Paper Info
Title
Content-Aware Efficient Video Uploading For Crowdsourced Multi-View Video Streaming
Abstract
Integration of video streams captured by many mobile cameras at a crowded event into a multi-view video enables viewers to experience the different perspectives of the event. The streaming of such crowdsourced multi-view videos has been popular in many applications such as entertainment, surveillance, social sharing and is potentially applicable to medical, educational and military areas. However, due to the resource-constrained nature of wireless network and the huge amount of video traffic, simultaneous uploading of video streams from many mobile cameras limits the quality of the video streams. In this paper, we propose a content-aware video uploading scheme for crowdsourced multi-view video streaming. Our work considers correlation-based differential encoding with multiple reference streams by packet overhearing to keep better quality under the reduced video traffic. The proposed scheme consists of two parts: correlation estimation and transmission order determination. First, we exploit the correlation among the mobile cameras based on the content features of the captured videos. Second, we decide the encoding behavior of each camera based on the correlation degrees among the cameras. Based on the encoding behavior, we schedule the transmission order of the cameras that can realize the differential encoding by packet overhearing. Evaluation results show up to 2.5 dB quality improvement with the traffic reduction of up to 30% in a strongly correlated network of mobile cameras.
Year
Venue
DocType
2018
2018 INTERNATIONAL CONFERENCE ON COMPUTING, NETWORKING AND COMMUNICATIONS (ICNC)
Conference
ISSN
Citations 
PageRank 
2325-2626
0
0.34
References 
Authors
0
3
Name
Order
Citations
PageRank
Than Than Nu101.35
Takuya Fujihashi200.68
Takashi Watanabe316522.64