Italian Chapter Conference 2017 - Smart Tools and Apps in computer Graphics

Permanent URI for this collection

https://diglib.eg.org/handle/10.2312/2631683

Browse

Browsing Italian Chapter Conference 2017 - Smart Tools and Apps in computer Graphics by Subject "I.3.8 [Computer Graphics]"

Now showing 1 - 2 of 2
  • No Thumbnail Available
    Item
    A Digital Approach for the Study of Roman Signacula From Syracuse, Sicily
    (The Eurographics Association, 2017) Tanasi, Davide; Milotta, Filippo L. M.; Gradante, Ilenia; Stanco, Filippo; Kaplan, Howard; Andrea Giachetti and Paolo Pingi and Filippo Stanco
    In the last decade the epigraphists have grown a new interest in signacula, a class of artifacts for a long time neglected. This has brought numerous contributions devoted to the different regional contexts, along with reflections on methodological questions, not to mention the momentum towards the digitizing of a corpus which counts at least 3,500 pieces, confirming the great potential of these artifacts in providing information related not only to the economy and to the administration of the ''res'', both in public and private sphere, but also about the profile of the signacula holders. In this scenario, a specific research question has been inspired by the Sicilian seals - about 60 signacula and a dozen impressions left by seals on mortar in burial contexts: it is possible to identify unequivocally a signaculum through its impression? Given for granted that the use of 3D documentation will bring along effective results in terms of improved readability of signacula and seals, the aim of this contribute is to establish a protocol for a semi-automatic matching between 3D models of seals and 3D models of impressions. As part of a preliminary scanning campaign of Late Roman impressions on mortars and metal seals from the catacombs of Syracuse, two bronze metal seals were digitized with a NextEngine 3D triangulation laser scanner and subsequently 3D printed with liquid resin with a Formlabs Form 2 SLA high resolution printer. The casts obtained, were experimentally used to create a set of impressions on mortar using different degrees and angles of pressure, in order to create similar but still different stamps. During the next step, the impressions were 3D scanned and used as ground truth for the outlined semi-automatic procedure of matching with the seals. In MeshLab environment, the 3d models of seals and impressions were manually aligned and then the distance between two sets of 3D points was measured using the filter Hausdorff distance in order to validate a matching. This successful exercise could open the way to the proposal of creating a virtual edition of signacula with 3D models metadata. Furthermore, a research agenda may include the design of a machine learning algorithm for matching of 3D meshes.
  • No Thumbnail Available
    Item
    The Social Picture: Advanced Image Analysis Applications
    (The Eurographics Association, 2017) Milotta, Filippo L. M.; Bellocchi, Michele; Battiato, Sebastiano; Andrea Giachetti and Paolo Pingi and Filippo Stanco
    In The Social Picture (TSP) an huge amount of crowdsourced social images can be collected and explored. We distinguish three main kind of events: public, private and cultural heritage related ones. The framework embeds a number of advanced Computer Vision algorithms, able to capture the visual content of images and organize them in a semantic way. In this paper we employ VisualSFM (VSFM) to add new features in TSP through the computation of a 3D sparse reconstruction of a collection within TSP. VisualSFM creates a N-View Match (NVM) file as output. Starting from this NVM file, which characterizes the 3D sparse reconstruction, we are able to build two important relationships: the one between cameras and points and the one between cameras themselves. Using these relationships, we implemented two advanced Image Analysis applications. In the first one, we consider the cameras as nodes in a fully connected graph in which the edges weights are equal to the number of matches between cameras. The spanning tree of this graph is used to explore images in a meaningful way, obtaining a scene summarization. In the second application, we define three kinds of density maps with relation to image features: density map, weighted-density map and social-weighted-density map. Results of a test conducted on a collection from TSP is shown.