HAL will be down for maintenance from Friday, June 10 at 4pm through Monday, June 13 at 9am. More information
Skip to Main content Skip to Navigation

Region-based tracking in an image sequence

François Meyer 1 Patrick Bouthemy 1
1 TEMIS - Advanced Image Sequence Processing
IRISA - Institut de Recherche en Informatique et Systèmes Aléatoires, INRIA Rennes
Abstract : This paper addresses the problem of motion tracking in a sequence of monocular images. We want to establish and maintain the successive positions of objects in a sequence of images. The use of regions as primitives for tracking enables us to directly handle consistent object-level entities. On one hand, a motion-based segmentation process based on normal flows and first order motion models provides us with instantaneous measurements of the geometry of each region present in the segmented images. Shape and position of each projected object are estimated with a recursive algorithm along the sequence. On the other hand, a motion filter based on a multiresolution estimation scheme and temporal filtering generates reliable estimates of the motion parameters of each region. The proposed approach relies on adequate modeling and measurement of the geometry and kinematics of object projections. In particular no 3-D information is required. It realizes a good trade-off between tractability and efficiency. Occlusion situations can be handled. We have carried out experiments on both sequences of synthetic and real images depicting complex outdoor scenes to illustrate the performance of this new method.
Document type :
Complete list of metadata

Contributor : Rapport de Recherche Inria Connect in order to contact the contributor
Submitted on : Monday, May 29, 2006 - 11:41:55 AM
Last modification on : Friday, February 4, 2022 - 3:21:55 AM
Long-term archiving on: : Friday, May 13, 2011 - 10:15:10 PM


  • HAL Id : inria-00076962, version 1


François Meyer, Patrick Bouthemy. Region-based tracking in an image sequence. [Research Report] RR-1723, INRIA. 1992. ⟨inria-00076962⟩



Record views


Files downloads