| Abstract | Statistical density estimation techniques are used in manycomputer vision applications such as object tracking, back-
ground subtraction, motion estimation and segmentation.
The particle filter (Condensation) algorithm provides a gen-
eral framework for estimating the probability density func-
tions (pdf) of general non-linear and non-Gaussian systems.
However, since this algorithm is based on a Monte Carlo ap-
proach, where the density is represented by a set of random
samples, the number of samples is problematic, especially
for high dimensional problems. In this paper, we propose
an alternative to the classical particle filter in which the un-
derlying pdf is represented with a semi-parametric method
based on a mode finding algorithm using mean-shift. A mode
propagation technique is designed for this new representa-
tion for tracking applications. A quasi-random sampling
method [14] in the measurement stage is used to improve
performance, and sequential density approximation for the
measurements distribution is performed for efficient compu-
tation. We apply our algorithm to a high dimensional color-
based tracking problem, and demonstrate its performance by
showing competitive results with other trackers.
|
