Duration of project
18 months
Funding source
Swiss National Science Foundation
Funding number
SNSF contract nb: 200021-113827
Description
This project addressed the smoothing of tracked pose. Nonlinearities of ego motion motivate the application of particle filters (Bayesian tracking) for pose smoothing. Moreover, state-of-the-art results together with our own results on ego motion tracking directed our attention towards the problem of adaptive tuning (adaptive motion modelling). We proposed a research plan to find an accurate modelling of ego motion in order to impact on the usability of mixed reality technologies. In general, any MR application would benefit from the results envisioned in this grant. In particular, we were targeting user-centred mobile environments and we proposed to demonstrate it by developing a video-based tracking system. This system was, at a first stage, a hybrid of a marker-based and a particle filter based ego tracker. Further research was going to be pursued to change this marker-based tracker with a markerless one.
Research activities
-
- Rotation-discrimnative template matching: method to perform recognition of feature points based on gradient orientation and intensity information.
-
- Video-based camera tracking fusing a top-down and a bottom-up approach with a particle filter.
Related publications
Rotation correlation maps
2007
Advances in top-down and bottom-up approaches to video-based camera tracking
Lausanne, EPFL, 2007.Efficient rotation-discriminative template matching
2007. 12th Iberoamerican Congress on Pattern Recognition (CIARP), Viña del Mar-Valparaiso, Chile, November 13 to 16. p. 221–230. DOI : 10.1007/978-3-540-76725-1_24.Orientation histogram-based matching for Region Tracking
2007. Eighth International Workshop on Image Analysis for Multimedia Interactive Services (WIAMIS 2007), Santorini, Greece, 6-8 June. DOI : 10.1109/WIAMIS.2007.67.Combination of video-based camera trackers using a dynamically adapted particle filter
2007. 2nd International Conference on Computer Vision Theory and Applications (VISAPP07), Barcelona, Spain, 8 – 11 March. p. 363–370.Particle filter-based camera tracker fusing marker- and feature point-based cues
2007. DOI : 10.1117/12.703150.Feature point tracking combining the Interacting Multiple Model filter and an efficient assignment algorithm
2007. DOI : 10.1117/12.702800.Tracking and User Interface for Mixed Reality
3D Videocommunication : Algorithms, concepts and real-time systems in human centred communication; John Wiley and Sons Ltd, 2005. p. 315-332.Online Registration Tool and Markerless Tracking for Augmented Reality
2005.Partners