Rate-distortion Optimized Streaming of Compressed Light Fields with Multiple Representations
Prashant Ramanathan, Bernd Girod · 2004
Image-based rendering has been proposed as a way of enabling interactive photorealistic viewing of objects and scenes without the complexity of traditional computer graphics rendering techniques. Image-based rendering, however, relies on a large amount of image data to achieve photorealistic quality and freedom in viewing directions and position. This poses several challenges for remote viewing of these data sets over a network, the first of which is efficient compression of the image data. When streaming this data to a remote user who is interactively viewing the light field, random access to images also turns out to be an important consideration. With conventional light field coding techniques, there is typically a trade-off between random access and compression efficiency. Recently, a new encoding scheme using multiple representations based on SPframes from video coding has been proposed. This new scheme provides both good random access and compression efficiency. In this paper, we propose a method for doing rate-distortion optimized streaming of light fields that have been encoded with multiple representations. We demonstrate that using multiple representations in a streaming scenario can provide better ratedistortion performance to the remote user.