Jean-Luc Mari

Using 3D Spline Differentiation to Compute Quantitative Optical Flow

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.

Computational Topology in Image Context

Controlling the construction of geometric objects is important for several Geometric Modeling applications. Homology (groups and generators) may be useful for this control. For such incremental construction processes, it is interesting to incrementally compute the homology, i.e. to deduce the homological information at step s of the construction from the homological information computed at step s−1. We here study the application of effective homology results [13] for such incremental computations.