Figure 4: Numerical demonstration of sparsity-based super-resolved 1D CDI of objects consisting of rectangles with only approximately known widths (allowing 20% variations).
From: Sparsity-based super-resolved coherent diffraction imaging of one-dimensional objects
(a) The ‘original’ 1D object which consists of five rectangular functions with widths (left to right) 2.4, 1.6, 2.4, 2 and 1.6 μm. (b) Power spectrum of the original object. (c) Truncated power spectrum used to simulate the measured data, with 35 dB noise added. (d) The blurred reconstruction calculated by inverse Fourier transform of the ‘measured’ power spectrum presented in c assuming complete knowledge of the spectral phase. (e–g) Sparsity-based reconstruction (dashed red) compared with the original object (solid blue). The reconstruction uses the ‘measured’ power spectrum (of c) and performed with GESPAR in the basis of shifted rectangles with a fixed width of 2 μm. (h–j) Sparsity-based reconstruction (dashed red) compared with the original object (solid blue). The reconstruction uses the same ‘measured’ power spectrum (of c) but performed with a modified GESPAR algorithm implemented in a basis of shifted bars with variable widths (see Supplementary Note 3).
