论文部分内容阅读
本文论述白色噪声淹没下有限个二维正弦信号的谱估计问题。导出了自回归(AR)予测误差滤波器法的谱估计封闭公式,同时也导出了BARTLETT窗周期图及最大似然比谱估计的封闭公式,这些公式在研究分辨频率很接近的正弦信号方面是很有用的。在一个窄频域上,直接抽取可以用来改善分辨率同时降低运算量,仿真实验结果证明:在一个(N_1,N_2)的支持上,作(D_1,D_2)的抽取,其结果和未经抽取但支持扩大为(D_1N_1,D_2N_2)所得结果具有几乎相等的分辨率。采用抽取也能显著地降低运算量。
This paper deals with the problem of spectral estimation of a limited number of two-dimensional sinusoidal signals submerged by white noise. The closed formula of spectral estimation of autoregressive (AR) predictor error filter method is derived. At the same time, closed formulas of BARTLETT window period graph and maximum likelihood ratio spectrum estimation are also derived. These formulas are useful in the study of sinusoidal signals with very close resolution frequency Is very useful. In a narrow frequency domain, the direct extraction can be used to improve the resolution and reduce the computational complexity. The simulation results show that the extraction of (D_1, D_2) on a (N_1, N_2) The extracted but supported expanded (D_1N_1, D_2N_2) results have almost equal resolution. Extracting can also significantly reduce the amount of computation.