arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 1601.04699cond-mat.mtrl-sciphysics.data-an
收藏

体界面网络的拓扑忠实非参数估计与跟踪

Topology-faithful nonparametric estimation and tracking of bulk interface networks

Siddharth Maddali, Shlomo Ta'asan, Robert M. Suter

摘要:

本文的主要重点是一个非参数滤波技术的散装材料从现代成像测量可获得的界面几何形状的估计。过滤方法依赖于一个假设的层次结构的拓扑特征存在于一个典型的界面网络,如泡沫界面和晶界网络多晶材料。每种类型的拓扑特征在层次中按等级顺序处理,较低级别的特征根据较高级别特征施加的位置约束进行过滤。这样的计划是一个替代现有的表面平滑/估计技术在微观结构材料科学,其中明确的处理网络拓扑结构的不同元素是不存在的,或在最好的任意参数化。我们描述了这种技术在通常的微观结构的应用,其中重要的物理量的计算是基于精确估计的接口功能的后果。作为一个额外的应用程序,我们描述了一种新的前端跟踪算法,用于量化这些接口的传输。

英文摘要:

The main focus of this paper is a nonparametric filtering technique for the estimation of interface geometry in bulk materials obtainable from modern imaging measurements. The filtering methodology relies on an assumed hierarchy of topological features present in a typical interface network, such as foam interfaces and grain boundary networks in polycrystalline materials. Each type of topological feature is treated in order of rank in the hierarchy, with the lower-level feature being filtered subject to the positional constraints imposed by the higher-level features. Such a scheme is an alternative to existing surface smoothing/estimation techniques in microstructural materials science, in which the explicit treatment of different elements of the network topology is absent, or at best arbitrarily parameterized. We describe the ramifications of this technique in the usual microstructural applications in which the computation of important physical quantities is predicated on the precise estimation of the interface features. As an additional application, we describe a novel front-tracking algorithm for quantifying the transport of such interfaces.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑