The Study Of Image Denoising Based On The Undecimated Discrete Wavelet Transform

As an implement method of the Wavelet Transform,the Undecimated Discrete Wavelet Transform becomes an effective tool in the fields of multiscale geometric analysis because of its good time-frequency localized features and the translation invariance.The Undecimated Discrete Wavelet Transform has a perfect mathematical theory foundation and rapid discretization method.It can not only process data fast and efficiently,but also have an important research significance in the signal processing field.Basing on the characteristics of the Undecimated Wavelet and the statistical property of the coefficients after transform,this pepper puts forward and implements the following image denoising algorithm.(1)According to the curve fitting method to determine the bayesian image denoising threshold has been proposed.First,decompose the image by using the Undecimated Discrete Wavelet Transform.On the basis of translation invariance to keep more details and information of the image.Then choose the generalized gaussian distribution for coefficients modeling which based on the non-gaussian characteristics.Next,according to the correlation between the noise variance and image edge standard deviation,use the curve fitting to determine the new threshold.Finally,through the simulation results of some images verify the algorithm can better keep the edge and details with a higher PSNR.(2)The denoising method,combine the Undecimated Discrete Wavelet Transform and the fourth order partial differential equation effectively,which is based on their advantages and shortcomings.First,the Undecimated Discrete Wavelet Transform decomposition method is applied to get the low frequency coefficients and high frequency coefficients of the image.Then apply the improved fourth order partial differential equation model to deal with the high frequency coefficients.Finally,reconstructing the low frequency coefficients and the high frequency coefficients after processing with the inverse Undecimated Discrete Wavelet Transform.The simulation results show that this method can not only makes full use of the translation invariance and the ability to protect the image detail information after the Undecimated Discrete Wavelet Transform and effective play the characteristics of the improved fourth order partial differential equation to keep the edge and seize of the details,but also restrain the false edge and blocking artifacts effectively.Through the analysis of Undecimated Discrete Wavelet Transform theory and the validation of the two kinds of image denoising method which was proposed in this paper,we found that the translation invariance of the Undecimated Discrete Wavelet Transform has good applicability with image denoising.And the processing result has obvious improved than the original algorithm.