Enhancement of Multi-Light Photographs

Thesis in collaboration with the Natural History Museum of London

2017

This project investigates the acquisition and modelling of enhanced images for the assessment of Heritage artifacts. Based on Photometric Stereo, our acquisition system uses a hemisphere mounted with several light sources and a fixed camera to capture the scene under varying lighting conditions. While being quicker than a full 3D scan, this technique further allows to reconstruct both the color and the surface’s normals of the object. This inferred informations on the geometry allows to apply processing algorithms such as smoothing and curvature estimation to create enhanced rendering.



In collaboration with the Natural History Museum of London, we tested our system on an old herbarium from one of the earliest expeditions in the North Pole. We developed a web interface to enter the digital catalogue of plants and manipulate them on the fly using our best rendering techniques.



University College of London
Supervisor: Prof. Tim Weyrich
External Supervisor: Dr. Steen Dupont