Regenerated phase-shifted sinusoids assisted EMD for adaptive analysis of fringe patterns

Abstract

Fringe patterns are often produced from optical metrology. It is important yet challenging to reduce noise and remove a complicated background in a fringe pattern, for which empirical mode decomposition based methods have been proven useful. However, the mode-mixing problem and the difficulty in automatic mode classification limit the application of these methods. In this paper, a newly developed method named regenerated phase-shifted sinusoids assisted empirical mode decomposition is introduced to decompose a fringe pattern, and subsequently, a new noise-signal-background classification strategy is proposed. The former avoids the mode-mixing problem appearing during the decomposition, while the latter adaptively classifies the decomposition results to remove the noise and background. The proposed method is testified by both simulation and real experiments, which shows effective and robust for fringe pattern analysis under different noise, fringe modulation, and defects.