Using 3D Spline Differentiation to Compute Quantitative Optical Flow

John Barron (1) , Marc Daniel (2) , Jean-Luc Mari (2)

1 UWO - University of Western Ontario
2 GMOD-LSIS - GMOD-LSIS

We show that differentiation via fitting B-splines to the spatio-temporal intensity data comprising an image sequence provides at least the same and usually better 2D Lucas and Kanade optical flow than that computed via Simoncelli’s balanced/matched filters.

Domaines: Autre [cs.OH]

John Barron, Marc Daniel, Jean-Luc Mari. Using 3D Spline Differentiation to Compute Quantitative Optical Flow. Third Canadian Conference on Computer and Robot Vision, CRV 2006, CIPPRS, Jun 2006, Québec, Canada. ⟨10.1109/CRV.2006.84⟩. ⟨hal-01311458⟩