This thesis presents a new approach to
background subtraction for multimodal
systems. The work is an extension
to the Gaussian mixture model background
subtraction of Stauffer and Grimson.
The background conformity values
of two image sources, namely thermal
and RGB, are fused in order to enable
stable background subtraction for persistent
surveillance. Image quality heuristics
based on image characteristics and
external sources are specified to evaluate
the usefulness of the modalities and
perform the fusion context aware. Extensions
for the use of the system for
the purpose of traffic monitoring are presented.
Therefor modulations of a new
image representation of the conformity
of pixels with the background model are
made. The potential of the proposed
method has been shown during excessive
tests of quantitative and qualitative characteristics.