Nonlinear Restoration of Diffused Images via Seeded Instability

We demonstrate a nonlinear method for restoring the quality of diffused images. It works by introducing a self-focusing medium into the optical path and allowing the underlying correlations to grow as they propagate. More specifically, the method is a dynamical stochastic resonance in which a weak signal seeds an instability in the diffuse background. The resonance is determined by the transition point between competing instabilities, with optimal growth occurring when the spatial scale of the instability matches that of the object of interest. The results are presented within the general framework of nonlinear statistical optics and have implications for information theory, basic wave physics, and nonlinear system design.