Local fractallity as a feature in database image retrieval

An important aspect encountered in archaeological field consists in the identification of the broken pieces part of an object. A preliminary analysis presupposes the grouping of the articles by material, but this is not enough. A second analysis will be done, this time considering the shape of the frontier of the broken piece. In order to describe the shape of an object's border, the evaluation of some local fractal features is proposed. The algorithms for extracting fractal features were implemented into an original software. The final purpose consists in building a component for a content-based image retrieval (CBIR) which uses a classifier based on fractal properties.