Adaptive noise repression based on orthogonal empirical mode decomposition and correlation detection

This paper uses orthogonal empirical mode decomposition (OEMD) to decompose the mixed signal into a set of intrinsic mode functions (IMF). Different from traditional noise removing method, OEMD can both suitable for non-linear and non-stationary hypothesis with good adaptability. After processing of OEMD, IMFs will be analyzed for the correlation between each other. Finally, recovers the source signal from intermixed signal with noise.

1. INTRODUCTION

Signals that we received in a general condition are always mixed with several source signals and noises. It's not an easy work to extract signals what we need when source signals were mixed with noises. Noise makes the signal processing become a difficult work. The overlapping part of source signals and noises would guide our result to a wrong way. In general Blind Source Separation (BSS) problems, source signal is always mixed with noise, we can't avoid meeting the problem in general BSS problems with that source signal is always mixed with noise. That's why denoising of signal or noise repression become an important subject when dealing the signal analysis problems.