Free Viewpoint Video Streaming: Concepts, Techniques and Challenges

Huszák, Árpád (2016) Free Viewpoint Video Streaming: Concepts, Techniques and Challenges. IEEE Multimedia Communications Technical Committee (MMTC) Communications – Frontiers, 11 (2). pp. 17-21.

Preview

Text FVVstreaming.pdf Download (3MB) | Preview

Abstract

In contrast to traditional 3D videos, which offer the users only a single viewpoint, Free-Viewpoint Video (FVV) is a promising approach to allow free perspective selection while watching multi-view video streams. The user-specific views dynamically change based on the user’s position, and they must be synthetized accordingly. The unique views are synthesized from two or more high bitrate camera streams and the corresponding depth maps that must be delivered over the network and displayed with low latency. By increasing the number of deployed cameras and the density of the camera setup, the free-viewpoint video experience becomes more realistic. But on the other hand, more camera streams require higher network capacity. Therefore, viewpoint synthesis is a very resource hungry process and there is the need to find the best tradeoff between the quality of the synthetized view, which is related to the number of the delivered camera streams, and the processing time of the algorithm. The required camera streams may change continuously due to the free navigation of viewpoint, hence effective delivery schemes are required to avoid starvation of the viewpoint synthesizer algorithm and keep the network traffic as low as possible. Moreover, packet losses due to congestion and the increased latency can disturb the user experience. In order to support more multi-view videos in IP networks, a simple approach is to minimize the bandwidth consumption by transmitting only the minimal number of camera views.

Actions (login required)

Edit Item