arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 1702.05092math.NAcs.NA
收藏

相位恢复作为正则化问题

Phase-Retrieval as a Regularization Problem

Brazilian Synchrotron Light Laboratory/CNPEM(巴西同步辐射光实验室/CNPEM);Federal University of Humaitá, AM, Brazil(乌迈塔联邦大学);ICMC/University of São Paulo/São Carlos, Brazil(圣保罗大学伊拉斯谟斯·卡洛斯研究所)

Eduardo X. Miqueles, Nathaly L. Archilha, Marcelo R. Dos Anjos, Harry Westfahl, Elias S. Helou

AI总结:

本文提出将相位恢复算法与优化策略联系起来,以数值确定样品的物理参数,解决盲选择参数导致的错误问题。

摘要:

近期研究表明,样品的相位恢复成像可以建模为一个简单的卷积过程。有时,这种卷积依赖于样品的物理参数,这些参数难以提前估计。在这种情况下,盲目的参数选择通常会导致错误结果,例如在后处理图像分割过程中。本文提出了一种简单的相位恢复算法与优化策略之间的联系,这使我们能够数值确定物理参数。

英文摘要:

It was recently shown that the phase retrieval imaging of a sample can be modeled as a simple convolution process. Sometimes, such a convolution depends on physical parameters of the sample which are difficult to estimate a priori. In this case, a blind choice for those parameters usually lead to wrong results, e.g., in posterior image segmentation processing. In this manuscript, we propose a simple connection between phase-retrieval algorithms and optimization strategies, which lead us to ways of numerically determining the physical parameters

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。

↑