Title :
A method for the direct frequency-domain deconvolution without division
Author_Institution :
Dept. of Comput. Sci., Maribor Univ., Slovenia
fDate :
5/1/1993 12:00:00 AM
Abstract :
Describes a frequency-domain deconvolution that uses no division and does not suffer from singular kernels. A fast computational algorithm is derived which works equally well for nonsingular as for singular kernels and has a complexity which is proportional to the fast Fourier transform (FFT). Its noise behavior is also studied statistically on real signals, showing improved performance, on average, compared to the direct frequency-domain deconvolution
Keywords :
fast Fourier transforms; signal processing; computational algorithm; direct frequency-domain deconvolution; fast Fourier transform; noise behavior; nonsingular kernels; real signals; singular kernels; Computational complexity; Convolution; Deconvolution; Fast Fourier transforms; Iterative methods; Kernel; Signal processing algorithms; Signal resolution; Time domain analysis; Time invariant systems;
Journal_Title :
Circuits and Systems II: Analog and Digital Signal Processing, IEEE Transactions on
