Geometric hashing using 3D aspects and constrained structures

Geometric hashing, as an effective model retrieving method, acts as an important role in object recognition. The most of the current geometric hashing methods are suitable for the 2D scene recognition under affine transformation. In this paper, geometric hashing method is extended to 3D object recognition under perspective transformation. In which, 3D aspects of object and geometric constrained structures are used to construct hash table. In this way, geometric invariants of constrained structures can provide the hashing function, and the 3D aspects of object give the information of object pose, which can simplify matching procedure. In experiment, some artificial objects are used to verify the method and the experimental results show that the proposed method is correct and effective.