Image Recovery Through Scattering Media via GAN Reconstruction and SNES Optimization

Abstract

Optical image recovery through scattering media is a significant yet challenging problem. Iterative wavefront shaping is one of the powerful tools to re-distribute the diffusive light and compensate for the diffuser by controlling the incident wavefront. However, in the scenario that only a feedback signal on the camera can be obtained, this technology would fail due to the lack of target images. In this paper, we propose a new scheme for recovering images through scattering media in an absence of target images. In particular, we employ an improved Generative Adversarial Network (GAN) for computational reconstruction and separable natural evolution strategy (SNES) for wavefront shaping optimization. Both simulation and experimental results suggest that the proposed scheme will open up new opportunities in the applications of biomedical imaging, optical encryption, holographic display, etc.