| Geometric And Stochastic Error Minimisation In Motion Tracking (2008) | |||||||||||||||
Abstract | |||||||||||||||
| Tracking has been considered to be a single view problem conventionally, where one is interested in the projections of a particular object in a view over time. Even if many views of the same event are available, tracking often proceeds independently in each view. The geometric information due to the projection of the same object onto multiple image planes is not utilized. In this paper, we couple the stochastic error used by the Kalman filter with a geometric error term derived from multiview geometric constraints to achieve improved tracking in individual views. We present experimental results to evaluate the performance of the algorithm. 1. | |||||||||||||||
Publication details | |||||||||||||||
| |||||||||||||||