Synthetic aperture imaging by distributed arrays of space telescopes

We propose and numerically investigate a new methodology for long-range short-wavelength synthetic aperture imaging. Utilizing low-resolution intensity images captured by a distributed array of sparsely fly-scanning detectors, we reconstruct high-resolution images. We introduce and explore two reconstruction implementations: one leveraging Fourier ptychography and the other employing an untrained deep neural network to replace the update step in the iterative algorithm. These methods are compared to small-aperture low-resolution images and conventional Fourier ptychography reconstructions. The proposed techniques yield superior reconstructions in scenarios where traditional methodologies may be inadequate, with the neural network demonstrating generally superior performances over the ptychography-based approach. The proposed method exhibits four principal advancements: enhancement of images via sparsely sampled synthetic aperture, fly-scan sampling, better resistance to noise, and serialization and scalability—whereby additional samples can incrementally refine the solution without necessitating resolving for the entire dataset for each new sample.