A model-based pose determination algorithm for segmented 3-D images has been developed using the geometric hashing method and the interpretation tree. Its major advantage is an intensive off-line model preprocessing stage, where model information is indexed into a hash-table using minimal, transformation invariant features. This enables the on-line recognition algorithm, with its voting procedure, to be particularly efficient. Once the correspondence of a minimal set of features is established by the geometric hashing method, the interpretation tree is used to obtain the complete set of correspondences between the object's model and the segmented range image. The best pose estimate is then found using a constrained least-squares method.
展开▼