This paper addresses the multi-resolution implementation of a Cox&Roy optimal flow image matching algorithm. This minimization algorithm aims at solving the surface reconstruction problem formalized as a minimization of an energy. This multi-resolution approach is necessary for achieving reasonable processing times on extended areas and improving robustness by restraining matching ambiguities. Some very good results are shown on a 10m ground sample distance (GSD) SPOT5-HRS stereopair. Intermediate results at different steps of the reconstruction process show all the Level Of Details provided by the different pyramid scale levels. The tuning of the regularization parameter is also discussed for different landscapes.