| The wavelet transform has been successfully applied to denoising signals in the field of signal processing. This dissertation introduces a new technique for noise reduction using an adaptive wavelet threshold based on higher order statistics for denoising. We applied an algorithm based on higher order statistics to determine is wavelet coefficients contained mostly noise or signal. We also provided analytical results for some simple wavelets.; We performed experiments with one- and two-dimensional data sets to evaluate our approach with additive white Gaussian noise. We found that our approach worked better than a more conventional denoising approach for low SNR values. The adaptive wavelet threshold based on higher order statistics efficiently associates with the characteristics of input data, thus, it removes noise and preserves the detail parts without losing important information. |
