| Online shopping of clothes is becoming popular. However, it is hard for customers to choose among clothes due to various sizes and styles, which leads to a high return rate. In order to display the appearance of customers wearing clothes of different styles and sizes, we present a system that captures the human body size and reconstructs 3D human models with Kinect. In addition, the system allows customers to combine different garment components freely.In this thesis, we present an novel method to realize a 3D garment tailored system, which focuses on two aspects:1) 3D human body scanning and reconstruction with Kinect, personalized design and combination of cloth components. Firstly, we extract human body features automatically from depth and skeleton data captured by Kinect, and match the human models from the database of human body templates with the help of these features, and then deform the model to fit the captured data.2) in order to achieve personalized combination of clothing components, our system can provide the following functions:loading Drawing Exchange Files (DXF), adjusting the location of 2D patterns, setting seam lines and so on. We adopt a method for 2D mesh triangulation based on the Delaunay algorithm to get regular and unique triangular meshes. We use a method for clothes simulation based on Position Based Dynamics (PBD) to get stable and visually plausible cloth simulation results. We solve the dis-matching of triangular meshes and sewing lines to get combined clothes. At last, we propose a method to calculate the degree of matching for combined clothes and design a component storage database for fast combination. |
PDF Full Text Request