This paper presents a new cooperative matching algorithm based on the
integration of stereo and motion cues. In this algorithm, stereo dispa
rity and image flow values are recovered from two successive pairs of
stereo images by solving the stereo and motion correspondence problems
. Feature points are extracted from the images as matching objects. Th
e entire matching process composes of a network of four subprocesses (
two for stereo and two for motion). Each of the subprocesses can acces
s information from connected nodes to perform the disambiguation. The
''best'' matches are obtained in a relaxation manner using the 3-D con
tinuity constraint. Experimental results are presented to illustrate t
he performances of the proposed method.