A Markov random field approach for modeling spatio-temporal evolution of microstructures
A Markov random field approach for modeling spatio-temporal evolution of microstructures
复制标题
用于建模微观结构时空演化的马尔可夫随机场方法
DOI：
10.1088/0965-0393/24/7/075005
复制
发表时间：
2016
影响因子：
1.8
通讯作者：
V. Sundararaghavan
中科院分区：
文献类型：
--
作者：
P. Acar;V. Sundararaghavan
解决了以下问题：在较小区域测量的实验电影的情况下，能否合成大范围内的微结构演变？我们的输入是一个小样本窗口中微结构演变的电影。开发了一种马尔可夫随机场(MRF)算法，该算法使用这些数据来估计更大区域内的微观结构的演变。与基于静止图像的标准微结构重建问题不同，该算法还能够重建晶粒长大等随时间演化的现象。这样的算法将通过将数学估计与目标小规模时空测量相结合来降低全面微结构测量的成本。为了验证该方法的有效性，将合成的多晶组织在不同时间的晶粒度、形状和取向分布统计数据与原始电影进行了比较。
The following problem is addressed: ‘Can one synthesize microstructure evolution over a large area given experimental movies measured over smaller regions?’ Our input is a movie of microstructure evolution over a small sample window. A Markov random field (MRF) algorithm is developed that uses this data to estimate the evolution of microstructure over a larger region. Unlike the standard microstructure reconstruction problem based on stationary images, the present algorithm is also able to reconstruct time-evolving phenomena such as grain growth. Such an algorithm would decrease the cost of full-scale microstructure measurements by coupling mathematical estimation with targeted small-scale spatiotemporal measurements. The grain size, shape and orientation distribution statistics of synthesized polycrystalline microstructures at different times are compared with the original movie to verify the method.