In this paper,a new method is presented for 3D motion estimation by image region correspon-dences using stereo cameras.Under the weak perspectivity assumption,we first employ the momenttensor theory(Cyganski and Orr)to compute the monocular affine transformations relating imagestaken by the same camera at different time instants and the binocular affine transformations relatingimages taken by different cameras at the same time instant.We then show that 3D motion can herecovered from these 2D transformations.A space-time fusion strategy is proposed to aim at robustresults.No knowledge of point correspondences is required in the above processes and the computa-lions involved are linear.To find corresponding image regions,new affine invariants,which showstronger invariance,are derived in term of tensor contraction theory.Experiments on real motionimages are conducted to verify the proposed method. In this paper, a new method is presented for 3D motion estimation by image region correspondence-dences using stereo cameras. Unders the weak perspectivity assumption, we first employ the momenttensor theory (Cyganski and Orr) to compute the monocular affine transformations relating image staken by the same camera at different time instants and the binocular affine transformations relatingimages taken by different cameras at the same time instant. We then show that 3D motion can herecovered from these 2D transformations. A space-time fusion strategy is proposed to aim at robustresults .No knowledge of point correspondences is required in the above processes and the computa-lions involved are linear. To find corresponding image regions, new affine invariants, which shows tronger invariance, are derived in term of tensor contraction theory. Experiments on real motion images are conducted to verify the proposed method.