Fringe pattern normalization using the Bidimensional Empirical Mode Decomposition method

BERNINI, M. B.; FEDERICO, A; KAUFMANN, G. H.

Stuttgart, Germany

Congreso; Fringe 2009; 2009

Stuttgart University

Phase extraction techniques in fringe analysis are very sensitive to noise but also to modulation defects mostly due to non-uniform illumination. For this reason, in many cases the fringe pattern modulation must be normalized before the phase map is extracted in order to avoid the introduction of large errors in the demodulated phase. The application of a normalization method can be also of great utility in Digital Speckle Pattern Interferometry (DSPI). When it is necessary to smooth the DSPI fringes with a denoising technique previous to phase extraction, frequently the smoothing process reduces the visibility of the fringes. Therefore, a normalization algorithm must be used to pre-process the smoothed DSPI fringes before the extraction of the phase distribution. Various fringe normalization techniques have been proposed, such as the one presented by Quiroga et al. Even though this method has been successfully applied in fringe normalization, its efficiency is reduced in images presenting regions with very low modulation or stepped contrast changes. In this paper we describe an approach to normalize fringe patterns based on the Bidimensional Empirical Mode Decomposition (BEMD) method adapted to DSPI and the partial Hilbert transform (PHT).