3-D shape indexing language

The authors propose an indexing method that allows a 3-D shape database to be accessed with descriptive words or 2-D images, both of which are very intuitive and powerful. The new 3-D shape-indexing method uses superquadrics. First, the requirements for 3-D indexing are given. From an estimation of 3-D modeling methods, the superquadric model is found to be most suitable, and its performance is discussed. Applications of this 3-D indexing method to man-machine communications are given. first, the authors show that a modeled 3-D object can be easily deformed by verbal instruction, such as `rounder´ and `squarer´. Second, they prove that the shape abstraction of a 3-D object is easily created from two rough sketches showing the cross-sections of the object. And finally, an experimental 3-D image database system that makes good use of this indexing method is created. This new indexing method promises to revolutionize 3-D visual communication and database systems