Removing Quasi-Periodic Noise in Strain Maps by Filtering in the Fourier Domain

Abstract

Quasi-periodic noise due to various reasons often corrupts strain maps obtained with full-field measuring systems. The aim of this didactic paper is to show how to remove this noise by changing some Fourier coefficients involved in the two-dimensional (2D) Fourier transform of these strain maps. The basics of the 2D Fourier transform of images, which is a common tool in image processing but that is only scarcely employed in the experimental mechanics community, are first briefly recalled. Several procedures employed for removing undesirable frequencies in strain maps are then discussed. Three different examples illustrate the benefit of this approach.