基于粒子群优化和原子特性的匹配追踪算法

Matching Pursuit Based on PSO and Atomic Property

摘要: 由于信号稀疏表示的优良特性，已被用于信号处理很多领域，但计算量大阻碍了它在实际中的应用。粒子群优化算法简单,易于实现,且搜索效果好。论文采用匹配追踪(Matching Pursuit, MP)算法实现信号稀疏分解，利用粒子群优化算法搜索MP过程中的最优原子。根据原子特性，优化改进后的算法。仿真结果证明了新算法的可行性。

Abstract: As sparse representation of signals has excellent characteristics, it has been applied in several fields of signal processing. But it has a large scale of computing, which hinders its application in practical signal processing. Particle swarm optimization is simple to be realized, and the searching result is good. In this paper, Matching Pursuit is used to realize sparse representation of signals, and particle swarm optimization is used to effectively search the best atom in the process of MP. According to the property of atoms, the improved algorithm is optimized. At last, the simulation results demonstrate the feasibility of the new algorithm.