1: \begin{abstract}

2: We quantify the performance of a non-linear Wiener filter,

3: constructed in wavelet space, at recovering some of the Fisher

4: information that was lost in the weak lensing convergence field. The

5: proposed method consists in a separation of the original field into

6: the sum of a Gaussian and a non-Gaussian contribution. After

7: filtering an ensemble of such fields, which are obtained from

8: $N$-body simulations, we find that we can recapture about four times

9: more Fisher information, an effect that can potentially improve by a

10: significant amount the constraining  power of weak lensing surveys

11: on cosmological parameters, including the dark energy equation of

12: state  $\omega$. We compare this performance with that of a

13: logarithmic mapping and find that the wavelet method can recover up

14: to three times more information.

15: \end{abstract}