3D Scene Modeling for Distributed Video Coding

1 The compression efficiency of Distributed Video-Coding2 (DVC) suffers from the necessity of transmitting a large num-3 ber of key-frames which are intra-coded. This paper describes4 a new 3D model-based DVC approach which reduces the key-5 frame frequency. The decoder first recovers a 3D model from6 the key-frames. It then predicts the intermediate frames by7 projecting it onto 2D image planes and applying image-based8 rendering techniques. This paper also introduces a new quasi-9 DVC method relying on a limited point tracking at the en-10 coder. It greatly improves the prediction PSNR, while only11 slightly increasing the encoder complexity. It also allows12 the encoder to adaptively select the key-frames based on the13 video motion-content.14

• Type: Conference
• Year: 2006
• Where: ICIP
• Authors: Matthieu Maitre, Christine Guillemot, Luce Morin
• Keywords: Encoder Complexity, ICIP 2006, Image Processing, Large Num-3 Ber, Limited Point Tracking, Model-based Dvc Approach, Quasi-9 Dvc Method