Analysis and Optimization of Aperture Design in Computational Imaging

TL;DR

This paper analyzes the fundamental limits of coded aperture imaging systems using mutual information, revealing optimal aperture designs under different noise conditions and comparing them to classical pinhole cameras.

Contribution

It introduces an analysis framework based on mutual information to determine optimal aperture coding for various noise regimes in computational imaging.

Findings

Spectrally-flat masks are optimal under thermal noise.

Randomly generated masks perform better under shot noise.

Comparisons show advantages over classical pinhole cameras.

Abstract

There is growing interest in the use of coded aperture imaging systems for a variety of applications. Using an analysis framework based on mutual information, we examine the fundamental limits of such systems---and the associated optimum aperture coding---under simple but meaningful propagation and sensor models. Among other results, we show that when thermal noise dominates, spectrally-flat masks, which have 50% transmissivity, are optimal, but that when shot noise dominates, randomly generated masks with lower transmissivity offer greater performance. We also provide comparisons to classical pinhole cameras.