Title :
Kernel-factorization deconvolution method
Author :
Kong, F.N. ; Li, Z.P.
Author_Institution :
R. Norwegian Council for Sci. & Ind. Res., Kjeller, Norway
fDate :
1/1/1990 12:00:00 AM
Abstract :
The deconvolution problem for a cycle convolution equation with the input (or the deconvolution kernel) spectrum containing zeros is discussed. It is shown that through factorization of the kernel the impulse can be obtained by combined use of the two conventional methods: the Z-transform division method and DFT (discrete Fourier transform) division method, if the length of the impulse response is ⩽N-K, where N is the cyclic period and K is the number of kernel-spectrum zeros. Only deconvolution with noiseless data is considered
Keywords :
Z transforms; fast Fourier transforms; spectral analysis; DFT division method; Z-transform division method; cycle convolution equation; deconvolution method; discrete Fourier transform; impulse length response; kernel-factorization; kernel-spectrum zeros; noiseless data; Acoustic signal processing; Autocorrelation; Convergence; Councils; Deconvolution; Finite impulse response filter; Signal to noise ratio; Speech processing; Surveillance; Upper bound;
Journal_Title :
Acoustics, Speech and Signal Processing, IEEE Transactions on
