3D Video Compression Using Depth Map Propagation

Abstract

We propose a method of 3D video compression based on 2D+depth representation. We use correlation between 2D video and depth map to reconstruct highly compressed depth map while decoding. Depth map is compressed with reduced spatial and temporal resolution. On decoding stage the algorithm restore original resolution using information from 2D video. We evaluated the influence of key frames' resolution, compression ratio and density on the performance of the algorithm. The proposed technique was compared to depth map compression using H.264 in compression pipeline for stereo video. © Springer-Verlag Berlin Heidelberg 2013.