Information Capacity and Estimation Enhancement in Imaging Systems

Abstract

The pursuit of high resolution, large field of view, and high parameter estimation accuracy has been a driving force in the field of computational imaging. This dissertation contributes to computational imaging by studying the imaging system from the perspective of Shannon capacity and Fisher information. Through the physical design of optics and sensors, we explore how these two fundamental measures of information can be extended to enhance imaging performance. We begin with the design of a multifocal array camera aimed at increasing depth-of-field coverage without relying on active focusing mechanisms. System-level design considerations are discussed, and detailed lens design and stray-light analyses are presented. The optical design should be tested. Therefore, Chapter 3 addresses this critical issue by developing optical metrology techniques for system testing. An on-axis deflectometric testing configuration is developed, demonstrating high accuracy, large dynamic range, and robustness to miscalibration errors. Finally, we investigate the design of sensors for measuring spatial coherence of the optical field at the image plane in incoherent imaging. Starting from the fundamentals of coherence theory, we establish a framework for coherence measurement and show through simulation that the Rayleigh criterion can be surpassed when coherence information is available. The Fisher information in coherence measurements for two-point sources is analyzed. Further more, the relationship between the modulation transfer function (MTF) and Fisher information in coherence-based imaging is further explored.