| Abstract Tracking of Probabilistically Coupled Features (2007) | |||||||||||||||
Abstract | |||||||||||||||
| In this paper an approach on tracking objects consisting of multiple parts is presented. Instead of tracking each part independently, all features are tracked simultaneously. Therefore, the spatial dependencies of the object’s parts are described by a probabilistic model that is called “coupled structure”. The tracking process is performed using an algorithm for propagating conditional probability density function over time, called CONDENSATIONalgorithm. The main advantage of our approach is, that object modeling and object tracking are embedded into a completely probabilistic framework, so uncertainty can be handled very powerful and elegant. Finally, we demonstrate the applicability of our approach by tracking a moving human face in a difficult environment and give some experimental results. 1 | |||||||||||||||
Publication details | |||||||||||||||
| |||||||||||||||