A wavelet algorithm for zoom-in tomography

Abstract

In zoom-in tomography, the aim is to image a region of interest lying partially or fully within the imaged object, using a high resolution tomographic scan covering only the ROI, and a low resolution scan covering the whole object. We analyze the problem from a multiresolution point of view and propose an algorithm for combining the two data sets using the discrete wavelet transform and the Haar wavelet. We compare the proposed algorithm to a previously reported method that involves padding of the high resolution data with a supersampled version of the low resolution data, to zero padding and edge extension, using synthetic data sets. We show that the proposed algorithm is insensitive to offsets between the two data sets, but that it is slightly more sensitive to noise.