Compressive light field photography using overcomplete dictionaries and optimized projections

By Kshitij Marwah, Gordon Wetzstein, Yosuke Bando, Ramesh Raskar. ACM Transactions on Graphics, 2013

392 citations recorded by the source corpus.

Light field photography has gained a significant research interest in the last two decades; today, commercial light field cameras are widely available. Nevertheless, most existing acquisition approaches either multiplex a low-resolution light field into a single 2D sensor image or require multiple photographs to be taken for acquiring a high-resolution light field. We propose a compressive light field camera architecture that allows for higher-resolution light fields to be recovered than previously possible from a single image. The proposed architecture comprises three key components: light field atoms as a sparse representation of natural light fields, an optical design that allows for capturing optimized 2D light field projections, and robust sparse reconstruction methods to recover a 4D light field from a single coded 2D projection. In addition, we demonstrate a variety of other applications for light field atoms and sparse coding, including 4D light field compression and denoising.

Subfields: Computer Vision and Pattern Recognition

View Paper: https://www.semanticscholar.org/paper/106be9dd2ac3f8baccf22e9f14d0c1b71e2bd69d

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Computer Vision and Pattern Recognition

Citations in the corpus

End-to-end optimization of optics and image processing for achromatic extended depth of field and super-resolution imaging (2018, methodology): “We repeatedly apply 4 rounds (i.e. 16-phase-level structures) of photolithography and reactive iron etching techniques [Morgan et al. 2004] to fabricate optics that are optimized with Fourier coefficient parameterizations of the discretized height map.”
Rank Minimization for Snapshot Compressive Imaging (2018, background): “The underlying principle is encoding the high-dimensional data on a 2D sensor with dispersion for spectral imaging [9], [10], [33], temporal-variant mask for high-speed imaging [3], [4], [5], [14], and angular variation for light-field imaging [32].”
Light Field Image Processing: An Overview (2017, background): “[33] 2013 Attenuation mask 480 × 270 × 5 × 5 – Learning-based Antipa et al.”
Light field video capture using a learning-based hybrid imaging system (2017, neutral): “Some require the input light fields to follow a specific format to reconstruct images at novel views (Levin and Durand 2010; Marwah et al. 2013; Shi et al. 2014).”
A Dataset and Evaluation Methodology for Depth Estimation on 4D Light Fields (2016, background): “We are aware of multiple smaller and larger collections, in particular the Stanford Light Field Archive [8], the Synthetic Light Field Archive [9], a collection of Lytro images [10], the 3D High-Resolution Disney Dataset [11], and the New Light Field Image Dataset [12].”