Neural Volumetric Reconstruction for Coherent Synthetic Aperture Sonar

Abstract

Synthetic aperture sonar (SAS) measures a scene from multiple views in order to increase the resolution of reconstructed imagery. Image reconstruction methods for SAS coherently combine measurements to focus acoustic energy onto the scene. However, image formation is typically under-constrained due to a limited number of measurements and bandlimited hardware, which limits the capabilities of existing reconstruction methods. To help meet these challenges, we design an analysis-by-synthesis optimization that leverages recent advances in neural rendering to perform coherent SAS imaging. Our optimization enables us to incorporate physics-based constraints and scene priors into the image formation process. We validate our method on simulation and experimental results captured in both air and water. We demonstrate both quantitatively and qualitatively that our method typically produces superior reconstructions than existing approaches. We share code and data for reproducibility.

Aperture Synthesis With Physics-Informed Neural Networks

“…Recently, implicit neural representations have been proposed for volumetric scattering field reconstruction in forward-looking sonar 11 and SAS imaging 12 . In these studies, a neural network is trained to infer the scatterers' distribution as a continuous function of spatial coordinates.…”

“…Specifically, the optimization objective involves finding the scatterers' distribution that, when forward propagated through a physical model, best fits the recorded data under some constraints such as scatterer sparsity and surface continuity. However, the neural backprojection is practically applicable for small-scale imagery and depends on the compromise between accuracy and efficient computational implementation of the forward propagation model 12 . Another group of studies has used PINNs for sound field reconstruction from a limited number of recordings of the room impulse response collected from linear 13 and planar microphone arrays 14 .…”