Geometric Approach for Egomotion Estimation Using Normal Flow

In this paper, we address the problem of egomotion estimation for a monocular moving observer, under arbitrary translation and rotation. The method we propose is uniquely based on the spatio-temporal image derivatives, and has O(n 2) complexity. The approach consists in searching the image for particular geometric properties of the normal ow. The algorithm is performed in two successive steps: the-line search algorithm determines the direction of the Focus of Expansion (FOE) and computes some information on the rotation parameters; the-circle search algorithm determines the FOE, and the individual values of the rotation. Various experiments with synthetic and real images are presented and discussed.