More Efficient, Adaptive and Stable, A Virtual Fitting System Using Kinect
| Autor: | Chang-Tai Xiong, Ruo-Yu Yang, Shun-Lei Tang |
|---|---|
| Rok vydání: | 2018 |
| Předmět: |
Computer science
business.industry Constraint (computer-aided design) 020207 software engineering 02 engineering and technology Human body Parallelism (philosophy) Virtual fitting Robustness (computer science) 0202 electrical engineering electronic engineering information engineering Human body modeling 020201 artificial intelligence & image processing Collision detection Computer vision Artificial intelligence business Gesture |
| Zdroj: | Advances in Multimedia Information Processing – PCM 2017 ISBN: 9783319773827 PCM (2) |
| DOI: | 10.1007/978-3-319-77383-4_18 |
| Popis: | Under the trend of online shopping, requirements for virtual fitting continue to increase. However, there still lacks a mature system that can display realistic garment effects in real-time. In this paper we use Microsoft Kinect to gather user data, and based on the real-time human body modeling technique designed by us, the human body model that adapts to different gestures and body sizes are generated. With suitable clothing models (4–6k vertices on average) for the target human body, we use Position Based Dynamics (PBD) constraints to model cloth physical properties and utilize GPU based parallelism method to accelerate constraint resolving and collision detection. As a result, our system can provide realistic effects for the virtual fitting while meeting the real-time and robustness requirements. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|