Stationary Wavelet Packet Programmable Filters for Real Time Signal Detection and Denoising

摘要

In this work we present a new methodology for Radar or Sonar pulses detection and filtering in presence of strong noise and jamming signals. The detection and denoising can be obtained through an hard thresholding or a pattern matching procedure performed on the coefficients resulting from a Stationary Wavelet Packet Transform analysis on the noisy signal. When the knowledge of the ideal received signal spectrum is given, we can select the optimus wavelet packet tree, and then filter the frequencies out of the sig-nal band. Moreover, we give useful guidelines to design hardware implementation of the proposed algorithm steps, to perform a real time pulse detection and denoising. The architecture is easily reconfigurable, so we can eventually redirect the analysis to different wavelet packet domains.