High-frequency random noise elimination using wavelet transform
Zhang Yu, Zhang Guanquan
National Laboratory of Scientific&Engineering Computation, Research Institute of Numerical Mathematics and scientific-Engineering Computation, Chinese Academy of Sciences, Beijing City 100080
Abstract：Wavelet transform is theoretically based on some conclusions which are derived from wavelet singularity analysis.By analysing the very different characteristics of signals and noises in running wavelet transforms,we can locate low SJN contents,remove the corresponding orthogonal wavelet component and achieve noise elimination after inverse transform.High-frequency random noise elimination using wavelet transform is characterized by automatically locating the low SJN intervals,and local noise eliminations in both time domain and frequency domain.Numerical experiment results show that the method produces the seismic section which shows improved effects in average signaljnoise ratio,visual signalJnoise ratio and visual resolution.
