1. INTRODUCTION
A light field [1] [2] is a data set used to render 3D interactive photorealistic graphics. A light field image is captured by a camera whose position in 3D space can be described by five variables: three for the planar viewing positions and two for the viewing angles [3]. Sampling of the 3D space yields a two-dimensional array of 2D light field views. Novel views are constructed by appropriately combing image pixels from existing views. Dense sampling leads to a large volume of data; for example, the light field data set of the statue David by Michelangelo has an uncompressed size of 36 gigabytes [3]. Thus, compression is imperative in any light field coding scheme.