Investigating Material Decay of Historic Buildings using Visual Analytics with Multi-

Temporal Infrared Thermographic Data

Maria Danese*,** Urška Demšar***, Nicola Masini*, Martin Charlton***

* National Counsil of Research Archaeological and Monumental Heritage Institute,

**Università degli Studi della Basilicata,Dipartimento di Architettura, Pianificazione ed

Infrastrutture di Trasporto***National Center for Geocomputation

Infrared Thermography: introduction

A remote sensing technique

Many application fields

IRT for Cultural Heritage (decay research)

Infrared Thermography: introduction

J = σ·T4 Black body model

- J = exitance, radiation emitted per unit of surface (W/m2)- σ is Stefan-Boltzmann’s constant (5.67 × 10-8 W/m2K4) - T is the absolute temperature (°K)

J = ε·σ·T4 Gray body model

- ε = emissivity

The problem: material characterization and decay research

Large number of parameters involved in the process of the heat transfer

• Spectral properties (absorption, reflection, transmission)• Thermal properties (conductivity, diffusiveness, effusiveness,

specific heat)• Geometric properties (porosity, volumetric mass)

Big size and dimensionality of multi-temporal IR dataset (ten thousand of pixel per thermogram…)

The problem: material characterization and decay research

Spatial continuity of materials: spatial clusters

Thermal inertia of materials: temporal clusters

Methods: Visual Analytics of multi-temporal infrared thermographic

imagery Definition: visual spatial data analysis as a part of exploratory spatial data analysis employs visual exploration of large data sets in order to identify spatio-temporal and other patterns that subsequently serve as basis for hypothesis generation and analytical reasoning about the data and the phenomenon that generated these data.

Environment built using Geovista Studio*:- Self-Organising Map (SOM)- Temporal Parallel coordinates- Parallel coordinates plot linked to SOM- A map linked to the SOM

*Gahegan et al. 2002

Methods: the Self-Organizing Map (SOM)

It maps a multidimensional space in a bidimensional one

The output space

• is a regular grid or hexagonal lattice

• Has two types of cells: node cell, distance cells GeoVISTA Studio SOM (Guo et al. 2005)

Methods: the Parallel Coordinates Plot (PCP)

Each polygonal line is the representation of a data element

Each axe represents a dimension of the problem

Methods: the PCP linked to SOM

Each polygonal line is the representation of a node cells of the SOM

Each axe represents a dimension of the problem

Case study: the façade of the Cathedral in Matera, Italy

1. calcarenite surface with a few shallow alveoli (ashlars 1, 2, 3, 4, 5 and 9); 2. light alveolisation (isolated and slightly deeper alveoli) and diffuse erosion of the surface (ashlars 10 and 12); 3. significant alveolisation (alveoli deeper than those of the pattern 2) that start to be connected (ashlars 7, 13 and14);4. strong alveolisation and irregular surface (ashlars 6, 8 and part of ashlar 11); 5. dark coloured crust probably attributable to a past protective treatment (ashlar 11);6. the behaviour of the mortar between ashlars;7. other phenomena that are not recognisable in the photo taken in visible light, such as for example the presence of humidity in the wall.

Acquisition of IR thermal images and pre-processing of the data

Thermal camera used characteristics• AVIO TVS 600 microbolometric• long wave spectrum (8 ÷14 μm,)• lens of 35 mm • target range of 3.30m • spatial resolution is 1.4 mrad

Thermograms : spatial resolution is 4.62mm

Results of first experiment

Results of first experiment

Results of first experiment

Results of first experiment

Results of first experiment

Results of first experiment

Results of first experiment

to use this approach to study•More materials•Different kind of decay

to map identified patterns

to give a practical help for restoration of the building (economic advantages)

to iteratively re-evaluate and control the restoration results at every step during the restoration process.

Future goals