In this paper we present an efficient 2-D object recognition method th
at uses multiscale tree representations. A planar object is represente
d by means of a tree, in which each node corresponds to a boundary seg
ment at some level of resolution and an are connects nodes correspondi
ng to segments at successive levels that are spatially related. The pr
oblem of matching an object against a model is formulated as the one o
f determining the best mapping between nodes at all levels of the two
associated trees. The proposed matching algorithm is based on dynamic
programming and has optimal O(\T parallel to T'\) time complexity, whe
re \T\ and \T'\ are the number of nodes in the two trees. (C) 1998 Pub
lished by Elsevier Science Ltd on behalf of the Pattern Recognition So
ciety. All rights reserved.