3D Sensing and Reconstruction (3W5A)

Presider: Hong Hua, University of Arizona
Authentication Required

Access to this content is only available to registered attendees. Please login now to access the live content.

13:00 - 13:45
(UTC - 07:00)

How to Pass the Visual Turing Test (3W5A.1)
Presenter: Douglas Lanman, Facebook

How can a display appear indistinguishable from reality? We describe how to pass this “visual Turing test” using AR/VR headsets, emphasizing the joint design of optics, display components, rendering algorithms, and sensing elements.

Authors:Douglas Lanman, Facebook

13:45 - 14:00
(UTC - 07:00)

Assessment of a Fresnel biprism-based digital holographic microscope for fast, high-sensitivity, high-resolution and polarization-sensitive phase imaging (3W5A.2)
Presenter: Ana Doblas, University of Memphis

We present a digital holographic microscope using a Fresnel biprism. The proposed system offers the five desired features in phase imaging: high stability high speed, high accuracy, high resolution, and sensitivity to the birefringence.

Authors:Charity Hayes-Rounds, University of Memphis / Brian Bogue-Jimenez, University of Memphis / Jorge Garcia-Sucerquia, Universidad Nacional de Colombia Sede Medellin / Omar Skalli, University of Memphis / Ana Doblas, University of Memphis


14:00 - 14:15
(UTC - 07:00)

Overview on convolutional neural network-based classification of red blood cells in lensless single random phase encoding (3W5A.3)
Presenter: Timothy O'Connor, University of Connecticut

We overview a previously reported system for red blood cell identification using convolutional neural networks in lensless single random phase encoding. The methods presented provide improved classification performance and increased robustness to various noise models.

Authors:Timothy O'Connor, University of Connecticut / Christopher Hawxhurst, University of Connecticut / Leslie Shor, University of Connecticut / Bahram Javidi, University of Connecticut


14:15 - 14:30
(UTC - 07:00)

Joint Reconstruction Strategy for Telecentric-based Digital Holographic Microscopes (3W5A.4)
Presenter: Ana Doblas, University of Memphis

We present a reconstruction algorithm for digital holographic microscopy (DHM) operating in the telecentric regimen. This strategy rests on the minimization of a cost function to estimate both the numerical reference wave and the quantitative phase information. This algorithm paves the way to a universal DHM reconstruction tool.

Authors:Raul Castañeda Quintero, University of Memphis / Ana Doblas, University of Memphis


14:30 - 14:45
(UTC - 07:00)

Impact of non-linear distortion on learning based approaches for depth estimation (3W5A.5)
Presenter: Julie Buquet, Universite Laval

In this work, we evaluate the impact of non-linear distortion on learning based approaches for depth estimation. We use real wide angle images with locally increased resolutions to see how it compares with perspective images.

Authors:Julie Buquet, Universite Laval / Jinsong Zhang, Universite Laval / Patrice Roulet, Immervision / Simon Thibault, Universite Laval / Jean-François Lalonde, Universite Laval


14:45 - 15:00
(UTC - 07:00)

An Overview of MPEG Immersive Video (3W5A.6)
Presenter: Basel Salahieh, Intel Corporation

MPEG immersive video is an upcoming standard for data access and delivery of immersive content. A brief introduction of its encoding and decoding processes is presented along with a simplified playback implementation of it demonstrated on Intel GPUs.

Authors:Basel Salahieh, Intel Corporation / Mengyu Chen, Intel Corporation / Jill Boyce, Intel Corporation