Noise reduction in digital speckle pattern interferometry using bidimensional empirical mode decomposition

Abstract

We developed a bidimensional empirical mode decomposition (BEMD)method to reduce speckle noise in digital speckle patterninterferometry (DSPI) fringes. The BEMD method is based on asifting process that decomposes the DSPI fringes in a finite setof sub-images represented by fast and slow oscillations, which arenamed modes. The sifting process assigns the high frequencyinformation to the first modes, so that it is possible todiscriminate speckle noise from fringe information, which iscontained in the remaining modes. The developed method is a fullydata-driven technique, therefore neither fixed basis functions noroperator intervention are required. The performance of the BEMDmethod to denoise DSPI fringes is analyzed using computersimulated fringes and results are also compared with thoseobtained by means of a previously developed one dimensionalempirical mode decomposition approach.