Robust Real-Time Tracking of Non-Rigid Objects

Xu, R.Y.D., Allen, J.G. and Jin, J.S.

    Several approaches to real-time video object tracking are reviewed. A new alternative approach for fast real-time object tracking based on colour thresholding is presented. Tracking is performed on non-rigid objects in a sequence of video frames based on a user-selected region of the initial frame. Details of the tracking algorithm, including colour cluster representation and pixels region grouping using run-length and noise filtering algorithm are discussed in detail. The exact contours of the tracked objects are then extracted by minimizing snake energy of thresholding results. We also experimented alpha blending the thresholding results with Canny filter edge maps to achieve more robust tracking. Foreground object colour cluster extraction at the initial frame using K-means algorithm and filtering via Foreground Extraction Mask is also discussed.
Cite as: Xu, R.Y.D., Allen, J.G. and Jin, J.S. (2004). Robust Real-Time Tracking of Non-Rigid Objects. In Proc. 2003 Pan-Sydney Area Workshop on Visual Information Processing (VIP2003), Sydney, Australia. CRPIT, 36. Piccardi, M., Hintz, T., He, S., Huang, M. L. and Feng, D. D., Eds. ACS. 95-98.
pdf (from crpit.com) pdf (local if available) BibTeX EndNote GS