A Novel Method of Wavelet Threshold Shrinkage Based on Genetic Algorithm and Sample Entropy

In order to denoise different type of noisy signals, the genetic algorithm and sample entropy are applied to the wavelet transform threshold shrinkage (WTS) method. As the genetic algorithm is used, the parameters needed in WTS, such as the wavelet function, decomposition levels, threshold functions and threshold can be optimized automatically. For the sample entropy can measure the complexity of different signals, it is adopted in the fitness function with the root mean square error of wavelet coefficients after and before denoising process, and so the individual of genetic algorithm can be evaluated more effectively. The proposed method can achieve the optimal denoising results for different type signals, and finally the effectiveness of this method is validated by the results of simulation for four benchmark signals.