Research And Application Of Signal De-noising Based On Wavelet Analysis

Wavelet analysis has been widely used in the information technology and other subjects, and many issues are concerned by technology workers. In theory, new ideas, methods are emerging. This dissertation aims to consummate the wavelet theory, present some new de-noising algorithms and develop the new scopes of wavelet application.Typical signal processing means, such as Fourier transform, Short term Fourier transform and so on, always have their own limitations. As the result, their application extents are restricted. As a new signal processing method, wavelet analysis decomposed different kinds of frequency element to non-overlapped frequency bands, and this method puts forwards an effective way to signal filter, signal-noise separating and character picking-up, especially it has well de-noising capability. This paper introduced the classic de-noising method, and it's applied of scope and results to carry on the analysis and comparison.The fundamental theories of wavelet analysis are discussed in detail. Continuous wavelet transform, discrete wavelet transform, and dyadic wavelet transform are introduced. The fast algorithm of discrete dyadic wavelet transform is given. Finally an analysis is made on the influence of the wavelet bases on practical applications by studying their mathematical properties. Then studied the wavelet transform theories, analyzed the characteristics of the wavelet transform, aim at the non-stable signal, and the system introduced the common use wavelet de-noising methods: coefficient of high frequency placed to zero method, maximum de-noising method, threshold de-noising method, and the spatial correlation algorithm de-noising method. The coefficient of high frequency placed to zero method is convenient for de-noising but it will lose some details in the process of signal reconstitution. Although it is difficult to choose the wavelet bases only depending on experiences, it is better than filter. The principles of modulus maximum de-noising and the analysis in the choice of some parameters in the process are made in detail. Based on the threshold de-noising, this paper puts forward a de-noising method that combines mean approximation and threshold together, and the method comes true through the experiment simulations. The results indicate that it increases the SNR, and de-noising effect is better than the threshold de-noising only. A combination de-noising algorithm based on the spatial correlation algorithm is presented, and the experimental results show that the filtered wavelet coefficient in the proposed algorithm has the advantage of good continuous, high accuracy and convenient for signal reconstruction.The analysis of de-noising algorithms is respectively introduced in the dissertation, and carried out on the experiments. The simulation results show that using the wavelet analysis to signal de-noising is efficient and practical.