Abstract

Reconstruction of a coherent (complex-valued) image of a general object from a single snapshot of Fourier (far field) laser intensity data is very difficult with existing algorithms. For the special case of an object having one or more glints (delta functions), a set of three algorithms, used in succession, has been demonstrated to be capable of reconstructing a high fidelity image. The first algorithm employs the component of the autocorrelation function (computed from the Fourier intensity) of the image due to the glints. It reconstructs the glints (complex values and positions) using a triple-intersection approach. The second algorithm is the AF synthesis (Baldwin-Warner) algorithm which, starting with the reconstructed glints and using the Fourier intensity data, reconstructs the remainder of the image, but with some residual artifacts. The third algorithm is the iterative Fourier transform algorithm which completes the reconstruction using both the Fourier intensity data and a support constraint (derived from the autocorrelation function). Performance of the approach is quantified as a function of glint strength and photon levels.

© 1989 Optical Society of America