 24.

c. Limb modeling

Using the extrude in the body trunk, make the cube, use the insert edge loop tool to make the details of the limb, adjust the vertex to the body image plane, and model it as shown in Fig.

d. Detailed appearance modeling

In order to make the human body more detailed or add lines to the detail of the human body, modeling is performed as shown in Fig. 26 by expressing the curved surface characteristics of the human body to the utmost by using only the minimum line before using the smooth function.

e. Apply Smooth

To make the completed human body curve smoother, the smooth surfaces are used to curve the planarized surfaces as shown in FIG. 27 like the human body and the surface. After a smooth run, there are a lot of lines, so it is difficult to fine-tune the details. All of the detailed modeling should proceed.

f. Human modeling left complete

In human body modeling, only one side is designed based on the median of the human body and is symmetrical. Therefore, after one side is designed, the entire human body modeling state is confirmed, and the final correction operation is completed as shown in FIG.

Next, use the duplicate tool to mirror the pivot of the object and copy it to the other side.

g. Completed 3D human body modeling

The process of checking for completeness of the human body modeling. If there is a face whose nomal direction is inverted or where the vertexes are not merged, correct it. When the final check is completed, the history generated by modeling is deleted to reduce the capacity of the modeling data file, and then the modeling storage is performed as shown in FIG.

Step 4: Evaluate and complement 3D avatars

The designed 3D avatar is verified by the Maya 2013 3D numerical measurement system to check the fit of human body by selecting 9 items of circumference 4 items and 5 items of length 5 items which are judged as essential items for determining the size of an adult woman, . The detailed procedure is as follows.

a. Scale up to key

The avatar that is designed to be easy to model is enlarged like the actual key in the 3D virtual space as shown in FIG. 30, and the length of the key is measured by measuring the bottom point and the head end point using the distance tool.

b. Landmark settings

In order to verify the accuracy of the human body dimension, the locator tool is used for the measurement reference point (landmark) in the 3D avatar as shown in FIG.

c. Circumferential dimension measurement

: Cut the face straightly with the cut face tool based on the set landmark, select the edge to deform it with the curve, measure the circumference of the modified curve by using the arc length tool as shown in Fig. 32 .

d. Length measurement

Cut the face with the cut face tool based on the set measurement point (landmark), select the edge between the locator and locator, transform it into a curve, and measure the dimension using the arc length tool as shown in Fig.

e. Modify human body modeling

: After measuring the size, make sure that it matches the size of Korean adult female body. At this time, if the error range exceeds the range of ± 0.5 cm, execute the image plane again, re-check the shape of the image plane and the human body curve by using the soft selection tool, and modify the curved surface shape as shown in FIG.

f. Final assessment of human body modeling

: The dimensions of avatar's four circumference and five length-long avatars are revised and complemented, and 3D representative avatar of Korean adult female with excellent final fit is shown as FIG.

Fig. 36 shows the result of shaping 3D standard avatar human body model group according to body shape of Korean adult female. For each of the four body types, it can be seen that the characteristics of each body type are reflected as the result of the statistics.

First, Full & Short is the largest shape among all four types except for hip width, and it can be said that it is obese body shape. In addition, the heights of the breasts were smaller than those of the other breasts except for breasts height, and body flexion was the least, reflecting the characteristics of a full & short body that is obese and taller than the average body.

The second body type (Slim & Short) was the least in the volumetric group except the hip circumference, the knee circumference, and middle in the height category. The overall body shape was Slim, and the height items were shorter than the average body shape and 3, 4 body shape. The shoulder skeleton factor was the smallest, and the shoulders were narrow and the shoulder inclination angle was also less.

The third and fourth body types (Full & Tall) and 4 were significantly higher than the first and second body types in height, respectively. However, the third body type in volume was Slim compared to the first body type And it is the largest among the four body parts of the shoulder, the body bend, and the body.

The fourth body type reflects the human body characteristics of Slim & Tall, which is the body shape that is less weight than the height and is the thinnest among the four body types.

Among the virtual fitting programs, CLO 3D can select the avatar body type before converting the avatar size. The avatar body type provided by the program is shown in FIG. In comparison with the avatar body type of the CLO 3D, the appearance and size of the 3D standard avatar human body model group according to the Korean adult female body were compared with the avatar body type of the CLO 3D, which is considered to reflect a more realistic size and appearance .

 In order to verify the dimensional suitability of 3D standard avatars of Korean adult women designed by this study, the size of adult women is determined by referring to the dimensions of KS K 0051 adult womens out of the perimeter and length dimension items excluded from the Image Plane Fig. 38 shows the results of analyzing the dimensions of four items of circumference, five items of length and nine items of which the items are considered necessary.

Analysis of the circumference of the research avatar using Maya 2013 numerical analysis tool revealed that the range of error was ± 0.5cm and the dimensional suitability of the avatar was verified.

Among the 3D virtual fitting programs, DC Suite has Maya to DC function and can import 3D avatar created by Maya program as a virtual model. In this study, body import was attempted in the DC Suite to verify whether the 3D research avatar can be used as a virtual fitting model. The result screen is shown in FIG.

Claims (9)

(A) statistically analyzing an average dimension including a two-dimensional item and a three-dimensional item according to a person's body type;
(B) constructing a front, an upper, and a side image plane of a person using the two-dimensional average dimension;
(C) performing 3D human body modeling using the image planes; And
(D) verifying the human body dimensional fitness using the three-dimensional item analyzed through the step (A) with respect to the 3D avatar modeled through the step (C), and supplementing the size of the 3D human body modeling; A method of generating 3D representative avatars by body type of a person who includes.
The method according to claim 1,
The step (A)
Dimensional avatar generation method according to the sixth aspect of the present invention is a method for statistically analyzing an average size including a two-dimensional item and a three-dimensional item according to a body shape of a Korean person using a sixth person's position survey data.
3. The method of claim 2,
The body shape of the Korean,
Wherein the 3D representative avatar is divided into four categories according to statistical techniques of human body type analysis.
The method according to claim 1,
In the step (A)
Wherein the two-dimensional item includes at least one of width, thickness, length, and angle,
Wherein the three-dimensional item includes at least one of a circumference and a length for modeling evaluation.
5. The method of claim 4,
Wherein the three-dimensional item includes a body length, a hip circumference, a waist circumference, a perimeter circumference, a shoulder length, an anterior center length, a back length, How to create a star 3D representative avatar.
The method according to claim 1,
The step (B)
(B-1) setting a base line of width, height, and thickness using the average dimension as the two-dimensional item;
(B-2) setting a reference point essential for avatar production on the base line; And
And (B-3) setting an outline based on the base line and the base point information.
The method according to claim 1,
The step (C)
(C-1) modeling the torso and limb using image planes;
(C-2) modeling a curved surface characteristic of the human body;
(C-3) performing left human body modeling through the steps (C-1) and (C-2); And
(C-4) a step of performing 3D human body modeling by symmetrically copying the left human body modeling; and generating a 3D representative avatar for each human body type.
The method according to claim 1,
The step (D)
(D-1) scale-up the 3D avatar according to an actual key;
(D-2) setting a measurement reference point (landmark) for verifying human body dimensional conformity of the 3D avatar;
(D-3) measuring perimeter and length dimensions based on the measurement reference point (landmark); And
(D-4) checking the image plane and the shape information of the human body curve to compensate for the size of the 3D human body modeling when the measurement result obtained through the step (D-3) is out of the tolerance range How to create a 3D representative avatar for each human body type.
(A) statistical analysis of the average dimensions including the two-dimensional item and the three-dimensional item of the four body types of the Korean adult woman using the 6th Korean position survey data;
(B) constructing a front, an upper, and a side image plane of a person using the two-dimensional average dimension;
(C) performing 3D human body modeling using the image planes; And
(D) verifying the human body dimensional fitness using the three-dimensional item analyzed through the step (A) with respect to the 3D avatar modeled through the step (C), and supplementing the size of the 3D human body modeling; Including,
In the step (A)
The two-
Base point items including neck point, neck point, neck back point, shoulder point, shoulder point, groin point, axillary point, folding point on axillary point, nipple point;
Height including neck height, neck height, neck height, shoulder height, armpit height, breasts height, breasts height, waist height, navel level waist height, hip height, Item;
Length items, including arm length, hip vertical length, length between armpit folds, horizontal length between nipples, and head vertical length;
Width items including neck width, shoulder width, breasts width, breasts width, waist width, navel level waist width, hips width, head width;
Thickness items, including armpit thickness, breast thickness, breast thickness, waist thickness, navel level waist thickness, butt thickness; And
The left shoulder includes an angular item including a side inclination,
Wherein the three-dimensional item includes at least one selected from among a breasts, a hip circumference, a waist circumference, a perimeter circumference, a shoulder length, an anterior center length, a back length, A method of generating 3D representative avatars by body type.

Images