Enhancing single-shot fringe pattern phase demodulation using advanced variational image decomposition

Abstract

In many full-field optical metrology techniques, i.e. interferometry, moiré, and structured light, the information about the measurand, e.g. displacement, strain, or 3D shape, is stored in the phase distribution of a recorded two-dimensional intensity pattern—the fringe pattern. Its analysis (phase demodulation) therefore plays a crucial role in the measurement procedure, significantly affecting the total accuracy of the optical system. Phase demodulation methods based on just a single fringe pattern are especially interesting, due to their robustness to environmental disturbances and ability to examine dynamic events. However, the calculated phase map is easily spoiled by errors, which appear mainly because of fringe pattern imperfections, i.e. random noise, parasitic interferences, a nonsinusoidal fringe pattern profile or a non-uniform image background. In this contribution, an advanced variational image decomposition scheme is proposed to reduce these phase demodulation errors. The reported purely phase domain method can be easily adopted to aid virtually any fringe analysis method, including single-frame and multi-frame phase-shifting, possibly enhancing retrieved phase distribution without the need for hardware manipulation. We employed it to improve single-frame Hilbert–Huang transform-based fringe analysis. The remarkable efficiency and versatility of the developed algorithm are verified by processing synthetic and experimental fringe patterns and phase maps. The demonstrated approach compares favorably with the very capable 2D empirical mode decomposition reference method.