Computational Optical Imaging Overview

Principles

Definition

Computational optical imaging is a research field and methodological approach that integrates physical optics and image formation processes with advanced computational algorithms for image acquisition, processing, and reconstruction. It investigates methods to overcome limitations inherent in traditional optical systems, enabling enhanced image quality, quantitative information extraction, and the development of novel imaging modalities.

Ontological type

Core Methods

Imaging Modalities

Reconstruction Algorithms

Key developments

Key Figures

PSF and Phase Computation era

Marc Levoy [1], affiliated with Stanford University [3] and the University of North Carolina at Chapel Hill [4], is a leading figure in the PSF- and phase-centric imaging of this era. The 1988 paper Display of surfaces from volume data [7] advanced volume rendering and surface extraction from volumetric data, providing quantitative PSF-based visualization that underpins later deconvolution and phase-sensitive imaging. Ichirou Yamaguchi [2], affiliated with Tokyo University of Science [5] and Hokkaido University [6], is another key figure in this era. His 1997 paper Phase-shifting digital holography [8] advanced phase-sensitive holography and complex-field measurements, enabling improved depth retrieval in scattering media and strengthening PSF-based phase computation frameworks.

Wavefront and Holographic Control era

Lei Tian [1], affiliated with Massachusetts Institute of Technology [3] and University of California, Berkeley [4], helped shape computation-enabled wavefront control during the Wavefront and Holographic Control era. His key contributions include multiplexed coded illumination for Fourier Ptychography with an LED array microscope [5], and 3D intensity and phase imaging from light field measurements in an LED array microscope [6], enabling high-resolution reconstructions from low-NA captures and robust phase retrieval essential to this era. Laura Waller [2], likewise affiliated with MIT [3] and UC Berkeley [4], co-developed the same computational approaches that underlie Fourier Ptychography and LED-based light-field imaging demonstrated in Multiplexed coded illumination for Fourier Ptychography with an LED array microscope [5] and 3D intensity and phase imaging from light field measurements in an LED array microscope [6]. Together, these contributions epitomize the era's program of computation-enabled wavefront control and holographic reconstruction, enabling extended depth encoding, improved resolution, and volumetric imaging through iterative phase retrieval and holographic algorithms.

Meta-Optical Co-Design era

Tie Jun Cui[1] is a leading figure in metasurface-enabled computational optical imaging during the Meta-Optical Co-Design era (2018–2024), with affiliations at University of Science and Technology of China[3] and Nanyang Technological University[4]. His 2022 work A programmable diffractive deep neural network based on a digital-coding metasurface array[7] demonstrates end-to-end programmable optical computation using a digital-coding metasurface, a contribution that advances differentiable optical design and compact, adaptable sensing pipelines essential for the era. Edmund Y. Lam[2] is a prominent contributor during this era, with affiliations at University of Hong Kong[5] and Chinese University of Hong Kong[6]. His 2019 End-to-end deep learning framework for digital holographic reconstruction[8] provides an end-to-end learning approach for holographic reconstruction, accelerating robust retrieval and enabling efficient processing in constrained optical channels as part of the era's end-to-end holographic sensing paradigm.