Towards Optimal Multirate Encoding for HTTP Adaptive Streaming

Amirpour, Hadi; Çetinkaya, Ekrem; Timmerer, Christian; Ghanbari, Mohammad

doi:10.1007/978-3-030-67832-6_38

Cited by 7 publications

(2 citation statements)

References 17 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In [3], the authors show different approaches for efficient multirate encoding with respect to parallel encoding performance. They propose an improved method by using the middle quality representation as the reference representation, which achieves the best overall results in terms of encoding time-complexity including significant improvements while encoding the highest quality representation.…”

Section: Related Workmentioning

confidence: 99%

See 1 more Smart Citation

Multi-resolution quality-based video coding system for DASH scenarios

Moina-Rivera

Aguado

García-Pineda

2021

Proceedings of the 31st ACM Workshop on Network and Operating Systems Support for Digital Audio and Video

View full text Add to dashboard Cite

Nowadays the use of Internet has become pervasive in many aspects of our life. Among the delivered content, multimedia has a large share and for this reason, suitable techniques for supporting the streaming services are needed. In this context, Dynamic Adaptive Streaming over HTTP (DASH) is one of the most widely adopted technique. The idea that underlies DASH is the fragmentation of the video content in smaller segments that are then encoded at different resolutions and bit rates. At the client side, the segment with the highest possible bit rate that can be downloaded and played without causing delays or stalls is automatically chosen by an Adaptive BitRate (ABR) algorithm. In this framework, one important aspect that should be considered is the quality perceived by the viewer of the streamed content. For this reason, Variable bitrate (VBR) encoding is used to select a bit rate for each segment, based on its complexity. In this paper, Wilmer Moina-Rivera and his colleagues present a method that is based on encoding the video in several resolutions, based on the scores of a selected objective quality metric. More specifically, they address a specific case that is the encoding of raw 4K videos with DASH at five selected resolutions (240p, 480p, 720p, 1080p and 2160p) while keeping the selected quality metric value fixed. In the considered case they adopted the Video Multimethod Assessment Fusion (VMAF) with a value of 90 (VMAF scores range from 0 to 100, with 0 indicating the lowest quality, and 100 the highest). The proposed approach is pretty straight forward. It relies on applying an adaptive, iterative Constant Rate Factor (CRF) encoding scheme for each segment rather than a constant CRF factor throughout the video encoding. The experimental results support the claims of the authors and show an expected improvement with respect to the case in which a constant CRF is selected. Further investigation will be of interest by considering different video encoding schemes and quality metrics. We believe that this paper will especially be of interest for engineers and industry practitioners.

show abstract

Section: Related Workmentioning

confidence: 99%

“…In our performance evaluation several tests have been carried out over three open source movies: Big Buck Bunny (BBB) 1 , Sintel (STL) 2 and Tears of Steel (TOS) 3 . These videos have a spatial resolution of 4K with a frame rate of 24 fps, and are stored in YUV4Mpeg format (y4m).…”

Section: Test-bedmentioning

confidence: 99%