This paper presents an approach to the local stereo matching problem u
sing edge segments as features with several attributes. We have verifi
ed that the differences in attributes for the true matches cluster in
a cloud around a center. The correspondence is established on the basi
s of the minimum squared Mahalanobis distance between the difference o
f the attributes for a current pair of Features and the cluster center
(similarity constraint). We introduce a learning strategy based on th
e Self-Organizing feature-mapping method to get the best cluster cente
r. A comparative analysis among methods without learning is illustrate
d. (C) 1998 Elsevier Science B.V. All rights reserved.