项目名称: 多类型数据驱动的智能形状建模
项目编号: No.61303136
项目类型: 青年科学基金项目
立项/批准年度: 2014
项目学科: 自动化技术、计算机技术
项目作者: 陈翔
作者单位: 浙江大学
项目金额: 22万元
中文摘要: 几何形状建模技术在影视业、视频游戏、艺术设计、工程设计与仿真、产品设计与制造、医学、建筑学、地质学等许多方面都起到了不可估量的作用。然而现有的几何形状建模方法对用户具有较高的要求,往往需要艺术家和设计者的深厚专业知识和大量时间投入。因此,提升几何形状建模的智能性和易用性对于业界和普通民众而言具有十分重要的意义。本项目拟基于最新的几何形状集合的分析建模研究,结合机器学习和计算机视觉等技术,使用二维图像、三维形状等多类型数据源作为驱动,并以简洁的用户交互方式作为引导,最终实现高效的智能几何形状建模。
中文关键词: 智能形状建模;数据驱动;机器学习;模型降价;渐进数值方法
英文摘要: Shape modeling plays a quite important role in a variety of areas, such as movie&television, video games, art design, engineering design and simulation, product deisgn and manufacturing, medicine, architecture and geology. However, existing shape modeling approaches have many unnecessary requirements of users, which need deep knowledge of design specialty and plenty of time. Therefore, it is significant to improve the intelligence and ease of use of shape modeling, for both industry and people. In this project, we leverage on the lastest research on shape set analysis and modeling, and integrate machine learning and computer vision techniques to make shape modeling intelligent and effective, which is driven by multitype data (e.g. 2D images and 3D shapes) and guided by concise user interactions.
英文关键词: intelligent shape modeling;data driven;machine learning;model reduction;asymptotic numerical method