Spectral substraction and spectral estimation

Abstract

The problem of spectral subtraction, to estimate the parameters of a single source in colored noise, is used to show the relationships between the likelihood formulation and spectral density estimation. Reported previously as a filter bank processing for spectral estimation, it is shown that the normalized Capon estimate is the natural tool for source location in I-d and 2-d scenarios when the noise background estimate is faced as a spectral subtraction problem. Several simulations selecting I-d and 2-d apertures are used to show the degree of quality achieved with the proposed formulation. Also, the Periodogram test for incoherent detection is analyzed in front of the optimum test and the herein referred to as the Capon's test.