Denoising of digital speckle pattern interferometry fringes by means of Bidimensional Empirical Mode Decomposition

摘要

We present an introduction to the Bidimensional Empirical Mode Decomposition (BEMD) and its application to the denoising of DSPI fringes. The BEMD is based on the decomposition of an image in high and low frequency zero-mean oscillation modes, called intrinsic mode functions (IMFs). The decomposition is carried out through a sifting process which produces significantly fewer basis functions than the ones generated by the Fourier or the wavelet transforms. The denoising approach is based on the removal of the first IMFs, so that the filtered image is given by the residue. A normalization algorithm is then applied to the denoised fringes to reduce the oversmoothing caused by the filtering. The performance of this denoising approach was evaluated using computer-simulated DSPI fringes with different fringe density and speckle size, in order to calculate a figure of merit through the comparison with the noise-free fringes. The obtained results are also compared with those produced by other smoothing methods, and the advantages and limitations of the proposed approach are finally discussed.