R.Collins, Y.Liu, and M.Leordeanu,
"On-Line Selection of Discriminative Tracking Features,"
to appear, IEEE Trans Pattern Analysis and Machine Intelligence (PAMI), 2005.

Shorter version appeared earlier as

R.Collins and Y.Liu,
"On-Line Selection of Discriminative Tracking Features,"
IEEE International Conference on Computer Vision, 
ICCV'03, Nice, France, October 2003, pp.346-352. 

also appeared as Technical Report CMU-RI-TR-03-12, 
Robotics Institute, Carnegie Mellon University, April 2003.

Abstract

This paper presents an on-line feature selection mechanism for evaluating multiple features while tracking and adjusting the set of features used to improve tracking performance. Our hypothesis is that the features that best discriminate between object and background are also best for tracking the object. Given a set of seed features, we compute log likelihood ratios of class conditional sample densities from object and background to form a new set of candidate features tailored to the local object/background discrimination task. The two-class variance ratio is used to rank these new features according to how well they separate sample distributions of object and background pixels. This feature evaluation mechanism is embedded in a mean-shift tracking system that adaptively selects the top-ranked discriminative features for tracking. Examples are presented that demonstrate how this method adapts to changing appearances of both tracked object and scene background. We note susceptibility of the variance ratio feature selection method to distraction by spatially correlated background clutter, and develop an additional approach that seeks to minimize the likelihood of distraction.

Full Paper

Click here for full paper (972515 bytes, pdf file).

shorter, previous ICCV'03 Version (1066642 bytes, pdf file).

older Tech report Version TR03-12, (919263 bytes, pdf file).

Download Movies

These two mpegs show some sample results.

car.mpeg, (1.6Mb). Tracking a vehicle through rapid changes in illumination (sunlight, shadows) and partial occlusion by trees.

flag.mpeg, (8.3Mb). Tracking of a flag blowing nonrigidly in the breeze, though background contrast changes (sometime the flag appears bright against dark trees, sometimes it is backlit by the bright sky).