Font Size: a A A

Denoising Method Based On Wavelet Transform And Independent Component Analysis

Posted on:2011-06-09Degree:MasterType:Thesis
Country:ChinaCandidate:J ChengFull Text:PDF
GTID:2208360305497484Subject:Circuits and Systems
Abstract/Summary:PDF Full Text Request
The real signal may be subjected to varying degrees of random noise pollution in the process of stimulation, transmission and detection. Particularly in the small signal acquisition and measurement, noise interference is more serious. It can bring large measurement error and affect the subsequent analysis process. The noise is random and unpredictable, often widely distributed and its bandwidth always overlaps with the signals'. How to eliminate noise from the noisy signal to extract useful information has been the focus of signal processing area. The traditional Fourier transform is no longer applicable. The wavelet analysis rise in recent years with its unique time-frequency and multi-resolution properties have a great advantage in de-noising.Independent Component Analysis is a new analysis method which developed with the blind signal separation. The principle of independent component analysis algorithm is to find the mutual independent underlying components, to remove the higher-order redundant between components and to extract the independent original signals according to the analysis of higher-order statistical relationships among the multidimensional observed data. It can better characterize the internal structure oof the signal. So it can achieve good result in de-noising.This paper firstly describe the basic theory of wavelet transform, introduce several major methods of wavelet de-noising and their respective principles, the de-noising effect on their simulation experiments, and summarizes the advantages and disadvantages of their application. Then it proposes an improved spatial filtering method which achieves good result.Second, due to the shortcomings of the wavelet de-nosing, it introduce the theory of independent component analysis. The basic theory and algorithm of the ICA are firstly introduced. Then we focus on the algorithms of the FastICA based on negentropy and relative gradient. An adjustable rate of relative gradient algorithm then is proposed. With the changes of iteration number, the learning rate of relative gradient algorithm corresponding changes, which solves the problem on the contradiction between the convergence rate and stability well. And then the ICA method is applied to signal de-noising, the experiment results show that it has better de-noising performance.Finally, it summarizes the de-noising work and the future work.
Keywords/Search Tags:Wavelet Transform, Wavelet De-noising, Spatial Filtering, Independent Component Analysis, FastICA, Relative Gradient, Adjustable Rate
PDF Full Text Request
Related items