Title :
On the inversion of singular operators
Author :
Casares-Giner, Vicente
Author_Institution :
Universidad Politécnica de Barcelona, Barcelona, Spain
Abstract :
This paper is concerned with the inversion of digital singular convolution operators defined by a polynomial in the z-transform domain. For that purpose, in the one dimensional case (1D), the method given here proceeds by first approximating the polynomial by a rational transfer function, Which numerator results to be an autocorrelation functions, free of zeros on the unit circle. This is obtained by using the least square inverse concept of a polynomial. The extension to the 2D case, can be achieved using the well known 2D planar least square inverse polynomial (PLSI) in semicausal form. With this approach, the inversion deals with a non singular autocorrelation functions, so we can use the asymmtotical method given recently |1|.
Keywords :
Autocorrelation; Equations; Large scale integration; Least squares approximation; Linear systems; Mean square error methods; Polynomials;
Conference_Titel :
Acoustics, Speech, and Signal Processing, IEEE International Conference on ICASSP '82.
DOI :
10.1109/ICASSP.1982.1171484