海量数据驱动的高维材质外观建模方法

项目名称： 海量数据驱动的高维材质外观建模方法

项目编号： No.61303135

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： 吴鸿智

作者单位： 浙江大学

项目金额： 22万元

中文摘要： 高维材质外观是计算机图形学领域近年来一个重要的研究方向。它从本质上刻画了光与材料之间的交互作用，因此越来越多地被运用到真实感绘制中，给电影和游戏中的物体赋予了逼真的材质效果。然而目前对真实世界材料的外观进行数字化建模是一件十分耗时耗力的任务，面向普通用户的低成本快速高维材质外观建模仍然是一个有待探索的领域。受互联网上相关海量数据不断快速增长的启发，本项目以构建一个海量数据驱动的、包含常用自然和人造材料的大规模高维材质外观数据库为目标，拟对互联网上材质外观相关数据的高效采集处理、基于海量图片的高维材质外观建模方法、真实材料外观空间的理论分析和大规模材质外观数据库的相关应用这几个关键科学问题开展研究，并提出海量数据驱动的基于最大后验概率估计的高维材质建模核心算法。本项目拟根本上解决阻碍真实材质外观实用化的障碍，将有力推进高维材质外观在工业界中的大规模应用。

中文关键词： 外观建模；双向反射分布函数；数字材质外观；；

英文摘要： High-dimensional material appearance is an important research topic in computer graphics in recent years. Material appearance fundamentally characterizes light / material interactions. Thus, high-dimensional material appearance is more and more widely used in photorealistic rendering, in order to add realism to objects/characters in feature films and computer games. However, it is tedious and costly to model material appearance from the real world, even with the help of state-of-the-art techniques. Therefore, more effort is needed in this area to provide average users with low-cost, rapid high-dimensional material appearance modeling techniques. Inspired by the fast-growing visual data on the Internet, we would like to build a large-scale, high-dimensional material appearance database, which includes a wide variety of both natural and man-made materials. We will focus on the following key research problems, including efficient collection and preprocessing of visual data on the Internet, high-dimensional material appearance modeling method from a large number of tagged photographs, theoretical analysis of real material appearance space and related applications for the large-scale material appearance database. We propose a novel large-data-driven, max-a-posteriori-based high-dimensional material appearace modeling

英文关键词： appearance modeling；BRDF；digital material appearance；；