Light source estimation for augmented reality

This internship was part of a project on augmented rendering for live video, at the computer vision laboratory of CEA-Saclay. The goal was to detect the light sources of the scene from specular reflections on shiny objects, such as cars.

This lighting estimation was meant to make the reflections and colors of the augmented object more photorealistic. I carried out the literature review and built the prototypes. My first reconstructions on simulated data demonstrated the feasibility of the approach.