Theoretical Study on Plenoptic Sampling and Equivalent Depth of Field for Multiple Camera Arrays

Abstract

Image Based Rendering(IBR) is a technique to generate arbitrary views from images acquired by multiple cameras. Quality of synthesized scenes is known to be greatly affected by the density of image acquisition. Each theory, Plenoptic Sampling or Equivalent Depth of Field, quantifies the relationship between camera interval and synthesis quality independently. In this paper, we clarify the relation of the two theories, and propose Generalized Equivalent Depth of Field, which extends applicable scope of Equivalent Depth of Field. Generalized Equivalent Depth of Field is expected to be a useful measure to evaluate image acquisition and scene generation processes in IBR.