The automation of imagery analysis processes leads to the need to detect ch
ange between pairs of aerial reconnaissance images. Approximate camera mode
ls are available for these images, accurate up to a translation, and these
are augmented with further constraints relating to the task of monitoring v
ehicles. Horizontal, bilateral, Euclidean symmetry is used as a generic obj
ect model by which segmented curves are grouped, first in a 2-d approximati
on, and then in 3-d, resulting in a sparse 3-d Euclidean reconstruction of
a symmetric object from a single view. The method is applied to sample imag
es of parked aircraft.