Pants design and pattern generation based on body images

Abstract

The personalized pattern generation method based on 2D body-measuring technology has considerable application potential in clothing e-commerce, remote clothing customization, clothing production, and other aspects. By inputting the front and side body images, this study proposed a new method of generating personalized patterns automatically. The silhouettes could be extracted from the body images to estimate body sizes and design style. The basic rules between the patterns and the body sizes were analysed, and the rules of the general pattern generation were established through a knowledge-based combination of the basic pattern and the style parameters. The sizes extracted from images were compared with the manually measured values, and the errors of these sizes were analysed. Sample pants were made and tried on with the automatic pattern generation system, and the experiments showed that the sample pants have a good fit at some key landmarks. As a result, this system can automatically generate personalized patterns and style designs based on 2D human body images, to improve garment fit and accelerate clothing customization.