Cyclic pure greedy algorithms for recovering compressively sampled sparse signals

摘要: The pure greedy algorithms matching pursuit (MP) and complementary MP (CompMP) are extremely computationally simple, but can perform poorly in solving the linear inverse problems posed by recovery of compressively sampled sparse signals. We show that applying a cyclic minimization principle, performance both significantly improved while remaining simple. Our simulations CompMP may not be competitive with state-of-the-art algorithms, their variations are. discuss ways which complexity further reduced, our these hurt performance. Finally, we derive exact condition algorithms.