Object Tracking Method Using PTAMM and Estimated Foreground Regions
This chapter proposes a new approach for tracking moving objects in videos taken by a hand-held camera. The proposed method is based on the particle filter. The method is robust to occlusion by other objects. The 3D point map calculated by the Parallel Tracking and Multiple Mapping (PTAMM) is used for obtaining the positional relation between the target object and other moving objects. This causes improving the accuracy of the judgement of occlusion and being able to track the target object robustly when it is hidden by the others. The method uses the estimated foreground regions for calculating a part of likelihood. This increases the robustness of the tracking when the camera moving with rotation is used. The effectiveness of the proposed method is shown through the experiments using real videos.
KeywordsObject tracking Particle filter PTAMM 3D point map
Fukui’s research is supported by JSPS Grant-in-Aid for Young Scientists (B) (23700199). Iwahori’s research is supported by JSPS Grant-in-Aid for Scientific Research (C) (26330210) and a Chubu University Grant. Woodham’s research is supported by the Natural Sciences and Engineering Research Council (NSERC).
- 2.Castle, R., Klein, G., Murray, D.W.: Video-rate localization in multiple maps for wearable augmented reality. In: 12th IEEE International Symposium on Wearable, Computing, pp. 15–22 (2008)Google Scholar
- 7.Khan, Z., Balch, T., Dellaert, F.: An MCMC-based particle filter for tracking multiple interacting targtes. In: Proceeding of ECCV2004, vol. 4, pp. 279–290 (2004)Google Scholar
- 8.Lucas, B., Kanade, T.: An iterative image registration technique with an application to stereo vision. In: Proceedings of IJCAI, pp. 674–679 (1981)Google Scholar
- 9.Perez, P., Hue, C., Vermaak, J., Gangnet, M.: Color-based probabilistic tracking. In: Proceedinf of 7th ECCV, Vol. 1, pp. 661–675 (2002)Google Scholar
- 12.Watanabe, G., Fukui, S., Iwahori, Y., Bhuyan, M.K., Woodham, R.J.: Robust tracking method based on particle filter for crossing of targets with similar appearances. In: SCIS-ISIS 2012, F1–54-4, pp. 1–4 (2012)Google Scholar