Distributed Compression of Light Fields

We address the problem of compression for large light field data acquisition systems and present a distributed scheme based on Wyner-Ziv coding. The proposed scheme allows independent encoding of each view with low-complexity cameras, and performs centralized decoding at the more sophisticated decoder, which takes advantage of side information obtained from additional views acquired by conventional cameras. We provide experimental results of the proposed scheme in comparison with the use of JPEG2000 for two light field data sets. The Wyner-Ziv codec yields superior compression performance at low bit-rates but inferior efficiency at higher bit-rates. Additionally, we see a reduction in computation complexity for Wyner-Ziv encoder even when compared with JPEG2000.