Abstract
This paper presents a new approach for the estimation of motion trajectories from image sequences. The approach is not limited to the tracking of features points, but relies on the matching of image patches in consecutive image frames. Motion trajectories are parametrically modelled and the problem is considered from am optimisation perspective using a Markov Random Field (MRF) framework. A three stage optimisation procedure is used to determine the trajectory parameters and a recursive estimation approach is developed. A method for the robust detection of occlusion regions is also presented. Although robustly computed motion trajectories show great promise in many motion-related areas of computer vision, this paper illustrates their use in the context of motion compensated prediction as part of a video compression algorithm.