Viewpoint-invariant indexing for content-based image retrieval

Current methods for shape- based image retrieval are restricted to images containing 2-0 objects. We pro-pose a novel approach to querying images conta,ining 3-0 objects, based on a view-based encoding of a finite domain of 3-0 parts used to model the 3-0 objects ap-pearing in,images. To build a quevy, the user manually identifies the salient parts of the object in a query im-age. The extracted views of these parts are then vsed to hypothesize the 3-0 identities of the parts which, in turn, are used to hypothesize other possible views of the parts. The resulting set of part views, along with their spatial relations (constraints) in the query im-age, form a composite q%ery that is passed to the im-age database. Images containing objects,with the same parts (in any viezu),with similar spatial relations are returned to the user. The resulting viewpoint invariant indexing technique does not require training the system for all possible,views of each object. Rather, the sys-tem requires only kno,wledge of the possible views for a finite vocab,ulary of 3-0 parts from which the objects are constructed