Open AccessAccelerating decomposition of light field video for compressive multi-layer display
Author(s) -
Xuan Cao,
Geng Zhang,
Tuotuo Li,
Mei Zhang,
Zhaoxing Zhang
Publication year - 2015
Publication title -
optics express
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.394
H-Index - 271
ISSN - 1094-4087
DOI - 10.1364/oe.23.034007
Subject(s) - light field , computer science , coherence (philosophical gambling strategy) , optics , frame rate , field (mathematics) , cuda , computation , random access , algorithm , computer vision , artificial intelligence , physics , mathematics , parallel computing , quantum mechanics , pure mathematics , operating system
Compressive light field display based on multi-layer LCDs is becoming a popular solution for 3D display. Decomposing light field into layer images is the most challenging task. Iterative algorithm is an effective solver for this high-dimensional decomposition problem. Existing algorithms, however, iterate from random initial values. As such, significant computation time is required due to the deviation between random initial estimate and target values. Real-time 3D display at video rate is difficult based on existing algorithms. In this paper, we present a new algorithm to provide better initial values and accelerate decomposition of light field video. We utilize internal coherence of single light field frame to transfer the ignorance-to-target to a much lower resolution level. In addition, we explored external coherence for further accelerating light field video and achieved 5.91 times speed improvement. We built a prototype and developed parallel algorithm based on CUDA.