A method of estimating an orientation of one or more of a plurality of objects disposed on a plane, from one or more video images of a scene, which includes the objects on the plane produced from a view of the scene by a
video camera. The method comprises receiving for each of the one or more objects, object
tracking data, which provides a position of the object on the plane in the video images with respect to time, determining from the object
tracking data a plurality of basis vectors associated with at least one of the objects, each basis vector corresponding to a factor, which can influence the orientation of the object and each basis vector being related to the movement or location of the one or more objects, and combining the basis vectors in accordance with a blending function to calculate an estimate of the orientation of the object on the plane, the blending function including blending coefficients which determine a
relative magnitude of each basis vector used in the blending function.Using the estimate of the orientation of each object, a
three dimensional model can be generated which can, as far as possible, represent the scene of the objects on the plane. For example, the scene may be that of a sporting event such as a football match, the objects being football players. The method can generate an
estimation of the orientation of each of the players on the plane of the football
pitch. The
estimation of the orientation can be combined with the object
tracking data for each player to generate a three dimensional representation, which models, as far as possible, the real football match as captured by the
video camera.