Determination of the attitude of 3-D objects without correspondence

Presents a model-based approach for solving the problem of determining the attitude of a 3-D object from a single perspective image. This approach does not require the knowledge of point-to-point correspondences between 3-D points on the model and the 2-D points in the observed image. The attitude of the object is iteratively estimated and updated from the values of feature functions globally defined over the model image and the observed image. The proposed method has been tested on images representing polyhedral objects in known attitudes.<<ETX>>