A ROBUST TECHNIQUE FOR MATCHING 2 UNCALIBRATED IMAGES THROUGH THE RECOVERY OF THE UNKNOWN EPIPOLAR GEOMETRY

This paper proposes a robust approach to image matching by exploiting the only available geometric constraint, namely, the epipolar constrai nt. The images are uncalibrated, namely the motion between them and th e camera parameters are not known. Thus, the images can be taken by di fferent cameras or a single camera at different time instants. If we m ake an exhaustive search for the epipolar geometry, the complexity is prohibitively high. The idea underlying our approach is to use classic al techniques (correlation and relaxation methods in our particular im plementation) to find an initial set of matches, and then use a robust technique-the Least Median of Squares (LMedS)-to discard false matche s in this set. The epipolar geometry can then be accurately estimated using a meaningful image criterion. More matches are eventually found, as in stereo matching, by using the recovered epipolar geometry. A la rge number of experiments have been carried out, and very good results have been obtained. Regarding the relaxation technique, we define a n ew measure of matching support, which allows a higher tolerance to def ormation with respect to rigid transformations in the image plane and a smaller contribution for distant matches than for nearby ones. A new strategy for updating matches is developed, which only selects those matches having both high matching support and low matching ambiguity. The update strategy is different from the classical ''winner-take-all' ', which is easily stuck at a local minimum, and also from ''loser-tak e-nothing'', which is usually very slow. The proposed algorithm has be en widely tested and works remarkably well in a scene with many repeti tive patterns.