The invention relates to a method for coding an image that is to be coded, characterized in that said method comprises the following steps: definition of a hierarchical mesh structure having at least two levels of interconnected meshing formed from mesh units which are defined by peaks; determination, for each of said mesh units, of information concerning an error between the image which is to be coded and an interpolated image obtained from the peaks of the mesh units pertaining to the level of the meshing of the mesh unit in question; termination (33) of the refinement of mesh units having error information below a first predetermined threshold; implementation of a specific coding (35) for the mesh units having error information above a second predetermined threshold; continuation (34) of the refinement of mesh units having error information above the first predetermined threshold and below the second predetermined threshold.