Shape's Related 3D Objects Indexing and Image Database Organization

The present paper considers the 3D shape as a sequence of 2D sections cut from the object. A part of the 2D sections (shapes) is essential for the 3D shape and the others are similar to the essential. Therefore a 3D object could be indexed through a sequence of 2D shapes, each of them coded by a numerical sequence with 4 parts. The 1
^st
one describes the convex hull. The 2
^nd
shows the number of 3
^rd
and 4
^th
level concavities. The 3
^rd
and 4
^th
parts are finite numerical sequences, describing the support of the convex set of the 2
^nd
level concavity and the support of the 2
^nd
level concavity. To comply with the sequence concept the image database is designed as a tree. The maximum number of leafs, the calculation complexity of the retrieval, and a comparison with methods close to the present one are given at the end of the paper.