Qualitative multiscale feature hierarchies for object tracking

This paper shows how the performance of feature trackers can be improved by building a hierarchical view-based object representation consisting of qua litative relations between image structures at different scales. The idea i s to track all image features individually and to use the qualitative featu re relations for avoiding mismatches, for resolving ambiguous matches, and for introducing feature hypotheses whenever image features are lost. Compar ed to more traditional work on view-based object tracking, this methodology has the ability to handle semirigid objects and partial occlusions. Compar ed to trackers based on three-dimensional object models, this approach is m uch simpler and of a more generic nature. A hands-on example is presented s howing how an integrated application system can be constructed from concept ually very simple operations. (C) 2000 Academic Press.