Individualizing women’s suit patterns using body measurements from two-dimensional images

Abstract

The goal of this study was to develop an approach that could automatically generate the customized patterns for women’s suits based on the body measurements taken from two-dimensional (2D) frontal and side images of a subject. The 26 important pattern dimensions relevant to certain body dimensions were first chosen, and the mapping relationships between the body and pattern dimensions were then established for pattern alterations. For the body dimensions (e.g. girths) that could not be directly measured in the 2D images, prediction models were created based on the available width and depth measurements. The body measurements from the 2D images (auto-measurements) of 295 subjects were compared with the corresponding manual measurements, which showed a good correlation between the auto and manual measurements. The try-on test of five suits made with the altered patterns demonstrated the good fitting effects of the customized suits at important characteristic landmarks of five participating subjects through a visual evaluation. The subjective test also showed a satisfactory result of clothing fit under five different postures. Since this pattern-making method is originated from the relationship between the features of a human body and the elements of a pattern prototype, the generated patterns are individualized by unique body shapes to attain a good fit. This method can also accelerate the pattern-making process, reducing human efforts, costs, and production time.