From compression to compressed sensing

Abstract

Can compression algorithms be employed for recovering signals from their underdetermined set of linear measurements? Addressing this question is the first step towards applying compression algorithms for compressed sensing (CS). In this paper, we consider a family of compression algorithms ${\mathcal{C}}_r$, parametrized by rate $r$, for a compact class of signals $\mathcal{Q} \subset \mathds{R}^n$. The set of natural images and JPEG at different rates are examples of $\mathcal{Q}$ and ${\mathcal{C}}_r$, respectively. We establish a connection between the rate-distortion performance of ${\mathcal{C}}_r$, and the number of linear measurements required for successful recovery in CS. We then propose compressible signal pursuit (CSP) algorithm and prove that, with high probability, it accurately and robustly recovers signals from an underdetermined set of linear measurements. We also explore the performance of CSP in the recovery of infinite dimensional signals.

References in corpus (6)

• Compressive Imaging using Approximate Message Passing and a Markov-Tree Prior
• Block-length dependent thresholds in block-sparse compressed sensing
• Asymptotic Analysis of Complex LASSO via Complex Approximate Message Passing (CAMP)
• Recovery from Linear Measurements with Complexity-Matching Universal Signal Estimation
• Minimum Complexity Pursuit: Stability Analysis
• Minimum Complexity Pursuit for Universal Compressed Sensing