Second-order differential based matching pursuit method for compressive sensing signal recovery

摘要

In this paper, we propose a novel sparsity adaptive greedy algorithm, called second-order differential based adaptive matching pursuit (SDAMP) method, for sparse signal reconstruction. By applying a novel sparsity sensing technique based on the second-order differential of correlation vector Φ y, SDAMP is capable to achieve real-time recovery without K as a prior condition. Further, SDAMP innovatively incorporates a backtracking framework to extend the algorithm to an adaptive greedy strategy. Experiments demonstrate that SDAMP runs considerably faster than other sparsity adaptive algorithms while maintaining superior performance to most of its predecessors.