Chen Shichao, Li Chengshuai, Zhu Yizheng
Opt Lett. 2017 Mar 15;42(6):1088-1091. doi: 10.1364/OL.42.001088.
Sensitivity is a critical figure of merit to quantify measurement performance in quantitative phase imaging. It is affected by various noise sources in the system and by signal processing algorithms. Here we propose a three-level framework for sensitivity evaluation, including the Cramér-Rao bound (CRB), algorithmic sensitivity, and experimental sensitivity. Comparing the first two determines the theoretical efficiency of an algorithm, while inspecting the gap between the latter two reveals system efficiency. As an example, we apply this framework to wavelength shifting interferometry, an important category of quantitative phase imaging techniques. In a shot-noise-limited regime, the CRB is derived, and the performance of a four-step Carré algorithm is studied in simulations and experiments. Importantly, the proposed procedure allows the algorithmic sensitivity to be conveniently estimated from a single set of measurement data, which serves as a basis for system efficiency evaluation.
灵敏度是定量相位成像中量化测量性能的关键品质因数。它受到系统中各种噪声源以及信号处理算法的影响。在此，我们提出了一个用于灵敏度评估的三级框架，包括克拉美 - 罗界（CRB）、算法灵敏度和实验灵敏度。比较前两者可确定算法的理论效率，而检查后两者之间的差距则可揭示系统效率。作为示例，我们将此框架应用于波长移位干涉测量法，这是定量相位成像技术的一个重要类别。在散粒噪声受限的情况下，推导了克拉美 - 罗界，并在模拟和实验中研究了四步卡雷算法的性能。重要的是，所提出的方法允许从单组测量数据方便地估计算法灵敏度，这为系统效率评估奠定了基础。
