The Continuous Interval Multi-scale Fuzzy Threshold De-noising Theory And Application

Wavelet analysis theory has a wide range of applications and has a great advantage in signal de-noising.In real life,signal de-noising is the precondition for extracting informa-tion and solving problems accurately.Among the variety of methods,the wavelet threshold de-noising method is not only easy to operate,but also has obvious effect on de-noising.Besides,it has received wide attention from researchers at home and abroad.Subsequently,the researchers carried out in-depth research on it,and made a series of improved methods,which made the de-noising effect become better and better.However,considering the com-plicated threshold functions of improved schemes which were proposed in some domestic and foreign literatures,and the large amount of calculation,it violated the simple's principle of the wavelet threshold de-noising method.Consequently,a simpler adaptive fuzzy thresh-old method was proposed by any other researchers,which lead to a new research direction for wavelet de-noising method.In this paper,two new fuzzy threshold de-noising methods are proposed to enhance the de-noising effect after exploring the adaptive fuzzy threshold method.In addition,some improved schemes are proposed for the lazy wavelet transform in the lifting algorithm.The main contents of this paper are as follows:First of all,the adjustment factor y in the adaptive fuzzy threshold de-noising method is improved from taking the discrete points to taking points in a continuous interval,and observe the differences of the de-noising effect from the two methods.On this basis,the continuous interval fuzzy threshold de-noising method can be proposed.The semi-soft threshold func-tion and fuzzy threshold value of continuous interval are the keys of the method.In this paper,four common simulation signals are de-noised through the Matlab toolbox.After the simulation,we can find the continuous interval fuzzy threshold method is better than the adaptive fuzzy threshold method.The validity and stability of the de-noising method for continuous interval fuzzy threshold is verified.Secondly,the principle of Mallta algorithm is studied and two new multi-scale thresh-olds are proposed.By Matlab,the new thresholds' de-noising effect is better than other thresholds.The multi-scale fuzzy threshold method is proposed under combinating with the adjustment factor and multi-scale threshold.In addition,a weighted prediction operator is proposed to improve the lazy wavelet transform of wavelet lifting algorithm.The new de-composition method is used to decompose the noise signal,and the de-noising processing is carried out to obtain the optimal weighting factor.Finally,three kinds of the noise in ECG signal mainly is Muscle artifacts,Electrode motion and Baseline wander.The hard threshold method,soft threshold method,adaptive fuzzy threshold method,the continuous interval fuzzy threshold method proposed and two multi-scale fuzzy threshold method in the paper are used to ECG de-noising.According to the experiment of the sample signal in the database,the proposed method filters out the noise successfully and the de-noising effect is better than other methods.