Kernel Methods for Interferometric Imaging

TL;DR

KRISP is a new kernel-based interferometric imaging algorithm that reconstructs high-fidelity images from sparse data without prior training, effectively handling various array configurations.

Contribution

The paper introduces KRISP, a statistically robust, data-driven interferometric imaging method that does not require prior images or tuning, improving image reconstruction from sparse arrays.

Findings

Reconstructs Fourier maps up to maximum baseline length.

Works efficiently with sparse array configurations.

Produces high-fidelity images even with significant structure.

Abstract

Increasing the angular resolution of an interferometric array requires placing its elements at large separations. This often leads to sparse coverage and introduces challenges to reconstructing images from interferometric data. We introduce a new interferometric imaging algorithm, KRISP, that is based on kernel methods, is statistically robust, and is agnostic to the underlying image. The algorithm reconstructs the complete Fourier map up to the maximum observed baseline length based entirely on the data without tuning by a user or training on prior images and reproduces images with high fidelity. KRISP works efficiently for many sparse array configurations even in the presence of significant image structure as long as the typical baseline separation is comparable to or less than the correlation length of the Fourier map, which is inversely proportional to the size of the target image.