In the case of works of art or objects coming from the field of Cultural Heritage (CH), the data collected during digitization allows a complete chain of processing to be performed from the gathering of digital data to statistical analysis, including various techniques of pattern recognition and pattern classification to facilitate the work of art historians and museums curators. These techniques can be refined and adapted to numerous particular cases, for example, as we will see, to the analysis of vases.
In the analysis of an object by a CH expert, there is often the need to search for similar 3D objects. This search may be carried out using text queries that aim at retrieving objects with similar 32 metadata, but the CH expert is often more interested in searching for similar objects based on criteria that cannot be described in a few words. For instance, a CH expert may want to retrieve objects with similar shape (global or local), colour or texture.
- Shape is the information about its size and its form. In the case of a collection of complete vases, the analysis of the shape can help to classify automatically the objects according to a notion, which is not easy to describe in words, but is immediately comprehensible by our brain. For instance, some shape characteristics are good discriminators for a particular region of production. In the case of vases for example the angle of the “neck” is characteristic of different families. The 3-dimensional scans of the objects may contain data about the micro structures of the vases, which can be also be analysed in different ways to classify the vases in separate groups. The thickness of the patterns can be particularly studied to discriminate groups of vases. Other information contained in the patterns can be used as well – for example the form of the designs - and can be combined with other data to refine the analysis.
- An analysis of the information carried by the pigments (Colour and Chemical composition) of the vases can also be used for the classification of vases. The colour histogram of an object corresponds to the spectrum of the colours used to paint the vase. Such histogram is often a good discriminator of the period and/or region of product ion. Texture refers to the 2D artistic details/patterns on the surface of the object. For instance, in the case of Greek vases, the designs on red-figure vases or on black-figure vases from distinct periods obey distinct styles. The temporal evolution of the designs can be measured by the use of an adequate classification of the patterns, by composition, component detail, craftsmanship, etc.
"Oenochoé Levy", DAGER dept, INV. E658, Louvre Museum, France
