Comparison of the different approaches to generate holograms from data acquired with a Kinect sensor

Authors
Kang, Ji-HoonLeportier, ThibaultJu, Byeong-KwonSong, Jin DongLee, Kwang-HoonPark, Min-Chul
Issue Date
2017
Publisher
SPIE-INT SOC OPTICAL ENGINEERING
Citation
Conference on Holography - Advances and Modern Trends V, v.10233
Abstract
Data of real scenes acquired in real-time with a Kinect sensor can be processed with different approaches to generate a hologram. 3D models can be generated from a point cloud or a mesh representation. The advantage of the point cloud approach is that computation process is well established since it involves only diffraction and propagation of point sources between parallel planes. On the other hand, the mesh representation enables to reduce the number of elements necessary to represent the object. Then, even though the computation time for the contribution of a single element increases compared to a simple point, the total computation time can be reduced significantly. However, the algorithm is more complex since propagation of elemental polygons between non-parallel planes should be implemented. Finally, since a depth map of the scene is acquired at the same time than the intensity image, a depth layer approach can also be adopted. This technique is appropriate for a fast computation since propagation of an optical wavefront from one plane to another can be handled efficiently with the fast Fourier transform. Fast computation with depth layer approach is convenient for real time applications, but point cloud method is more appropriate when high resolution is needed. In this study, since Kinect can be used to obtain both point cloud and depth map, we examine the different approaches that can be adopted for hologram computation and compare their performance.
ISSN
0277-786X
URI
https://pubs.kist.re.kr/handle/201004/114678
DOI
10.1117/12.2266357
Appears in Collections:
KIST Conference Paper > 2017
Files in This Item:
There are no files associated with this item.
Export
RIS (EndNote)
XLS (Excel)
XML

qrcode

Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.

BROWSE