1. Introduction
Each measurement method has its advantages and disadvantages, and the parameters after the measurements will be used in different fields in the garment industry. In Vietnam, having a sizing system table, which was establishment in 2009. This size chart table has been 12 years old by now (1). For men’s sizes, there are a total of 26 sizes, of which 14 are for boys and 12 for adult men. According to a research (2), fuzzy clustering data mining may be used to create a size system using the anthropometric data of girls between the ages of 20 and 30. Every form of apparel comes in a variety of sizes, such as the shirtdress in Technical Reference (3), which includes 15 sizes and 5 main measurements. Technical Reference JML 3,247 women’s pants (4) have 11 sizes and 4 primary dimensions, whereas Technical Reference CERVO pants (5) have 11 sizes and 5 primary dimensions. The size chart of Novelty shirt has 10 sizes and 4 primary dimensions. Novelty trousers has 9 sizes and 5 primary dimensions. The Sanding shirt has 9 sizes and 5 primary dimensions. The symbol for each size number has four information about height, bust, waist circumference, buttock circumference, and each height group have three size group: A, B, and C. The new sizing standards in the world also show that there are many sizes for each group of objects, such as ISO 8959-3: 2017 with 16 sizes for three kinds of body groups (6). In recent years, there have been many studies on the establishment of the size chart, such as in the study (7). The authors measured a sample of 500 men ranging from18–35 years old with 30 measurement parameters. The size chart for the Vietnam People’s Army includes 52 sizes (8). In Japan, JIS L4004: 1997 men’s clothing sizing system, has 10 body shapes (9). Another study correlates the classification of the body shapes which is the size chart developed in Korea after scanning on a 3D human body scanner. There are four body types, in which the body shape is classified based on the only drop of bust circumference (10, 11). The authors of a research (12) developed a sizing method for 7,800 kids, aged 6 to 18, who were separated into two age groups for the investigation of body forms. The authors of the study on (13) created an 11-size women’s size chart with the major measurements of breast girth and waist girth. The sizing system tables of each country were different. For example, the American size chart has 43 dimensions (14), the UK size chart has 20 dimensions (15), and the Australian size chart has 17 dimensions. It shows that each sizing system table has different sizes, so choosing the right size for your body shape will take a long time, causing fatigue as well as damage to the product because of full testing. Until now, choosing the size of ready-to-wear clothes was still based on the parameters printed on the product packaging. On this basis, the research of an algorithm to extract the size of the costume to choose the fit size in a short time from the size chart table is very urgent in the field of clothing trading and is in line with the development trend of 4.0 today.
2. Material and methodology
2.1. Material
There are five contents for this study. Firstly, choose the primary dimensions in the sizing system table. Secondly, create a simulation model for selecting the fit size. Thirdly, establish a simulation model of selecting the fit size. Fourthly, test of the simulation model’s results. Finally, draw the flowchart to extract the size.
2.2. Methodology
The study uses the fuzzy logic method, which is used in the design of the algorithm to extract the size. There is one output and three inputs in this model. Fuzzy sets serve as the foundation for fuzzy logic’s rules. In this study, a triangular fuzzy set is a particular kind of fuzzy set that is utilized for input variables. The Simulink simulation method is uses in the design of the sizes extraction model. The experimental method is applied in checking the results of size extraction through customers who buy new clothes and who are wearing these clothes.
3. Results and discussions
3.1. Choosing primary dimensions in the sizing system table
The research database is extracted from the technical document T-Shirt brand f. house of Phuong Dong Garment Joint Stock Company. This is an oversize T-Shirt (Figure 1). The measurement parameters (Table 1) have been coded and the measurement positions are presented according to table 5.24. This table has 4 sizes and the size labels are signed by S, M, L, and XL. From the dimensions of the table, it shows that there are 2 primary horizontal dimensions (the weight dimension and the bust dimension). Vertical primary dimensions are not available for this product group. Therefore, an additional vertical primary dimension is required; the body height is chosen based on standards (16–19). The height dimension comes from the size chart (1), and it is divided three groups. Group 1 arranges from 150 to 160 cm, group 2 arranges from 160 to 170 cm, and the group 3 arranges 170–179 cm.
Figure 1. T-Shirt’s measurement dimension positions.
Table 1. The size chart of the T- Shirt for f. house branding.
3.2. Designing the algorithm to extract the size ready-to-wear
The size chart table has 3 different height groups, each of which has 4 sizes (Table 2).
Table 2. The code table of the T- Shirt for f. house branding.
The MISO model was selected for the fuzzy controller in this study. In there, the input has three variables (body height, weight, and chest circumference), one output (size to find), and is passed through the Fuzzy Logic Controller (Figure 2). The size chart has three heights corresponding to four sizes. The first input variable has three membership functions, the second input variable has four membership functions, and the third input variable has four membership functions corresponding (Figure 3). All three use triangular fuzzy sets and parameter intervals for each membership function of each variable as shown in Table 3.
Figure 2. The fuzzy logic model for extracting the T-Shirt size.
Figure 3. The graph of membership functions for input variables: (A) The first input, (B) The second input, and (C) The third input.
Table 3. Measurement arrange of sizes for 2 output variables.
The output is the size of a lookup table. There are four sizes in total in the size chart, so there will be four membership functions for the output variable (Table 4).
Table 4. The value of the output for the size to look for.
The matrix of fuzzy logic rules is built based on the “max-min” inference method (Table 5). This fuzzy logic model has 48 rules. In there, there are 16 rules for size S, 8 rules for size M, 12 rules for size L, and 12 rules for size XL.
Table 5. The matrix of fuzzy logic rules.
The structure to extract the size is shown in Figure 4. There are three inputs, each of which has three membership functions for input 1, four for input 2, and four for input 3. According to the rule, one input membership function will link to one output membership function, producing one output that is the desired size.
Figure 4. The Anfis Model Structure.
To see fuzzy logic rules in the space, choose Surface in the menu View show in Figure 5.
Figure 5. Testing controller rules in Surface.
3.3. Establishing the Simulink simulation model
The model has three inputs (the height, the weight, and the bust measurement) and one input (the size) as shown in Figure 6. The height’s measurement range is 145–185 cm. The weight’s measurement range is 52–93 kg. The bust’s measurement range is 72–100 cm.
Figure 6. The Simulink simulation for T-Shirt-f. House.
3.4. Testing the result to extract the T-Shirt’s size
Testing is done in two ways. The first way is to take the correct measurements from the size chart and input them into the model, see the results after running, and then compare them with the size in the table (Table 6).
Table 6. The result of testing for extracting the size by the first method.
The results show that they are true with the coding size in the table. The second way, testing the objectivity by two methods of extracting the size: the traditional method and the fuzzy logic method by inputting measurements of 3 dimensions of height, weight, and bust. In there, 7 men are using T-Shirts (f. house) and 3 people are trying on new products. The fit of products is evaluated scientifically according to the analysis of product appearance when T-Shirts are worn on 3 customers (Figure 7). The images show in Figure 7 have the following serial number in Table 7. The sample (a-XL) is sample 10. These people’s primary dimensions have similar to the primary dimensions of the size XL in the size chart. The sample (b-L) is sample 9. Primary dimensions have similar to the primary dimensions of the size L in the size chart. The sample (c-S) is sample 8. Primary dimensions have similar to the primary dimensions of the size S in the size chart. Besides that, it is evaluated through the comments of 10 people about the fit by 5 levels (very tight, tight, medium, wide, and very broad) and is analyzed with the results of algorithm simulation.
Figure 7. Images try on T-Shirt, f. house branding: size L (a), size M (b), and size S (c).
Table 7. The result of testing for extracting sizes of the t-shirt for f-house branding.
3.5. The flowchart for extracting the ready-to-wear
The flowchart to extract the ready-to-wear size needs three input variables (height, weight, and bust dimensions). The three variables must be in the range of values in the size chart, then the program shows the output of the fit size (Figure 8).
Figure 8. The flowchart to extract the ready-to-wear size.
4. Conclusion
The study presents the results of the algorithm to extract the size of ready-to-wear T-Shirt with f. house branding from the sizing system table and give suggestions on choosing the fit size for the customer. Each of the four sizes on this table includes three distinct metrics for body height. The three most important parameters for choosing input variables for the fuzzy simulation model are waist circumference, bust, and body height. This outcome demonstrates the viability of using fuzzy logic to select the fit size. We may use the experimental measurement data to determine a good size for males based on the fuzzy. In addition, the study analyzed the experimental results in each research area as well as discuss the experimental data of the study. Measurements of height, weight, and bust are selected as the three input variables of the fuzzy model. The output variable is the required size. The study expanded the size selection range when extracting the size and body shape according to the fuzzy logic method and created a basis for businesses to easily calculate the number of sizes chosen by customers to balance production. Furthermore, it can be applied to other fields in garment technology as well.
References
2. Hsu C, Lee T, Kuo H. Mining the body features to develop sizing systems to improve business logistics and marketing using fuzzy clustering data mining. Wseas Trans Comput. (2009) 8:1215–24.
6. ISO 8559-3:2017.Size designation of clothes — Part 3: Anthropometric definitions for body measurement. Switzerland: ISO (2017).
7. Nga B. Researching and application of 3D measuring equipment in building a standard pattern for men’s pants and shirts. Hanoi: Institute of Textile Garment (2011).
8. Chau NTH. Researching of establishing sizing system table of the military by an anthropometric method. Hanoi: Institute of Textile Garment (2001).
10. KS 0050.Sizing Systems for Men’s and Boy’s Garments. Seoul: Korean Standards Association (2001).
12. Chung M, Lin H, Wang M. The development of sizing system for school students. Int J Ind Ergonom. (2007) 37:707–16.
13. Gupta D, Gangadhar BR. A statistical model for development body size charts for garment. Int J Cloth Sci Technol. (2004) 16:458–69.
16. ISO: 4415.Size designation of clothes-Men’s and boys’ underwear, nightwear and shirts. Switzerland: ISO (1981).
17. ISO/TC 133.Clothing sizing systems – size designation, size measurement methods and digital fittings. Switzerland: ISO (1991).
18. ISO 8559-1.Size designation of clothes-Part 1: Anthropometric definitions for body measurement. Switzerland: ISO (2017).