A NEW RECONSTRUCTION FILTER FOR UNDERSAMPLED LIGHT FIELDS

by
Jason Eliot Stewart

A thesis submitted to the faculty of the University of North Carolina at Chapel Hill in partial fulfillment of the requirements for the degree of Master of Science in the Department of Computer Science.

Chapel Hill
2004

ABSTRACT

JASON ELIOT STEWART: A New Reconstruction Filter for Undersampled Light Fields
(Under the direction of Leonard McMillan)

    This work builds on previous research in the light field area of image-based rendering. I present a new reconstruction filter that significantly reduces the "ghosting" artifacts seen in undersampled light fields, while preserving important high-fidelity features such as sharp object boundaries and view-dependent reflectance. By improving the rendering quality achievable from undersampled light fields, the presented method allows acceptable images to be generated from smaller image sets. I give both frequency and spatial domain justifications for the proposed techniques. I also describe a practical framework for implementing the reconstruction filter in multiple rendering passes.

The full thesis is available in PDF format (20.7 MB).