مقاله انگلیسی رایگان در مورد بازیابی فاز بر اساس مدولاسیون تقسیم کد شده – وایلی 2017

A new coded splitting imaging technique is proposed to reconstruct the complex amplitude of a light field iteratively using a single-shot measurement. In this technique, a specially designed coded splitting plate is adopted to diffract the illuminating beam into multiple beams of different orders and code their wavefronts independently and differently. From the diffraction pattern array recorded on the detector plane, both the modulus and phase distributions of the illuminating beam can be reconstructed iteratively using known transmission functions of different orders of the coded splitting plate. The feasibility of the proposed technique is verified both numerically and experimentally.

Introduction

Coherent diffraction imaging (CDI) method is a lensless imaging technique that can measure the phase and modulus distributions of a light field from the intensity of recorded diffraction patterns using an iterative approach. Since it does not require high-quality optics and can obtain diffraction limited spatial resolution theoretically, CDI can be adopted to observe a sample with a broad range of radiations including visible light, high-energy electron, and X-ray photons and has become an independent tool in many research fields (Fienup, 1993; Kostli ¨ and Beard, 2003; Zuo et al., 2003; Roy et al., 2011). The first widely accepted CDI technique is the G–S algorithm proposed by Gerchberg and Saxton (Gerchberg, 1972). Moreover, it was further developed using Fienup algorithms and an oversampling method (Fienup, 1982; Miao et al., 2003) and it can retrieve complex amplitude with a single frame of diffraction pattern intensity. Although these CDI algorithms based on the G–S algorithm demonstrated significant achievements in imaging samples with short wavelengths including X-ray and electron beam, they also suffer from problems such as low convergence speed, limited field of view, and low reliability, especially in imaging large objects with complex phase distributions (Fienup and Wackerman, 1986; Fienup, 1987). Many modified CDI algorithms have been proposed to overcome these drawbacks, and they can be generally divided into two categories. The first group, termed as multishot CDI, involves using a multiwavelength source for illumination (Bao et al., 2008), scanning the diffraction field axially (Ivanov et al., 1992; Allen and Oxley, 2001; Almoro et al., 2006), scanning the sample transversely (Faulkner and Rodenburg, 2004; Maiden and Rodenburg, 2009), and scanning the illuminating direction (Ou et al., 2013; Bian et al., 2014; Dong et al., 2015). Significant data redundancy in these algorithms offers several remarkable advantages over conventional CDI methods, including significant improvement in robustness to noise, no requirement for prior knowledge of the sample, faster convergence speed, and more reliable reconstruction. However, these methods require significant data acquisition time, precluding their application for imaging fast dynamics. Furthermore, since these methods rely heavily on the stability of the imaging system, both vibration and degeneration of the sample may result in failure of the experiments. Note that even tiny imprecisions in the scanning steps can degrade the reconstructed resolution. The second group, termed as singleshot CDI, can realize real-time measurement of a light field. The recently developed coherent modulation imaging method uses a random phase plate to modulate the light field, and it adopts a spatial constraint at the entrance plane which is usually the focal plane (Zhang and Rodenburg, 2010; Tao et al., 2016). However, since only one frame of diffraction patterns is applied, the reconstruction is noisy when compared to that of multishot CDI algorithms. In order to obtain high-quality reconstruction with single-shot measurement, a feasible approach is to record a diffraction pattern array with a single exposure of detector to increase data redundancy, and based on this idea, two kinds of single-shot ptychography methods have been demonstrated (Pan et al., 2013; Sidorenko and Cohen, 2016).