基于非正交分解的频率估计算法

高分辨率频率估计是信号处理许多领域研究的热点,并有着广泛应用.通过引入非正交分解思想,提出了一种基于稀疏分解的高分辨率频率估计新算法.由于原子库的冗余性,稀疏分解得到信号更加灵活的表示.算法中稀疏分解利用Matehing Pursuit(MP)算法实现,由分解原子的参数,可得到频率估计.与传统基于DFT频率估计方法相比,新算法可以获得更高的频率分辨率.尤其在低信噪比环境下,算法有明显优势,并且新算法在欠采样情况下仍然适用.理论分析及计算机仿真结果证实了算法的有效性.

Frequency estimation algorithm based on non-orthogonal decomposition

A new algorithm is proposed to achieve high-resolution frequency estimation which is intensively studied and has been widely applied to many areas in signal processing.In this paper,the idea of sparse decomposition,which is a non-orthogonal decomposition of signals,is introduced into high-resolution frequency estimation study.Since the choice of dictionaries is not limited,the sparse decomposition provides extremely flexible signal representations.The sparse decomposition is implemented by matching pursuit(MP)in the proposed algorithm.Frequency estimation can be then achieved according to the parameters of the decomposition atoms.The new algorithm can obtain higher frequency estimation resolution than most frequency estimation methods based on DFT,especially in the circumstance of lower signal to noise ratio.It is also applicable in the circumstance of under sampling.Theoretic analysis as well as experimental results illustrates the properties of the proposed algorithm.