Extracting information from multimodal images of documents and artworks

We present some image processing techniques that have been demonstrated useful to extract relevant information or improve the human readability of various artefacts of cultural interest. In many cases, the imaging data available for one such object are captured through different technologies and modalities, and in different times and places. Digital processing tools are essential to extract useful information from these data, especially when the physical objects are not accessible anymore, and further data acquisition campaigns are thus impossible. The first step to build a multimodal data cube from heterogeneous image data consists in relating each pixel to a single point in the physical space. Then, all the data layers are ready to be processed jointly to produce coherent information. Depending on the result expected, several processing strategies are possible. Some examples are reported here.