Research On Mvc Based Multi View Video Transport Mechanism

3D video brings new perception and rich user interaction experience to viewers.More and more3d video applications appear with the development of3d video tech-nology. With the advantage in visual efects and user interaction,3d video is expectedto replace the traditional2d TV, becoming the next generation of video media servicestandards. Multi view video, which is an important format of3d video, attracts exten-sive attention from the academia and industry. The huge amount of data produced bymulti view video poses a lot of challenges for deploying successful3D video service.How to reduce multi view video bit rate and preserve the video quality and the interac-tion experience at the same time becomes the concerns of industry. As one of the mostimportant multi view video coding standard, MVC introduces inter-view prediction,which efectively improves the multi view video compression efciency, and supportsview scalability simultaneously. However, the inter-view prediction increases the de-pendency between views. So we have to keep the integrity of the dependency structurewhen transporting and decoding MVC video. To deploy a successful3d video ser-vice system which supports a variety of application scenarios, the problems related tothe extraction, protection and packaging mechanism of view data should be studied.This paper focuses on the problems we face when transporting multi view video overMPEG transport stream, such as the view extraction of multi view video and view pro-tection and the packetization of MVC video over MPEG transport stream. The majorcontributions of this paper are as follows:Considering the huge amount of data produced by multiview video and the re-quirement of receivers difering in diferent scenarios. It is necessary to extractspecific views from MVC video for transporting and decoding. However, thecomplicated inter-view prediction introduced by MVC video brings difculty tothe extraction of views. In this paper, we propose a priority-based view extractionalgorithm to solve the problem. The proposed algorithm can efectively extract the related views and diferentiate the importance of views for decoding at thereceiver side.When transporting MVC video over unreliable channel, packet loss is inevitable.To ensure the correction and integrity of video data, the view data should be pro-tected. However, the importance of views in the MVC video is not identical dueto the inter-view prediction structure. So we should protect diferent views withdiferent priority. In this paper, we propose a view dependency aware forwarderror correct scheme to protect views during transport. The key of the protectionscheme is to consider the lower layer views when generate encode symbols ofhigher views. In this scheme, it is possible to recover the lower layer view withthe encoding symbols generated by the higher view. The experiment results showthat the proposed FEC structure can recover video data with higher probabilityand result in a better quality than the standard FEC.MPEG-2transport stream is an important packetization mechanism for transportof multimedia data over broadcast networks. We have implemented the muxerand demuxer for delivering MVC video over MPEG-2transport stream with thepriority-based extraction algorithm. The muxer and demuxer can be applied totransport MVC video in diferent scenarios.