These constraints can be used for checking the consistency of a system of
views involving motion in a multiview environment. This is easy to see as
the constraints we derived are view independent and should come out to be
the same in all views. If we can track sets of
points each for
many time instant in each view, we can compute
sets of
values from the reference view. The same
values will hold for
all views for each set of points at every time instant. Given the image velocities
in the other views and the
values computed from the reference
view, the matching time instants can be computed for each view, as in
Subsection 4.1. A necessary condition for all the
views to belong to the same motion configuration is that all computed
time
instants that align the views be congruent.