Fringe Pattern Denoising Using Two-Dimensional Variational Mode Decomposition (2D-VMD) Method for Inspection of Flatness of Reduced Surfaces

Abstract

The flatness is a major element for any functional precision surface. It becomes extremely important in the miniaturization of the systems, which require the components of reduced sizes. There is a need to develop appropriate methods to solve measurement problems in industrial design, manufacturing of production and process monitoring. In this work, we used an optical method based on the moire interferometry for topography in full-field and to control the flatness of the reduced surfaces having a size of 30 mm2 and more. This technique makes it possible to palpate optically the surface and to give an instantaneous topography of surface in real-time. We have developed an optical set-up which it is possible to inspect in a nondestructive way and without contact, the surface quality. Unfortunately, the moire fringes taken by camera contain excessive noise. That is why we proposed to use the Two-Dimensional Variational Mode Decomposition (2D-VMD) method to reduce the random noise from the fringe patterns and to ameliorate the profiles and the residual images from which we can control the presence of defects flatness of reduced surfaces. We have also used three objective image quality evaluation methods (PSNR, SNR and correlation coefficient) to compare between the denoising results given by 2D-VMD with those obtained by the Bidimensional Empirical Mode Decomposition (BEMD).