Ondaletas e modelagem estatística aplicados na redução de ruído em sinais

Abstract

In this work new methods are proposed to reduce noise in signals. These methods are based on the discrete wavelet transform and on the statistical modeling of the coeicients obtained from the application of this transform. Monte Carlo simulations will be performed considering the Donoho & Johnstone test signals, electrocardiogram signals obtained from the MIT-BIH database and seismic traces obtained from the R astsa package. The simulations will be performed considering diferent signal-to-noise ratios and the proposed new methods will be compared with other methods already well known in the literature.