Single-shot lensless imaging based on iterative denoising phase retrieval

Lensless imaging is a promising technique with a large field of view and high throughput imaging capacity. Limited by its imaging modulations, it suffers from twin image problems due to missing phases. Hence, it requires phase retrieval to recover the phase information as well as suppress the artifact. As the original algorithm, the Gerchberg-Saxton algorithm (GS) retrieves the phase data from at least two intensity measurements. It has low convergence and is sensitive to the initial guess. In this work, we apply denoising in the object field with a guided filter into the iterative framework of the GS algorithm named Guided filtering GS (GFGS). The approach in this paper requires only a single hologram measurement. To suppress artifact noise, guided filtering is performed on the real and imaginary parts of the complex-valued distribution with an appropriate smoothing parameter at the object field.