Stability of Scattering Decoder For Nonlinear Diffractive Imaging

TL;DR

This paper evaluates the robustness of the Scattering Decoder, a deep learning method for nonlinear diffractive imaging, demonstrating its stable performance across various imaging conditions.

Contribution

It provides a comprehensive analysis of ScaDec's robustness to different imaging parameters, confirming its reliability beyond initial performance claims.

Findings

ScaDec maintains high reconstruction quality across different permittivity contrasts.

Performance remains stable with varying numbers of transmissions.

Robustness is confirmed under different input signal-to-noise ratios.

Abstract

The problem of image reconstruction under multiple light scattering is usually formulated as a regularized non-convex optimization. A deep learning architecture, Scattering Decoder (ScaDec), was recently proposed to solve this problem in a purely data-driven fashion. The proposed method was shown to substantially outperform optimization-based baselines and achieve state-of-the-art results. In this paper, we thoroughly test the robustness of ScaDec to different permittivity contrasts, number of transmissions, and input signal-to-noise ratios. The results on high-fidelity simulated datasets show that the performance of ScaDec is stable in different settings.