Structure from Motion and History – PhD Scholarship

PhD in Computer Science
Coordinator: Riccardo Focardi
IIT Principal Investigator: Alessio del Bue
Ca’ Foscari University Contact Person: Andrea Torsello

Venice is the most imaged city of the world, in any historical period. Painters in the late 16th century were able to accurately depict the city geometry through realistic paintings that are masterpiece of their epoch. Moreover, there exists a massive photographic catalogue ranging late 1800 up to our days. This research theme has the aim to provide a 3D reconstruction of the city of Venice using images coming from different periods of time in order to make evident:

– Accurate localisation of historical images and paintings on a 3D map
– Structural changes in the architecture through time
– Missing or new 3D structures in the urban landscape
– Environmental variations in the city

To perform such tasks research would be dedicated to find similar images from different sources and periods (e.g. paintings, photographs, digital images), matching features between them, and provide a 3D reconstruction together with a semantic representations of the architecture of buildings, these existing now or disappeared through the ages. Machine learning approaches with deep architectures would be investigated in order to perform some of these tasks, such as point matching and image similarities estimation.