Feature point tracking combining the Interacting Multiple Model filter and an efficient assignment algorithm

• Published in: Proc. of the IS&T/SPIE Electronic Imaging Conf. on Visual Communications and Image Processing
• Series: Lecture Notes in Computer Science
• Publication date: 2007

An algorithm for feature point tracking is proposed. The Interacting Multiple Model (IMM) filter is used to estimate the state of a feature point. The problem of data association, i.e. establishing which feature point to use in the state estimator, is solved by an assignment algorithm. A track management method is also developed. In particular a track continuation method is presented. The evaluation of the tracking system on real sequences shows that the IMM filter combined with the assignment algorithm outperforms the Kalman filter, used with the Nearest Neighbor (NN) filter, in terms of data association performance and robustness to sudden feature point maneuvers.

Keywords: data association ; Extraction ; Feature Point ; Interacting Multiple Model ; LTS1 ; Tracking

Reference

• LTS-CONF-2006-062

• URL: http://electronicimaging.org/program/07/

Record created on 2006-10-27, modified on 2012-03-20