Image Denoising Using Multiwavelet Filterbanks

A plenty of research about thresholding methods in wavelet domain has been proposed by many authors. Multiwavelet domain can be decomposed as scalar wavelets and differently according to the structure of scaling functions. When scaling functions are symmetric-antisymmetric, the antisymmetric part is low pass filter in mathematical formula. But in practice it works as high pass filter because of antisymmetric structure. By the method of scalar wavelet decomposition, we cannot use the effect of the multiwavelet structure like this. In multiwavelets, 15 wavelet bands appear at each decomposition level. Therefore, we cannot use the many thresholding methods set in scalar wavelets. When using the multiwavelets, prefiltering process is needed. This paper presents the decomposition method effectively using the symmetric-antisymmetric structure. And the complexity is reduced by including the prefilter matrix to low pass filter and high pass filter respectively. Experimental results demonstrate that this method exhibits performance equal to, or superior to the traditional ways.