Probabilistic detection and tracking at high frame rates using affine warping

This paper addresses two vital issues that can affect real-time operation of a visual tracking system: the realization of an effective sub-sampling policy and the real-time initialization of the tracking algorithm. We propose to use affine warping to sub-sample the images selectively only in those regions that contain too much data for real-time operation. The automatic detection of objects of interest in images captured by a moving camera is based on a random search which enables us to set all thresholds automatically without any user support. Using these methods, we implemented a probabilistic tracker that can detect and track up to 10 objects at 60 Hz on a dual processor 933 MHz Pentium III PC.

[1]  Nebojsa Jojic,et al.  Tracking self-occluding articulated objects in dense disparity maps , 1999, Proceedings of the Seventh IEEE International Conference on Computer Vision.

[2]  Ales Ude,et al.  Real-time visual system for interaction with a humanoid robot , 2001, Proceedings 2001 IEEE/RSJ International Conference on Intelligent Robots and Systems. Expanding the Societal Role of Robotics in the the Next Millennium (Cat. No.01CH37180).

[3]  Christoph Bregler,et al.  Learning and recognizing human dynamics in video sequences , 1997, Proceedings of IEEE Computer Society Conference on Computer Vision and Pattern Recognition.

[4]  Shaogang Gong,et al.  Tracking colour objects using adaptive mixture models , 1999, Image Vis. Comput..

[5]  Ying Wu,et al.  A co-inference approach to robust visual tracking , 2001, Proceedings Eighth IEEE International Conference on Computer Vision. ICCV 2001.

[6]  Dorin Comaniciu,et al.  Real-time tracking of non-rigid objects using mean shift , 2000, Proceedings IEEE Conference on Computer Vision and Pattern Recognition. CVPR 2000 (Cat. No.PR00662).