Automatic Interference Term Retrieval From Spectral Domain Low-Coherence Interferometry Using the EEMD-EMD-Based Method

Low-coherence interferometry (LCI) has proved to be a useful tool in optical measurement and detection. However, the noise that is present in practical applications makes interference term retrieval (ITR) difficult. In this paper, we propose a novel technique based on ensemble empirical mode decomposition and empirical mode decomposition (EEMD-EMD) to achieve automatic ITR from the spectral domain LCI. In EEMD, the correlation coefficient of each intrinsic mode function (IMF) is calculated, and a characteristic parameter (CP) is introduced to recognize and remove the IMFs of noisy components. The remaining IMFs are used to reconstruct a new signal followed by EMD to extract the interference term. An experiment based on distributed polarization coupling detection in polarization-maintaining fibers (PMFs) was conducted; the PMFs having different lengths and coupling intensities were investigated. It was shown that by setting an appropriate CP, the interference term could be automatically extracted. The relative errors of the extracted coupling intensity were less than 2%. This suggests that our method may have potential applications in interferometric measurement.