Research On Empirical Mode Decomposition Theory And Its Application

EMD is a new method which is used for signal analysis. It was presented in1998by N. E. Huang. It has a special advantage in analyzing nonlinear and non-stationary signal and is considered a landmark change in the field of signal processing. Signal can be decomposed into a series of IMF by EMD method. IMF is transformed to get instantaneous frequencies which have physical significance. While EMD had been widely used in engineering, there is still having some problems. The paper studies EMD theory deeply and gives its application in practical engineering based on the problems. The main content of this paper is:(1) The principle and properties of the EMD are studied deeply, and then the definition of instantaneous frequency and the IMF are discussed too. The decomposition process of the EMD algorithm and the physical meaning of Hilbert spectrum and marginal spectrum are discussed. The problem in the decomposition process and hot spots on today’s research are pointed out.(2) EMD is extended to two-dimension fields and the principle of BEMD is researched. The decomposition of the BEMD algorithm is analyzed. There are two key technologies in the decomposition process of the algorithm which are extreme points getting and enveloping surface fitting. The pixel8neighborhood and morphological reconstruction law are compared through simulation results on getting extreme points. The triangulation interpolation and the interpolation based on the radial basis function are studied.(3) The EMD is used to remove the signal noise. Firstly, the EMD is used to remove Gaussian noise which is added in the signal. Secondly, BEMD is used in fingerprint image processing.Removing noise from images is achieved with BEMD.