An intelligent recommendation system for personalised parametric garment patterns by integrating designer's knowledge and 3D body measurements
Cheng Chi,Xianyi Zeng,Pascal Bruniaux,Guillaume Tartare,Cheng ChiXianyi ZengPascal BruniauxGuillaume Tartarea Wuhan Textile University,Wuhan,Chinab Ecole Nationale Superieure des Arts et Industries Textiles,GEMTEX Laboratory,Roubaix,Francec Wuhan Textile and Apparel Digital Engineering Technology Research Center,Wuhan,Hubei,China
DOI: https://doi.org/10.1080/00140139.2024.2332772
IF: 2.5612
2024-03-29
Ergonomics
Abstract:Garment pattern-making is one of the most important parts of the apparel industry. However, traditional pattern-making is an experience-based work, very time-consuming and ignores the body shape difference. This paper proposes a parametric design method for garment pattern based on body dimensions acquired from a body scanner and body features (body feature points and three segmented body part shape classification) identified by designers according to their professional knowledge. By using this method, we construct a men's shirt pattern recommendation system oriented to personalised fit. The system consists of two databases and three models. The two databases include a relational database (Database I) and a personalised basic pattern (PBP) database (Database II). The Database I is based on manual and three-dimensional (3D) measurements of human bodies by using designer's knowledge. And Database I is a relational database, which is organised in terms of the relational model of the body part shape and its key body feature dimensions. After a deep analysis of measured data, the irrelevant measured dimensions to human body shape have been excluded by designers and extract representative human body feature dimensions. In addition, the relations between body shapes and previously identified body feature dimensions have been modelled. From the above relational model, we label key feature point positions on the corresponding 3D body model obtained from 3D body scanning and correct the whole 3D human upper body model into the semantically interpretable one. The 3D personalised basic pattern is drawn on the corrected model based on these key feature points. By using three-dimensional to two-dimensional (3D-to-2D) flattening technology, a 2D flatten graph of the 3D personalised basic pattern of the interpretable model is obtained and slightly adjusted to the form suitable for industrial production, i.e., PBP and the PBP database (Database II) is built. In addition, the three models include a basic pattern parametric model (Model I) (characterizing the relations between the basic pattern and its key influencing human dimensions (chest girth and back length)), a regression model (Model II) which enables to infer from basic pattern to PBP for three body parts based on the one-to-one correspondence of key points between the PBPs and the basic patterns and a personalised shirt pattern parametric model (Model III) (characterizing the structural relations between the personalised shirt pattern (PBPshirt) and PBP). The initial input items of the recommendation system are the body dimension constraint parameters, including chest girth, back length and the body feature dimensions used to determine each body part shape as well as three shirt style constraint parameters (slim, regular and loose). By using Model I, the corresponding basic pattern can be generated through the user's chest girth and back length. Body feature dimensions determine the three body parts' shapes. Then, Model II is used to generate the PBP for the corresponding body parts shape. Based on the shirt style chosen by the user, Mode III is used to generate the PBPshirt from the PBP. The output of the recommendation system is a fit-oriented PBPshirt. Moreover, if the PBPshirt is unsatisfactory after a virtual try-on, four adjustable parameters (front side-seam dart, back side-seam dart, waist dart and garment bodice length) are designed to adjust the PBPshirt generated by the proposed recommendation system. The proposed recommendation system combines the designer's knowledge of manual measurement of the human body, traditional 2D pattern-making methods and 3D-to-2D flattening technology to generate personalised shirt patterns automatically and quickly, thus significantly improving pattern-making efficiency. The reliance on designers in the garment production process is reduced. Even users with no pattern-making knowledge can also develop professional shirt patterns by using our proposed system.
engineering, industrial,ergonomics,psychology, applied