The method of reducing the size of an initial image into a final image, each image being formed by a matrix of ordered pixels, each pixel being characterised by a characteristic quantity (Cn), the reduction being effected by a ratio P/Q with P and Q integers and Q not a multiple of P, comprising a pixel decimation step, in which the following steps are implemented: --effecting a partitioning of the initial image into groups of pixels; --allocating, successively to each group of pixels n of the initial image, a coefficient .mu.n the coefficients of the groups of pixels forming an arithmetic sequence of ratio P in Z/QZ defined by the recurrence relationship .mu.n+1.ident..mu.n+P(Q); --selecting only the groups of pixels whose coefficient allocated is a number less than P-1; --decimating the groups of pixels not selected: and--forming the final image of reduced size solely from the groups of pixels selected, each pixel in the groups of pixels selected being characterised by the filtered characteristic quantity (C'n).