Stability of Scattering Decoder For Nonlinear Diffractive Imaging

Published in Proc. iTWIST, 2018

Recommended citation: Yu Sun and Ulugbek Kamilov. "Stability of Scattering Decoder For Nonlinear Diffractive Imaging."(Oral) iTWIST 2018.


The problem of image reconstruction under multiple light scattering is usually formulated as a regularized nonconvex optimization. A deep learning architecture, Scattering Decoder (ScaDec), was recently proposed ‘Efficent and accurate inversion of multiple scattering with deep learning’ 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 the ScaDec is stable in different settings.

Download paper here


Y. Sun and U. S. Kamilov, “Stability of Scattering Decoder For Nonlinear Diffractive Imaging,” Proc. 4th International Traveling Workshop on Interactions between Sparse models and Technology (iTWIST 2018) (Marseille, France, November 21-23), in press.