High Order Statistics Analysis And Its Applications

High-order statistics (HOS),as a mathematical tool widely used in signal processing theory,with its superior to one-order,two-order statistics in antinoise and some important characteristics,has obtained extensive attention and application in recent years,especially in the case when the background noise signal is Gauss noise,its effect is more significant.This paper studies the characteristics estimation methods and properties of the high-order statistics, then based on Matlab high-order spectral analysis toolbox test and understand the outstanding performance of high-order statistics in anti noise.This paper applied high order statistics to signal processing,including typical one-dimensional speech signal processing and multidimensional signal in seismic exploration.In the speech signal processirg,the high order cumulants paremeters are used in speech activity detection,combined with the traditional characteristic parameters,in order to improve the anti noise properties of activity detection algorithm.Then test on the real data based on Matlab simulation verification,and get a good performance in speech activity detection under low SNR circumstance.According to the multidimensional signal of seismic reflection,in order to use seismic signal parameters to analysis the seismic fault,the traditional method tend to use the corss-correlation time delay method.This paper will use a time delay estimation method based on four order cumulant in the literature to detect the seismic fault.Compared with the traditional method,the new method that introduced four order cumulant to get seismic time delay profile,detection effect of seismic tomography subjective effect are obviously better than the traditional method.