Generator Surgery for Compressed Sensing

TL;DR

This paper proposes a novel method for image reconstruction in compressed sensing by modifying pre-trained generators at test time to reduce representation error, leading to improved recovery quality.

Contribution

It introduces a test-time generator modification technique that enhances the performance of deep generative priors in compressed sensing tasks.

Findings

Significantly improved reconstruction quality across various architectures

Effective for out-of-distribution images

Competitive with state-of-the-art methods

Abstract

Image recovery from compressive measurements requires a signal prior for the images being reconstructed. Recent work has explored the use of deep generative models with low latent dimension as signal priors for such problems. However, their recovery performance is limited by high representation error. We introduce a method for achieving low representation error using generators as signal priors. Using a pre-trained generator, we remove one or more initial blocks at test time and optimize over the new, higher-dimensional latent space to recover a target image. Experiments demonstrate significantly improved reconstruction quality for a variety of network architectures. This approach also works well for out-of-training-distribution images and is competitive with other state-of-the-art methods. Our experiments show that test-time architectural modifications can greatly improve the recovery quality of generator signal priors for compressed sensing.