Abstract—Compressive 3D display that can reproduce a light field, namely dense multi-view images, with several light-attenuating LCD panels stacked in front of a backlight has been investigated. Viewers can observe different images from different viewpoints because the light rays emitted from the backlight pass through different pixels at each light-attenuating layer depending on the outgoing directions. The transmittance patterns for the layer panels (layer patterns) are calculated so that the display can reproduce a given light field as accurately as possible. We suppose a 3D video transmission system where such compressive displays are adopted as the receiving terminals. Under this scenario, either light fields or pre-calculated layer patterns should be transmitted from the sender to a receiver. It should be noted that the layer patterns by themselves are compressive representations, because the entire light field (tens to hundreds of images) are reduced into only a few layer patterns. However, it is unclear how much the quality of the reproduced light fields is degraded due to the encoding errors of the layer patterns. To clarify this point, we compared the coding efficiencies of the light fields and layer patterns under this communication scenario. Experimental results show that the layer patterns have advantages over the light fields in terms of the rate-distortion performance in a lower bit range.
Index Terms—Compressive 3D display, light field compression, rate-distortion characteristics.
The authors are with Graduate School of Engineering, Nagoya University,
Nagoya-shi, Aichi, Japan (Corresponding author: Kohei Isechi; e-mail:
{isechi, kobayashi}@fujii.nuee.nagoya-u.ac.jp, {keita.takahashi,
t.fujii}@nagoya-u.jp).
Cite: Kohei Isechi, Yuto Kobayashi, Keita Takahashi, and Toshiaki Fujii, "Light Field Compression for Compressive 3D Display," International Journal of Machine Learning and Computing vol. 9, no. 2, pp. 208-212, 2019.