高真实感、个性化的三维虚拟试衣是虚拟现实三维内容生成的技术前沿，也是制约服装类电子商务发展的瓶颈问题。项目以数据驱动的个性化对象建模与仿真为切入点，致力突破基于集成学习的个性化人体三维点云语义特征点自动提取、数据驱动的个性化三维人体和服装几何模型自动生成、基于位置动力学的虚拟三维试衣过程实时交互仿真等关键问题，建立个性化三维虚拟试衣场景高效建模与实时仿真的技术框架。预期可形成涵盖不同性别、体型、常见服装材质类型的人体与衣物三维模型数据库(模型数量不少于500个)，实现可支持个性化三维人体即时构建(不超过30秒)与真实感实时动态试衣仿真(30帧每秒以上)的在线式三维虚拟试衣原型系统。本项目将为电子商务和信息消费产业的发展提供技术支撑。

3D virtual try-on with high real sense and personality is a pioneering topic leading the development of 3D content generation technology of Virtual Reality, and it’s also the bottleneck hindering the development of clothing e-commence. With automatic semantic feature extraction as the starting point, this project aims to solve the following key problems such as semantic feature extraction from personalized 3D body point cloud, automatic personalized 3D body and garment building based on data driven technology and real time interactive simulation based on position-based dynamic. This project is to build a 3D virtual try-on prototype system with 3D body and garment model database (not less than 500 3D models) which would identify different gender, body type and popular cloth materials to support for real time personalized body building within 30s and interactive dynamic simulation of virtual dressing with 30 fps or above. This project is expected to provide technology support for the development of e-commerce and information consumption industry.