CMSC 828Z::

Class Description :: Syllabus :: Reading List :: Projects :: Helpful Links ::

CMSC828Z: Description of Project 2 - Image-based Recovery of Illumination Environment

This project is closely related to project 1 and it might be a good idea that the two students who work on projects 1 and 2 closely collaborate. The goal of this project is to recover a spherical environment illumination map of the dome. The rigid structure of the dome was surrounded by cloth, thereby ensuring slowly varying illumination. Still there is more light entering the dome from some directions than others and the final result of this project should be an intensity map that describes this light intensity distribution. A good starting point for this project are the papers by Ravi Ramamoorthi about A Signal-Processing Framework for Inverse Rendering and An Efficient Representation for Irradiance Environment Maps (see also this page about spherical harmonics).

As described on the page for project 1 we moved a color checker chart around the room. The varying brightness values of the chart at different locations and orientations can be used to recover a low-dimensional approximation to the irradiance environment map that can be refined by using the background images of the environment that we recorded (see the following examples - the high-res image can be accessed by clicking on the low-res image:)

As said before, this project will need to be done in collaboration with project 1, since they mutually depend on eachother. Without a good radiometric calibration, we cannot recover the irradiance environment map accurately and without an idea about the directionality of the irradiance map, it is hard to account for the dependence of the recorded intensity on the orientation of the color chart.

The end result of this project will consist of a spherical map that maps orientations to environment irradiance intensities. This environment will then in later stages of the project to be used to create illumination invariant models and allow for a more accurate recovery of the surface reflectance properties.