The method is for processing a multiplex image, the multiplex including at least one first view intended to be viewed by a first eye of an observer and at least one second view intended to be viewed by a second eye of the observer. The two views are spatially sub-sampled according to complementary grids and mutually spatially shifted. The method includes a demultiplexing of the multiplex image so as to extract the first and the second views. And, for at least one missing pixel of the first view, there is a determination of a first window of the first view containing the location of the missing pixel and representing a first detail in the first view, a determination of a second window of the second view representing the same first detail in the second view, and a formulation of the missing pixel by using the pixels of the second window.