Improving object density determination in coherent diffraction imaging by preprocessing truncated discrete diffraction amplitudes

Abstract

We investigated the consequences of discrete Fourier transformation in coherent diffraction imaging (CDI). The object density reconstructed from the discretely sampled diffraction data within a truncated range is inherently aliased, blurred, and further aggravated in phase retrieval process. We devised a preprocessing procedure to correct input Fourier constraints using a convolution kernel and to exclude erroneous Fourier constraints. By applying the proposed preprocessing to both simulated and experimental data, we demonstrated that image reconstruction was substantially improved, effectively suppressing physically unsound fluctuations in the retrieved images. This procedure could improve the fidelity of the quantitative object density retrieved by CDI.