【作者】 马志刚； 汪国平； 董士海；

【Author】 Zhigang Ma Guoping Wang Shihai Dong (School of Electronics Engineering and Computer Science, Peking University,Beijing,100871)

【机构】 北京大学信息科学技术学院；

【摘要】 通过三维扫描设备获取的数字几何模型通常包含海量数据,对模型的存储、传输、处理都构成了挑战。本文提出了一种新的紧凑且易于处理的模型表示方法:法向插值偏移曲面。该方法将模型表示为一个简单的基曲面以及分布在基曲面上的标量偏移场。基曲面表达了模型的大致形状,而偏移场则描述了模型的几何细节。本文还设计了几何模型由三角网格表示转化为NIDS表示的转化算法,其核心在于找到一个合适的基曲面,使得原模型上的几何细节可以用该基曲面上的偏移量表示。利用基曲面上法向量分布的特点,在基模型的生成过程中建立其与原模型之间的一一映射,从而实现了上述特性。试验表明该算法精确而高效。更多还原

【Abstract】 We can accurately acquire highly detailed and dense sampled geometry models with millions,even billions of vertices. The huge data amount issue challenges on the models’ storage, transmission and processing.In this paper we introduce a new type of model representation,the normal interpolation displaced surfaces,which decomposes a geometry model into a simple domain mesh and a scalar-valued displacement field over it Then we present a fast scheme that converts geometry models from triangle mesh representation to NIDS.The challenge in this conversion process is to find an appropriate domain surface that still faithfully expresses the details of model as its offset This is accomplished by maintaining a one to one map between the original mesh and the current simplified mesh while obtaining the domain surface via a simplification process.This model representation is proven to be compact and easy to process by lots of examples.更多还原