Please use this identifier to cite or link to this item: doi:10.22028/D291-37841
Volltext verfügbar? / Dokumentlieferung
Title: Low complexity light field compression based on pseudo-temporal circular sequencing
Author(s): Hariharan, Harini Priyadarshini
Lange, Tobias
Herfet, Thorsten
Language: English
Title: 2017 IEEE International Symposium on Broadband Multimedia Systems and Broadcasting (BMSB)
Pages: 1-5
Publisher/Platform: IEEE
Year of Publication: 2017
Place of publication: Piscataway, NJ
Place of the conference: Cagliari, Italy
Free key words: Video Coding and Processing
3D and Multiview Video
Future technologies and services of Broadcasting and Video Processing
DDC notations: 004 Computer science, internet
Publikation type: Conference Paper
Abstract: Light fields are gaining more impact in today's filming technology due to post-processing possibilities such as software refocus, perspective shift, reconstruct 3D images, and depth scaling. Light fields are captured light rays at every point in space travelling in every direction, therefore a large amount of data is recorded for representation. For exchange, broadcast and storing of such imagery data, we lack efficient compression techniques. This paper proposes a light field compression approach based on low-complexity preprocessing by pseudo-temporal sequencing frames suitable for state-of-the-art codecs. An efficient novel approach of circular reordering is introduced where the redundancy between frames is maximized and exploited by low-delay HEVC predictive coding. Experimental results showcase a consistent increase in both PSNR and SSIM for several light fields and outperforms available schemes with pseudo-temporal sequencing using HEVC. An increase upto 1 dB gain compared to state-of-the-art reordering techniques using HEVC is achieved.
DOI of the first publication: 10.1109/BMSB.2017.7986144
Link to this record: urn:nbn:de:bsz:291--ds-378417
hdl:20.500.11880/34225
http://dx.doi.org/10.22028/D291-37841
ISBN: 978-1-5090-4937-0
978-1-5090-4938-7 (ISBN der Printausgabe)
Date of registration: 7-Nov-2022
Faculty: MI - Fakultät für Mathematik und Informatik
Department: MI - Informatik
Professorship: MI - Prof. Dr. Thorsten Herfet
Collections:SciDok - Der Wissenschaftsserver der Universität des Saarlandes

Files for this record:
There are no files associated with this item.


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