3D Content-based Search

                          3D shape matching has evolved to a wide research area during the last years. At the same time, a variety of emerging applications, such as CAD and games design, computer
animations, manufacturing and molecular biology applications, dictates the need for efficient 3D search and retrieveal tools. Among the several approaches introduced for 3D shape matching, the most well-known ones are based on low-level geometrical characteristics, which can be effectively extracted from the global shape of a 3D object.

The efficient and simple query- by-content approach has been almost universally adopted in the literature, until now. Any such method, however, must first deal with the proper positioning and orientation of the 3D models. The two common methods for the solution to this problem are the pose normalization, where models are placed in a normalized coordinate frame, and native descriptor invariance, where the models are described in a transformation invariant manner. Most of the existing methods for 3D content based search and retrieval, are utilizing the pose normalization method.

   Download :     Full Report (.pdf)