2D/3D Semantic Annotation of Spatialized Images for the Documentation and Analysis of Cultural Heritage

Abstract

The introduction of digital technologies in the documentation methods for cultural heritage has helped to design new tools for the acquisition and management of information collected for multidisciplinary studies. These tools are beginning to emerge as the preferred media for describing, analyzing and understanding the objects of study. Existing solutions for semantic annotation on images, on 3D models or with 2D/3D hybrid methods still reveal themselves today insufficient to tackle the complex problem of annotating heritage artifacts. In this field, the semantic description of the studied objects must be able to rely on a rich and structured representation by on one hand making explicit the morphological complexity of the object and on the other hand by reflecting all aspects conveyed by the acquisitions of scientific imaging. This paper introduces an approach for conducting semantic annotations on 2D images (photography, scientific imaging ...) while facilitating the annotation work with an automatic propagation of these annotations between other correlated representations (2D or 3D) of the object. It is based on a spatial referencing method aiming at the establishment of a continuous 2D/3D projective relationship. The goal of the approach is to define an informative continuum between all phases of observation and description processes ranging from the acquisition of images and spatial data up to the building of semantically-enriched 3D representations. The idea is to insert semantics at all phases of 2D/3D data processing while ensuring a continuous correlation of annotations from a spatial, temporal and morphological point of view.