COMMON-SIGNAL ESTIMATION

Let s(j) = w r(j) denote two or more time series represented by the convolution of a finite-length wavelet w with different finite-length time series r(j). Then the z-transform of w can be regarded as the pol ynomial factor that the z-transforms of s(j) have in common. This repr esentation is used to estimate w from the s(j). The two algorithms des cribed in this paper are more robust and much faster reformulations of an algorithm published earlier and used to estimate seismic wavelets.