Light field displays allow us to visualize 3D content without wearing 3D glasses. However, a class of objects (called non-Lambertian) such as mirrors or transparent objects are difficult to render. Non-Lambertian objects distort the light rays and are therefore difficult to render realistically in such 3D displays. In this thesis, we will work with 3D scenes acquired with a Plenoptic camera (e.g. RayTrix), a camera that is able to capture content for the light field display.
The student will develop a method to 1) detect the non-Lambertian objects of the scene (segmentation), 2) display the 3D scene on a simulated light field display.
The ideal candidate for this Master thesis is driven and a problem solver. The student is eager to work with state-of-the-art techniques on light field processing. The student will be able to work in a dynamic team.

Prerequisites: Programming skill (C++ and/or Python), familiarity with image processing is a plus.

Keywords: Computer vision, Image analysis, Segmentation

Promotor: Mehrdad TERATANI (mehrdad.teratani@ulb.be)

Support: Sarah Fachada, Daniele Bonatto