A Wavelet Threshold De-noising Algorithm Based on Adaptive Threshold Function

De-noising is an important application field of the wavelet analysis. It has advantage over the traditional filtering methods for its well localized time and frequency property. A central issue in the signal de-nosing research is how to obtain a good balance between shrinking noise and preserving the signal singularity features. This paper presents a wavelet de-noising method based on an adaptive threshold function. By tuning the parameter of the threshold function, the noise wavelet coefficients are shrunk while the signal details are preserved as much as possible on the small scales of the wavelet transform, and on the other hand, the noise coefficients are removed to their maximum extent on a large scale. The simulation results of the blocks, bumps and signals corresponding to the sonar returns from underwater targets, demonstrate that the signal singularity features by adopting the proposed method are better preserved with significant advantage than the traditional threshold filtering method.