Image reconstruction from projections under wavelet constraints
Berkman Sahiner, A.E. Yagle · IEEE Transactions on Signal Processing · 1993
First, the authors discuss how the wavelet transform can be used to perform spatially-varying filtering of an image, suppressing noise locally in smooth regions of the image, and discuss detection of such regions in a noise-corrupted image. Second, they show how to compute the minimum mean-square estimate of an image given: (1) noisy projections of the image; (2) statistics of additive noise in the projections; and (3) constraints on wavelet coefficients of the image. Examples illustrate the resulting procedure.>