By the term “fragmentation” we mean the identification of the original position of a fragment in relation to the object or building to which it belongs, as well as the selection of the adjacent fragments in order to be put together to a unit.

The fragmentation is traditionally approached on the basis of the ability of the human brain to combine fractured surfaces and characteristic decorative elements. The traditional techniques, however, are time consuming and require special equipment if, in particular, the relevant objects are bulky or heavy.

Some scholars have proposed the use of computers for solving the problem of fragmentation, however, a practical solution applicable to a wide variety of objects has not been set forth as yet.

The methodology proposed here is based on the creation of three-dimensional digital models of fragments and their analysis on the principles of Riemannian geometry and Mahalanobis distance. These models are segmented in fractured and decorative surfaces. The data concerning the object derive from decoration and from the available archaeological knowledge, while the fractured surfaces are numerically characterized and identified on the basis of their topography.

The combination of the aforementioned parameters decreases the possible associations of fragments considerably and contributes to the faster solution of fragmentation, without endangering, at the same time, the actual objects.