ABSTRACT

Phase retrieval plays a significant role in holographic imaging, optical metrology, and in other areas. There are various methods available to obtain satisfactory results for the general case, but the performance is often degraded when the data are grossly disturbed by outliers. To solve the problem with outliers, a phase retrieval algorithm based on median momentum reweighted amplitude flow (median-MRAF) is presented. The median-MRAF algorithm uses the median truncation rule to weaken the influence of arbitrary outliers in the initialization and gradient update of each iteration and then adopts the momentum parameter in gradient descent to accelerate the convergence rate. Simulation experiments demonstrate that the proposed algorithm can correctly and reliably detect the signals corrupted by outliers from square amplitude measurements; moreover, it can achieve higher success rate and tolerate larger proportion of outliers than others.

摘要

相位检索在全息成像，光学计量学和其他领域中发挥着重要作用。对于一般情况，可以使用各种方法来获得令人满意的结果，但是当数据受到异常值的严重干扰时，性能通常会降低。为了解决离群值问题，提出了一种基于中值动量加权幅度流的中间值检索算法（median-MRAF）。中值-MRAF算法使用中值截断规则来减弱任意离群值在每次迭代的初始化和梯度更新中的影响，然后在梯度下降中采用动量参数来加快收敛速度​​。仿真实验表明，该算法能够正确，可靠地检测出方波幅度测量中被异常值破坏的信号。而且，