Signal recovery and wavelet reproducing kernels
Jian Jiang Lu, D.M. Healy, John B. Weaver · IEEE Transactions on Signal Processing · 1994
A class of signal recovery problems can be formulated as finding missing data at the finest scale of a discrete wavelet transform. A unique and stable recovery can be obtained by solving the regularized wavelet-reproducing equation. We show that this approach has close relations to unconstrained and constrained least-squares techniques and derive a family of regularizing operators adapted to the degrading operator. Experimental results present restored images using regularizing operators of this type.>