Computational Imaging using Ultrasonically-Sculpted Virtual Lenses

Ultrasonically-sculpted waveguides provide interesting opportunities for in situ optical imaging in transparent and scattering media. The interference of ultrasonic waves can be designed to form a spatially-varying refractive index pattern in the target medium, acting as a virtual lens to guide light and relay images. The images formed by such lenses are subject to a large amount of spatially-varying blur, which significantly reduces their contrast. To alleviate this issue, the images can be computationally deblurred post experiment to restore the image. First, we demonstrate a brute force deconvolution technique to deblur the relayed images. While effective, this method proves to be computationally intensive due to the spatially-varying blur kernel. The reconfigurability of ultrasonically sculpted waveguides can be leveraged to address this issue by measuring line integrals of the image at multiple angles to form the Radon image of the target, which can be deblurred very efficiently with a simple linear model. We validate this method using simulated and experimental results. The tantalizing notion of using the reconfigurable ultrasonically-sculpted waveguides as part of the imaging systems, demonstrated in this paper, opens new opportunities for hybrid physical-computational optical systems.