6 Conclusion

Anisoplanatism is one of the most important limitation of large FOV high resolution imaging. We present a simple and analytical expression to model the OTF variation as a function of field angle. The OTF for an angle $\alpha$ is the product of the on-axis OTF by an anisoplanatic transfer function (ATF) which only depends on the angular decorrelation of the phase. The incorporation of this ATF in a post processing of large FOV images corrected by AO leads to good accuracy on the object parameter estimation. This method has been validated both on simulated and experimental data. In the case of a stellar field, the relative error on the magnitude estimation can be divided by more than an order of magnitude for large angles.

Nevertheless, the OTF estimation is never perfect and a residual error still limits the post-processing performance (error $\simeq 1\%$). This residual error is due both to the non-stationarity of the AO compensated phase in the pupil and to the turbulence noise (finite exposure time). Furthermore, the $C_n^2\$profile must be known to compute the ATF, even if there is a weak dependency of the results with this profile. A SCIDAR [Fuchs et al.1998] on the astronomic site may provide such an information, but this $C_n^2\$knowledge may be difficult to obtain in all cases. Another solution to these problems could be to use an approximated parametric model of the ATF in order to perform a joint estimation of the object and of the ATF parameters.

Acknowledgements

This work was supported by contracts from Service Technique des Technologies Communes, Ministère de la Défense, France. The authors wish to thank J.-P. Véran, F. Charbonnier and A. Blanc for fruitful discussions. Many thanks also to M. Azouit for the $C_n^2\$data acquisition and to P.-Y. Madec, D. Rabaud and B. Fleury who took part to the AO observing run.