Optimal inter-view rate allocation for multi-view video plus depth over MPEG-DASH using QoE measures and paired comparison


Özbek N. , Şenol E.

SIGNAL IMAGE AND VIDEO PROCESSING, vol.13, no.6, pp.1215-1223, 2019 (Journal Indexed in SCI) identifier identifier

  • Publication Type: Article / Article
  • Volume: 13 Issue: 6
  • Publication Date: 2019
  • Doi Number: 10.1007/s11760-019-01464-x
  • Title of Journal : SIGNAL IMAGE AND VIDEO PROCESSING
  • Page Numbers: pp.1215-1223

Abstract

Rapid advances in compression and transmission technologies enabled multi-view video streaming over the Internet through state-of-the-art MPEG standards such as dynamic adaptive streaming over HTTP (DASH) and high-efficiency video coding. Yet, standards for subjective and objective measurement of quality of experience (QoE) are lagging in the area. The MPEG-DASH system requires constant bitrate (CBR) coding and an efficient rate control strategy in order to increase QoE. Multi-view video plus depth format is preferable due to high compression gain but quality of intermediate view is crucial, strictly depending on performance of inter/intra-view rate allocation as well as rendering algorithms. We propose to find optimal ratio for inter-view rate allocation in CBR using our previously developed objective QoE measures using depth maps and structural similarities. To assess small differences and discriminate similar quality levels, a new subjective assessment methodology is also proposed as a multiple stimuli plus simultaneous presentation for paired comparison. The results show that the unequal rate allocation is superior to the equal rate allocation in terms of objective QoE as 0.012-0.04 (which translates into a bit rate saving of 15-24%) and subjective QoE as 0.3-0.5. Furthermore, a very high correlation is achieved between the new QoE measures and paired comparison results.