Generator Surgery for Compressed Sensing
This addresses the limitation of representation error in compressed sensing for image recovery, offering an incremental improvement for applications in medical imaging or surveillance.
The paper tackles the problem of high representation error in compressed sensing using deep generative models as signal priors by removing initial blocks of a pre-trained generator at test time to optimize over a higher-dimensional latent space, resulting in significantly improved reconstruction quality that is competitive with state-of-the-art methods.
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.