Empirical Mode Decomposition and Wavelet Decomposition in Respiratory Sounds Processing

Abstract

In order to process respiratory sounds to achieve acquiring available information of sound signals Empirical Mode Decomposition (EMD) and Wavelet Decomposition (WD) are applied separately to analyze lung sound signals. The de-noise of original signals is processed with spectral subtraction method. EMD divides the signal into independent Intrinsic Mode Functions (IMFs). Moreover, WD method decomposes the signal with wavelet transform. After receiving the decomposition signals of EMD and WD, a comparation is demonstrated. According to the results, WD has 85% signal information concentrate on layer 7 and Hilbert diagram shows the EMD owns more efficiency in decomposing signals with information keeping.