Input-Dependent FIR Prefiltering for Two-Dimensional Spectral Estimation
Tetsuya Shimamura, Jouji Suzuki · Transactions of the Society of Instrument and Control Engineers · 1992
In this paper, we propose an approach to two-dimensional spectral estimation, which involves inputdependent FIR prefiltering.This approach is an iterative algorithm in which the design of twodimensional zero-phase FIR filter and the operation of spectral estimation are sequentially performed.Although we use the periodogram for the purpose of calculating the power spectrum, an introduction of the input-dependent prefiltering leads to high spectral resolvability and reduced estimation variance.Any conventional Fourier-based method did not provide the both simultaneously.Also, the computational time wasted on this approach is moderate, since we can utilize iteratively the computational efficiency of the two-dimensional FFT algorithm.The effectness of the spectral estimation algorithm is shown by computer simulations.Furthermore, the two-dimensional spectral estimation approach could be extended to the multidimensional version.The extended algorithm would provide similar properties to those of the twodimensional algorithm.