Low complexity transform coding for depth maps in 3D video

Abstract

3D video is a new technology, which requires transmission of depth data alongside conventional 2D video. The additional depth information allows to synthesize arbitrary viewpoints at the receiver for adaptation of perceived depth impression and for driving of multi-view auto-stereoscopic displays. Depth maps typically show different signal characteristics compared to textured video data. Piecewise smooth regions are bounded by sharp edges resembling depth discontinuities. These edges lead to strong ringing artifacts when depth maps are coded with DCT-based transform codecs, such as AVC or its successor HEVC. In this paper alternative transforms are proposed to be used for coding depth maps for 3D video. By replacing the DCT with these transforms, ringing artifacts in the reconstructed depth maps are reduced and at the same time the complexity of the transform stage is lowered significantly. For high quality depth map coding the proposed alternative transforms can even increase coding efficiency.