Comparison of Phase Retrieval Methods in Computer Experiments
Comparison of Phase Retrieval Methods in Computer Experiments
复制标题
计算机实验中相位检索方法的比较
DOI：
10.1143/jjap.27.244
复制
发表时间：
1988
影响因子：
1.5
通讯作者：
N. Nakajima
中科院分区：
文献类型：
--
作者：
N. Nakajima
本文对三相检索方法的数值性能进行了比较。这些方法就是 Fienup 算法。 Walker 算法从迭代方法解决相位检索问题，以及从非迭代方法使用对数希尔伯特变换的方法。在二维目标函数重建的计算机模拟研究中，结果表明非迭代方法在每种情况下都会产生唯一的解；然而，在某些情况下，这些迭代算法在伪解上停滞不前。特别是，Fienup 算法的停滞现象比 Walker 算法更频繁地发生。
This paper presents a comparison of the numerical performances of three phase retrieval methods. These methods are Fienup's algorithm. Walker's algorithm from the iterative approach to the phase retrieval problem, and a method using the logarithmic Hilbert transform from the noniterative approach. In computer simulation studies for the reconstruction of two-dimensional object functions, it is shown that a noniterative method produces a unique solution in each case; however, these iterative algorithms stagnate on pseudo-solutions in some cases. Especially, the stagnation of Fienup's algorithm happens more frequently than with Walker's algorithm.