New Wiener Filter De-noising Algorithm Based On Wavelet And Wavelet Packet

With the development of informatization,the use of figure information is becoming more and more common,in addition,with the understandability of the figure information,the figure plays a very important role in communication field.But in the world with radio waves,the noise is one of the inevitable interferences in the process of producing figure.Therefore,in order to make the figure's subsequent processing be in good environment,it is necessary to make de-noise for the figure.However,in the current popular de-noising algorithm,the effect of the wiener filter in wavelet domain de-noising is not ideal,it is easy to have jitter,partial distortion,pseudo-Gibbs or slight fuzzy effect.This is because that the simple combination of wavelet transform and wiener filter can be considered as a special threshold de-noising algorithm,which is difficult to satisfy the demand of multi-scale precise denoising for non-stationary signals.Meanwhile,in the process of figure de-noising,researchers always acquiescence a thing that the mean of gaussian noise under the local wiener filter window is zero,which has not been proved and not fit the case,it may lead to the estimation of wiener filter coefficient have non-negligible deviation.Aiming at the above problems,this paper presents some improvement measures on the original algorithm:Firstly,On the basis of the wiener filter coefficient formula in the wavelet domain and the supposition that the mean of gaussian noise in the local window is not zero,introducing the noise adaptive correction factor? to the original algorithm,which is estimated by the coefficient distribution.In the second place,We can periodically de-noise the image with the wavelet domain wiener filter by the way of iterating wavelet basis and the window,which can make up for the deficiency of the fixed wavelet basis and the window in dealing with the signal frequency.Although the improved algorithm makes good de-noising effective,it still has disadvantages such as large amount of calculation,not exactly de-noising and multiplayer-divide for the high-frequent information.Finally,wiener filter de-noising algorithm based on wavelet packet has been proposed,which directly use the wavelet packet instead of the wavelet to get 32 energy regions by four-layers division.Moreover,it imports the adaptive correction factor? in de-nosing processing to protect that it cann't cause loss for image details when we do the detailed de-noising.In this article,the main content is the research of figure de-noising,and there are two de-noising algorithms have been proposed.Experiments have proved that they have very good de-noising effect and the figure detail protection ability.