Fast and exact method for computing a stack of images at various focuses from a four-dimensional light field

Mhabary, Ziv; Levi, Ofer; Small, Eran; Stern, Adrian

doi:10.1117/1.jei.25.4.043002

Cited by 3 publications

(5 citation statements)

References 0 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Unfortunately, the process of taking Fourier transforms, interpolating for slicing, and then taking inverse transforms introduces significant computational overhead, making FSP unsuitable for real-time rendering. This assumption was confirmed by Mhabary et al [21], who have worked to advance FSP by employing a fractional Fourier transform. However, the authors conclude that the integral projection operator in the spatialdomain is faster when computing only a single refocused image from a lightfield.…”

Section: A Backgroundmentioning

confidence: 66%

See 1 more Smart Citation

Real-Time Refocusing Using an FPGA-Based Standard Plenoptic Camera

Hahne¹,

Lumsdaine

Aggoun

et al. 2018

IEEE Trans. Ind. Electron.

View full text Add to dashboard Cite

Plenoptic cameras are receiving increased 5 attention in scientific and commercial applications because 6 they capture the entire structure of light in a scene, en-7 abling optical transforms (such as focusing) to be applied 8 computationally after the fact, rather than once and for all at 9 the time a picture is taken. In many settings, real-time inter-10 active performance is also desired, which in turn requires 11 significant computational power due to the large amount 12 of data required to represent a plenoptic image. Although 13 GPUs have been shown to provide acceptable performance 14 for real-time plenoptic rendering, their cost and power 15 requirements make them prohibitive for embedded uses 16 (such as in-camera). On the other hand, the computation 17 to accomplish plenoptic rendering is well structured, 18 suggesting the use of specialized hardware. Accordingly, 19 this paper presents an array of switch-driven finite impulse 20 response filters, implemented with FPGA to accomplish Q1 21 high-throughput spatial-domain rendering. The proposed 22 architecture provides a power-efficient rendering hardware 23 design suitable for full-video applications as required in 24 broadcasting or cinematography. A benchmark assess-25 ment of the proposed hardware implementation shows that 26 real-time performance can readily be achieved, with a one 27 order of magnitude performance improvement over a GPU 28 implementation and three orders of magnitude performance 29 improvement over a general-purpose CPU implementation.

show abstract

Section: A Backgroundmentioning

confidence: 66%

“…An early attempt at high-performance rendering was based on the projection slice theorem, which rendered images with lower-dimensional slices of the lightfield in the Fourier domain [3], [21]. This procedure is also known as Fourier Slice Photography (FSP).…”

Section: Introductionmentioning

confidence: 99%

Real-Time Refocusing Using an FPGA-Based Standard Plenoptic Camera

Hahne¹,

Lumsdaine

Aggoun

et al. 2018

IEEE Trans. Ind. Electron.

View full text Add to dashboard Cite

show abstract

“…The fidelity advantage of our method was demonstrated in Ref. 26. Here we focused on a computational load comparison.…”

Section: Discussionmentioning

confidence: 96%

“…In Ref. 26 we demonstrate that our method outperform FPS approximately 14dB in the PSNR value. Figures 5-7 summarize the computational burden benchmark results.…”

Section: Comparison Between the Refocusing Methodsmentioning

confidence: 95%

“…We have recently developed a new and efficient method 26 for computing a stack of images digitally focused at various lengths from a four dimensional (4D) light field. This method is fast, algebraically exact and does not require interpolation in the frequency or spatial domains, as alternative methods do.…”

Section: Fast and Eaxact Refocusinh Methods Using The Fractional Fourimentioning

confidence: 99%

See 1 more Smart Citation

An exact and efficient 3D reconstruction method from captured light-fields using the fractional Fourier transform

et al. 2016

Self Cite

View full text Add to dashboard Cite

Recently we have introduced a new method for refocusing images from captured light fields. The method is based on the fractional Fourier transform and allows refocusing a stack of images with a single step. The new techniques is computational efficient and more exact than alternative ones as it doesn't need any interpolations. Here we overview the method present a comparison of our technique to other refocusing techniques.

show abstract