Many of the existing image processing applications, in particular the construction of super-resolution videos, require an accurate high-speed motion tracking algorithm. This paper proposes an efficient recursive approach that recovers 3-D motion from a stereo image sequence with high precision based on the trifocal tensor. In the computation process, neither the 3-D structure of the scene nor its reconstruction is necessary. The validity of the proposed algorithm is demonstrated by applying it to upgrade the resolution of real stereo images. Empirical comparisons show that our novel approach outperformed traditional work in terms of both the accuracy of motion estimation and the quality of super-resolved images.