Interactive, Freeform Editing of Large-Scale, Multiresolution Level Set Models

由于医学和科学成像技术的进步，产生了越来越多的体数据集(标量值的3D数组)，产生了必须处理、查看和分析的大量原始3D数据。在计算机图形学中，几种激光扫描重建算法生成体积、隐式表示。电影中的许多高级特效利用物理模拟来产生计算机生成的洪水和流体流动。这些影响的计算流体动力学(CFD)计算通常在规则的3D网格上执行，并产生动态体数据集。由于体数据集具有多个来源，因此非常需要强大的隐式模型编辑功能。为了满足这一需求，PI正在开发用于大规模、多分辨率水平集模型的交互式自由编辑的技术和算法。水平集模型结合了低级体积表示、可变形隐式曲面的数学和稳健的数值技术，提供了一种强大的几何建模方法。本研究正在开发实现大量水平集建模能力所需的数学、算法和技术。这些功能包括：自由形式操作、拓扑控制、大规模表示、多分辨率编辑和交互响应时间。基于各种用户输入数据的水平集速度函数以及曲面特征和特性为自由形式编辑操作符提供了基础。大规模水平集曲面的高效表示(DT-GRID)的最新进展被用来表示详细的高分辨率物体。利用分层、多分辨率的体数据结构以及用于分解、重建和层间运动转移的相关算法。基于点的曲面的算法正在被扩展，以提供拓扑控制、局部参数化和交互显示。

Given the advances in imaging technology for medicine and science, an increasing number of volume datasets (3D arrays of scalar values) are being generated, producing an overwhelming flood of raw 3D data that must be processed, viewed and analyzed. In computer graphics several laser scan reconstruction algorithms generate volumetric, implicit representations. Many of the advanced special effects in movies utilize physical simulation to produce computer-generated floods and fluid flows. The computational fluid dynamic (CFD) calculations for these effects are usually performed on a regular 3D grid and produce dynamic volume datasets. Given these multiple sources of volumetric datasets, there is a great need for powerful implicit model editing capabilities. To address this need the PIs are developing techniques and algorithms for interactive freeform editing of large-scale, multiresolution level set models. Level set models combine a low-level volumetric representation, the mathematics of deformable implicit surfaces, and robust numerical techniques to provide a powerful approach to geometric modeling.This research is developing the mathematics, algorithms and techniques needed to implement numerous level set modeling capabilities. Those capabilities include: freeform manipulation, topology control, large-scale representations, multiresolution editing, and interactive response times. Level set speed functions based on a variety of user input data, as well as surface features and properties provide the foundation for the freeform editing operators. Recent advances in efficient representations for large-scale level set surfaces (DT-Grids) are employed for representing detailed high resolution objects. A hierarchical, multiresolution volumetric data structure is utilized along with associated algorithms for decomposition, reconstruction and inter-level movement transfer. Algorithms from point-based surfaces are being extended to provide topology control, local parameterizations and interactive display.