A wavelet based recursive reconstruction algorithm for linear measurements
Orhan Arıkan · 2002
A recursive algorithm is proposed to obtain an efficient regularized least squares solution to large linear system of equations which arises in many physical measurement models. The algorithm recursively updates the solution in an increasingly larger dimensional subspace whose basis vectors are chosen as a subset of a complete wavelet basis. Robust criterions on how to chose the basis vectors at each iteration, and when to stop the iterations are given.