A WAVELET SHRINKAGE APPROACH TO TOMOGRAPHIC IMAGE-RECONSTRUCTION

A method is proposed for reconstructing images from tomographic data w ith respect to a two-dimensional wavelet basis. The Wavelet-vaguelette decomposition (WVD) is used as a framework within which expressions f or the necessary wavelet coefficients may be derived. These coefficien ts are calculated using a version of the filtered back-projection algo rithm as a computational tool, in a multiresolution fashion. The neces sary filters are defined in terms of the underlying wavelets. Denoisin g is accomplished through an adaptation of the wavelet shrinkage (WS) approach of Donoho et al. and amounts to a form of regularization. Com bining these two steps yields the proposed WVD/WS reconstruction algor ithm, which is compared to the traditional filtered backprojection met hod in a small study.