Visual Tracking with Integral Histograms and Articulating Blocks
This paper proposes an algorithm for accurate tracking of (articulated) objects using online update of appearance and
shape. In this algorithm, the constantly changing foreground shape is modeled as a small number of rectangular
blocks, whose positions within the tracking window are adaptively determined. We show that robust object tracking is possible by adaptively adjusting the locations
of these blocks. Implemented in MATLAB, our tracker runs already at 3.7 frames
per second on a 3GHz machine and the algorithm is able to efficiently track
articulated objects undergoing large variation in appearance and shape.
- Tracking results of our algorithm with fixed scale window.
- Tracking results of our algorithm with adaptive scaling window.
- Comparison with mean shift tracker (yellow) and integral histogram tracker(green)
Source codes for our tracker can be obtained here. BHTracker-Jun-25-2009.zip
For all video sequences, individual frames have been extracted and saved as png images.
Please click here to download the pdf of poster presentation at CVPR 2008.