Virtual welcoming
Prof Carola-Bibiane Schönlieb
University of Cambridge, UK
"Research on the synthesis of forms and abstract images​"
"When de Prony Met Leonardo: An Automatic Algorithm for Chemical Element Extraction in Macro X-ray Fluorescence Data"
"From Inpainting to Shape Matching: Mathematics for Cultural Heritage Challenges"


Research on the synthesis of forms and abstract images

Prof Jean-Michel Morel

ABSTRACT: It is generally accepted by theorists and practitioners of the graphic and digital arts that forms have structure, that images are constructed from forms by laws of composition, and that said structures and laws of composition do not require “figuration” or imitation. 

The question then arises as to how to automatically create forms, textures and images, “abstract” in the sense that they are not imitating forms already seen. This question is central in decorative arts or in abstract art, and is therefore not new. In this talk I will discuss the principles of “automatic” digital synthesis of forms and images, and how such principles can finally be implemented by replacing subjective choices with dice strokes.

When de Prony Met Leonardo: An Automatic Algorithm for Chemical Element Extraction in Macro X-ray Fluorescence Data

Prof Pier Luigi Dragotti

ABSTRACT: The heritage sector is experiencing a digital revolution driven in part by the increasing use of non-invasive, non-destructive imaging techniques. These techniques range from visible images, images taken using different forms of radiation e.g. infrared and X-ray, as well as images derived using new spectroscopic imaging techniques such as for example macro X-Ray Fluorescence (MA-XRF). These new imaging methods provide a non-destructive way to capture information about an entire painting and can give us information about features at or below the surface of the painting. This is important to support art historical research to interpret and contextualise a collection and to help conserve and care for the collection. At the same time, these new imaging methods also provide new exciting ways of engaging with objects from our cultural heritage. However, the wealth of digital data generated by these instruments calls for new automatic approaches to extract as much information as possible from these often very large datasets.

In this talk we focus on Macro X-Ray Fluorescence (XRF) scanning which is a technique for the mapping of chemical elements in paintings. After describing in broad terms the working of this device, we introduce a method that can process huge-amount of XRF scanning data from paintings fully automatically. The method is based on connecting the problem of extracting elemental maps in XRF data to Prony’s method, a technique broadly used in engineering to estimate frequencies of a sum of sinusoids. The results presented show the ability of our method to detect and separate weak signals related to hidden chemical elements in the paintings. We then discuss results on the Leonardo’s “The Virgin of the Rocks” and show that our algorithm is able to reveal, more clearly than ever before, the hidden drawings of a previous composition that Leonardo then abandoned for the painting that we can now see.

This work is done in collaboration with the National Gallery and University College London and is supported by EPSRC

From Inpainting to Shape Matching: Mathematics for Cultural Heritage Challenges

Dr Simone Parisotto

ABSTRACT: In this talk, we will discuss about three research projects developed in the “Mathematical for Applications in Cultural Heritage” (MACH) group.

In the first part, we will show a carousel of mathematical imaging approaches for the virtual restoration (inpainting) of damages in illuminated manuscripts. Our aim is to support art conservators with alternative restoration solutions produced by mathematical and computer vision approaches, including local, nonlocal and deep-learning methods, also in view of future virtual exhibitions as a result of the COVID-19 pandemic.
This work is done in collaboration with the Fitzwilliam Museum (University of Cambridge, UK).
In the second part, we will detail about a workflow for clustering the profiles of common-ware Roman potteries, so as to bring an alternative order in the corpora of dedicated publications. This pipeline, based on the shape feature extraction via stacked sparse autoencoders (SSAE) and combined with hierarchical clustering algorithms, allows archeologists to unveil invisible relations in partial and fragmented shapes.
This work is done in collaboration with the Faculty of Classics (University of Cambridge, UK).
Last but not least we will discuss how to compare multiple layers in cross-section data extracted from paintings. After training the well-renowed deep learning U-NET network for the segmentation problem of the in-focus details, we describe the challenges in matching patches in the in-focus domain and made by the same chemical material but with different level of granularity.
This work is done in collaboration with the Hamilton Kerr Institute (Cambridge, UK).

Programme of the virtual event

- 18 November 2020 -