Abstract
A closed-form solution for motion estimation from first-order flow in two 'distinct' image regions is described. Uniqueness is guaranteed when these correspond to surface patches with different normal vectors. given an image sequence, the authors show how the image many be segmented into regions with the necessary properties, optical flow is computed for these regions, and motion parameters are computed. The method can be applied to arbitrary scenes and camera motion. The authors explain why it is more robust than other techniques that require the knowledge of the full flow or information up to the second-order terms of it. Experimental results are presented to support the theoretical derivations.< >