An Egomotion Algorithm Based on the Tracking of Arbitrary Curves
Abstract
We are interested in the analysis of non polyhedral scenes. We will present an original egomotion algorithm, based on the tracking of arbitrary curves in a sequence of gray scale images. This differential method analyses the spatiotemporal surface, and extracts a simple equation relating the motion parameters and measures on the spatiotemporal surface. When a curve contour line is tracked in image sequence, this equation allows to extract the 3D motion parameters of the object attached to the contour when rigid motion is assumed. Experiments on synthetic as well as real data show the validity of this method.