Modernization of adaptive matching pursuit method to analyze geophysical signals of pulse nature

Abstract

The paper is devoted to the development and comparison of different numerical methods which increase the adaptive property and improve the accuracy of matching pursuit algorithm in connection to geoacoustic and electromagnetic signals. At each step of adaptive matching pursuit, a function is chosen which has the highest correlation with an initial signal. Then parameters of a chosen function are refined. The refinement is performed by the help of different grid methods and methods based on gradient direction search. The paper considers the peculiarities of application of sparse approximation methods to geophysical signals of pulse nature and compares different variants of modification of adaptive matching pursuit algorithm.