Análise de uma Estratégia de Redução Computacional para o Algoritmo de Matching Pursuit com Dicionários Diádicos

This article presents an analysis of a computati- onal complexity reduction strategy of Matching Pursuit (MP) algorithm, based on a dyadic redundant dictionary. The strategy compute inner products between the signal and dictionary atoms, for different frequencies, fixing scale and time shift, by using Fast Fourier Transform (FFT). Computation time are presented so as to comparing performance of the classic and fast MP versions. The implementations were devised in C++ programming language and Linux environment.