Improved Image Compressive Sensing Recovery with Low-Rank Prior and Deep Image Prior

Yumo Wu, Jianing Sun, Wengu Chen, Junping Yin

Abstract: Highlights•We propose a novel patch-based prior, namely non-convex low-rank (NCLR) prior established from weighted Schatten p-norm. By plugging FFDNet-based deep prior as a composite part, we provide a new PnP model for image CS recovery.•We present an efficient algorithm to solve the proposed model by using alternating direction method of multipliers.•We apply the proposed model to image CS recovery, and improve the recovery effects. In addition, the proposed model can get better results with fewer iterations.