Fast Time-Space Tracking of Smoothly Moving Fine Structures in Image Sequences.

David Tschumperlé (1) , Yohan Bentolila (2) , Jen Martinot (2) , Jalal M. Fadili (1)

1 Equipe Image - Laboratoire GREYC - UMR6072
2 IFPEN - IFP Energies nouvelles

We address the problem of temporal tracking fine point-like and filamentary structures exhibiting smooth motions in \mbox{image} sequences. By taking these specific restrictions into account, we propose an original tracking method based on the search for integral lines in time-space structure tensor fields. The method appears to be simple and very efficient \mbox{regarding} computation time and tracking precision, allowing a sub-pixel accuracy in both spatial and temporal domains. We suggest a numerical implementation of the algorithm based on a modified constrained Runge-Kutta scheme. The performance and potential applicability of our algorithm is illustrated with two real cases : the tracking of internal linear features in composite materials observed with X-rays and the tracking of granular-shaped objects moving in a gas flow.

Domaines: Traitement des images [eess.IV]

David Tschumperlé, Yohan Bentolila, Jen Martinot, Jalal M. Fadili. Fast Time-Space Tracking of Smoothly Moving Fine Structures in Image Sequences.. IEEE International Conference on Image Processing (ICIP'07), 2007, San Antonio, United States. pp.636--639. ⟨hal-00332806⟩