1: \begin{abstract}

2: In this article we propose a novel {\em nonstationary iterated Tikhonov} (NIT)

3: type method for obtaining stable approximate solutions to ill-posed operator

4: equations modeled by linear operators acting between Hilbert spaces.

5: %

6: Geometrical properties of the problem are

7: used to derive a new strategy for choosing

8: the sequence of regularization parameters (Lagrange multipliers) for the NIT

9: iteration.

10: %

11: Convergence analysis for this new method is provided.

12: %

13: Numerical experiments are presented for two distinct applications: I) A 2D

14: elliptic parameter identification problem (Inverse Potential Problem);

15: II) An image deblurring problem.

16: The results obtained validate the efficiency of our method compared with

17: standard implementations of the NIT method (where a geometrical choice

18: is typically used for the sequence of Lagrange multipliers).

19: \end{abstract}