The problem of stereo vision has been of increasing interest to the co
mputer vision community over the past decade. This paper presents a ne
w computational framework for matching a pair of stereo images arising
from viewing the same object from two different positions. In contras
t to previous work, this approach formulates the matching problem as d
etection of a ''bright'', coherent disparity surface in a 3D image cal
led the spatio-disparity space (SDS) image. The SDS images represents
the goodness of each and every possible match. A nonlinear filter is p
roposed for enhancing the disparity surface in the SDS image and for s
uppressing the noise. This filter is used to construct a hyperpyramid
representation of the SDS image. Then the disparity surface is detecte
d using a coarse-to-fine control structure. The proposed method is rob
ust to photometric and geometric distortions in the stereo images, and
has a number of computational advantages. It produces good results fo
r complex scenes.