The invention relates to a method of computing the transformation which transforms
two different images (10, 10) of an object one into the other while
describing the motion or deformation of the object. In accordance with the method
first the local transformation parameters are computed for sub-regions (1, 2,
3, 4, 5). By choosing the sub-regions so as to be sufficiently small, a rigid
transformation can be used for this purpose. Starting from at least one first predetermined
sub-region (1), the local transformation parameters of the neighboring sub-regions
(2, 3, 4, 5) are successively computed, the starting values on which each
computation is based being the already determined local transformation parameters
of neighboring sub-regions. Using the local transformation parameters thus determined
for sub-regions, the overall transformation can subsequently be computed.